Information manipulation theory and perceptions of deception in Hong Kong
Communication Reports
Salt Lake City
Winter 1999

--------------------------------------------------------------------------------

Authors:                  Lorrita N T Yeung

Authors:                  Timothy R Levine

Authors:                  Kazuo Nishiyama

Volume:                   12

Issue:                    1

Pagination:               1-11

ISSN:                     08934215

Subject Terms:            Communication
                          Theory
                          Information
                          Culture

Geographic Names:         Hong Kong


Abstract:

This study tests McCornack's (1992) Information Manipulation Theory (IMT)
in Hong Kong.  IMT views deception as arising from covert violations of
one or more of Grice's four maxims (quality, quantity, relevance, and manner).
Copyright Western States Communication Association Winter 1999

Full Text:

This study tests McCornack s (1992) Information Manipulation Theory (IMT)
in Hong Kong. IMT views deception as arising from covert violations of
one or more of Grice's four maxims (quality, quantity, relevance, and manner).
Previous studies conducted in the United States have found that messages
violating one or more of the four maxims are rated as less honest than
messages that do not violate the maxims. Based upon cultural differences
in expectations and social roles, we predicted that only violations of
quality (i.e., outright falsification) would be seen as universally deceptive.
To test this prediction, McCornack, Levine, Solowczuk, Torres and Campbell's
(1992) original study was replicated in Hong Kong (N = 310). The results
indicated that violations of quality (falsification) and relevance (evasion)
were rated as deceptive in Hong Kong. However, message ratings along all
four dimensions were significantly correlated with deception ratings, suggesting
that perhaps the results stem from differences in what counts as a covert
violation rather than more fundamental differences in the appropriateness
of the maxims.

The prevalence of deception in everyday conversation is well documented
(DePaulo, Kashy, Kirkendol, Wyer, & Epstein, 1996; Turner, Edgley, & Olmstead,
1975), as are the potentially negative consequences stemming from the discovery
of deception (e.g., McCornack & Levine, 1990). Although research on deception
processes has long flourished, researchers have traditionally focused on
the falsification of information to the exclusion of more subtle forms
of deceptive messages (McCornack, 1992). Lies (i.e., presenting false information),
however, represent only one of many ways to deceive another (Bowers, Elliott,
& Desmond, 1977; Ekman,1985; Hopper & Bell,1984; Turner et al.,1975). By
limiting deception only to those acts involving the falsification of information,
many verbal acts that are functionally deceptive (e.g., equivocation: Bavelas,
Black, Chovil, & Mullett, 1990; evasion: Galasinski, 1994; Turner et al.
1975) are excluded.

While investigations of deceptive message design are currently in vogue
(e.g., Bavelas et al., 1990; Burgoon, Buller, Guerrero, Afifi, & Feldman,
1996; Jacobs, Dawson, & Brashers, 1996; Galasinski, 1994; McCornack, Levine,
Solowczuk, Torres, & Campbell, 1992), this research has characteristically
been concerned only with deception in Western cultures. This cultural myopia
is unfortunate. Because of cultural differences related to the individualism-collectivism
distinction, what counts as deception most likely differs across cultures.
With intercultural interactions becoming increasingly frequent due to advances
in communication technology and increases in mobility, differing views
of deception may increase the potential for misunderstanding, mistrust,
and ill will. It would seem, then, that a crosscultural examination of
deceptive message design is needed.

The current study examines the generalizability of Information Manipulation
Theory (IMT; McCornack, 1992) by attempting to replicate McCornack et al.'s
(1992) seminal work in Hong Kong. This investigation begins with a review
of research on information manipulation.

INFORMATION MANIPULATION THEORY

IMT offers a multidimensional approach to deceptive messages, integrating
Grice's (1989) theory of conversational implicature with research on deception
as information control (e.g., Bavelas et al., 1990; Bowers et al., 1977;
Metts, 1989; Turner et al., 1975). Specifically, IMT uses Grice's (1989)
Cooperation Principle (CP) and its maxims as a framework for describing
a variety of deceptive message forms. IMT views deception as arising from
covert violations of one or more of Grice's four maxims (quality, quantity,
relevance, and manner). Covert violations of quality involve the falsification
of information. Covert violations of quantity can result in "lies of omission."
Deception by evasion involves covert violations of relevance, and deception
by equivocation results from the covert violation of manner.

IMT also offers a pragmatic explanation for why deceptive messages deceive.
As McCornack (1992) wrote:

It is the principal claim of Information Manipulation Theory that messages
that are commonly thought of as deceptive derive from covert violations
of the conversational maxims... Because the violation is not made apparent
to the listener, the listener is misled by her/his assumption that the
speaker is adhering to the CP and its maxims. (p. 5-6)

Thus, covert violations of one or more of Grice's conversational maxims
(quality, quantity, relevance, and manner) are believed to result in messages
that are functionally deceptive.

To date, there have been three tests of IMT: McCornack et al.'s (1992)
original study and replications by Jacobs et al., (1996) and Lapinski (1995).
Each of the studies provided subjects with a hypothetical situation and
one of five message forms. Subjects were asked to rate the messages in
terms of honesty. The messages either violated one of Grice's maxims (quality,
quantity, relevance, and manner) or were baseline messages designed to
be honest (i.e., accurate, informative, clear, and relevant). In each of
the three studies, the baseline message was rated as significantly more
honest than the four messages violating one of the maxims. That is, each
of these three studies found that messages violating one or more of the
four maxims are seen as more deceptive than are messages that adhere to
Grice's maxims.

Research on Culture and Deception

At a very general level, the concept of deception may be universal. Saarni
and Lewis (1993), for example, argued that deception that is centered around
clandestine affairs, protecting one's possessions from a competitor, and
feigning emotion occur in most if not all cultures. Similarly, Buss and
Schmitt (1993) implied that deceptive sexual selection strategies might
be crosscultural.

The few cross-cultural studies of deception that exist, however, tend to
focus on cultural differences. O'Hair, Cody, Wang, and Choa (1990) investigated
vocal stress in the truthful and deceptive messages of Chinese immigrants.
Chinese had higher levels of vocal stress when revealing negative emotions.
Aune and Waters (1994) found that the more collectivistic American Samoan
participants indicated they would be more likely to deceive another on
an issue related to family or other ingroup concerns. U.S. Americans, in
contrast, were motivated to deceive when they felt an issue was private
or when they wanted to protect the target person's feelings.

Nishiyama (1994) discussed deception in a cultural framework from a business
perspective. Nishiyama suggested that there are a number of strategies
and behaviors that are considered everyday business practices in Japan
that may be interpreted as deceptive by U.S. American businesspeople. Commonly
misunderstood messages include official statements of policy (Tatemae),
which are different from true intentions (Honne), and certain nonverbal
behaviors that non-Japanese people find difficult to distinguish.

Finally, Lapinski (1995) investigated the relationships between honesty
ratings of the four information manipulation dimensions and self-reported
cultural orientations. Relevance violations were seen as significantly
less deceptive by those with a more collectivistic orientation.

China-U.S. Cultural Differences and Deception

The individualism-collectivism dimension (Hofstede, 1980) is perhaps the
most common way of distinguishing between cultures. Collectivism emphasizes
the goals of the ingroup over those of the individual. Asia, Africa, South
America, and the Pacific Islands have generally been considered to be the
locales of collectivistic cultures. Conversely, personal goals are emphasized
in individualistic cultures. Those cultures that are characterized by individualism
include many of those located in Australia, northern and western Europe,
and the United States.

Several studies offer empirical support for the Chinese collectivist orientation
(e.g., Bond and Kwang-kuo, 1986). There are five reasons why the Chinese
collectivist orientation should result in views of deception that diverge
from those of Westerners.

The first reason involves cultural differences in role expectations. The
Chinese have a strong tendency to act according to what is expected of
them by others. The Chinese concept of man, "ren," is said to be "based
on the individual's transactions with his fellow human beings" (Bond &
Kwang-kuo, 1986, p.220). With this expectation comes a greater concern
for the feelings of others in social interactions and a greater stress
on the obligations demanded by the social role. Thus, in interacting with
others, Chinese tend to give responses that fulfill the social expectations
of others, even if those responses are considered deceptive by Western
standards.

Second, cultural differences exist in conflict avoidance. There is both
theoretical and empirical support for the Chinese tendency to avoid conflict,
compared with Westerners (Tang & Kirkbride, 1986). Because deception is
an easy means of conflict avoidance, deceptive messages serving this purpose
may be more common among Chinese than Westerners.

Third, moral orientations differ between cultures. Chinese tend to base
their moral decisions on what they think is acceptable to their reference
group (Yang, 1986, p. 133). In contrast, Westerners more often form moral
judgments according to independently-held principles. Instead of being
universally applicable like Western moral principles, Chinese moral judgments
are related to particular roles and are therefore situational (Chiu, 1991).
It appears that role expectation is a predominant criterion among Chinese
in forming such judgments (Chiu, 1990). For example, sometimes the need
to act according to role expectation and concern for the other's feelings,
on the one hand, and the moral responsibility to be honest, on the other,
are in conflict. Manipulating message features by giving ambiguous or partial
information would appear to be a compromise to prevent the eventuality
of being a total liar.

Fourth, the concept of "face" is especially salient in the Chinese culture
(Chang & Holt, 1994). To the Chinese, the concept of face incorporates
two different notions: "lien" and "mianzi" (Hu, 1944). "Lien" refers to
the integrity of a person's moral character. "Mianzi," however, is the
personal prestige and reputation that comes with the person's achievement
and success in society. Both are essential for the operation of the Chinese
as a social being. Being socially skilled involves knowing how to do facework
(e.g., giving face, enhancing face, saving face, restoring face, etc.).
Many Chinese message strategies are motivated by the wish to do facework.
To Westerners, such message strategies may appear unnecessarily indirect.
For example, Chinese have a preference for refraining from criticism, especially
in public (Bond & Lee, 1981). Jokes and hints are commonly used for broaching
unfavorable information to the hearer (Du, 1995). Refusals to invitations
are used only rhetorically for sounding out the sincerity of the inviter
as well as avoiding being presumptuous by accepting invitations too readily
(Gu, 1990). Such indirect message strategies can also be considered as
violations of the conversational maxims and thus be construed as "deceptive."


Finally, the Chinese use of ritualistic message strategies may appear to
be deceptive in the eyes of Westerners. For example, initial refusals to
invitations are not meant to be, and are in fact not taken as, flat denials
in the Chinese context. But they may be interpreted literally by Westerners,
thus leading to cases of cross-cultural misunderstanding (Kasper & Zhang,
1995). Similarly, the ritualistic strategy of exaggeration is often used
among Chinese in doing facework. For example, exaggerated statements to
praise the accomplishments of the other and to denigrate the achievements
of oneself are frequently used in playing the face game.

Hypotheses

The integration of the theoretical distinctions between Chinese and Western
communicative styles with the research on cross-cultural deception leads
to predictions regarding perceptions of information manipulation among
Chinese in Hong Kong. We anticipate that blatant violations of quality
(i.e., outright intentional falsification) will be universally seen as
deceptive.

Falsification is the most direct and blatant form of deception and is therefore
the most likely to be universally seen as deceptive. Differences should
exist, however, in the perceptions of more subtle forms of deception (i.e.,
violations of quantity, relevance, and manner). Omission, equivocation,
and evasion involve indirectness, which may be typical of normal, honest
communication in Asian cultures. Further, because violations of these maxims
are required to fulfill expectations and social roles, these should not
be seen as particularly deceptive in Hong Kong. Hence, we hypothesize that
there will be a main effect for violation type such that violations of
quality will be seen as significantly more deceptive than other message
forms, and violations of quantity, relevance and manner will not differ
from the baseline honest message.

If the data are consistent with our hypothesis, two explanations for these
findings will be possible. First, IMT may simply not hold in collectivist
cultures because violations of quantity, relevance and manner are not seen
as constituting deception. Consistent with this speculation, Yum (1988)
argued that the maxim of manner is not the norm in East Asia. However,
an alternative explanation for the expected results is also possible. It
may be that covert violations of these dimensions are seen as deceptive,
but what counts as a violation differs with culture. That is, if a message
that violates a given maxim in North America is not seen as deceptive in
Hong Kong, it could be because either that maxim does not apply, or because
the message in question is not perceived as a covert violation of that
maxim. For this reason, we will also investigate the relationship between
message ratings of the four dimensions and perceptions of honesty.

METHOD

Participants

Three hundred and ten undergraduate students in Hong Kong completed a survey
very similar to that used by McCornack et al. (1992). Of these, 139 (44.5%)
were male, 169 (54.5%) were female, and 2 (0.6%) failed to answer the sex
question. The respondents ranged in age from 18 to 25 (M = 21.22, SD =
7.70). Differences between the current method and that of McCornack et
al. (1992) are noted below.

Design and Procedures

A 1 X 5 independent groups design was used. As in previous IMT studies
(Jacobs et al., 1996; Lapinski, 1995; McCornack et al., 1992), subjects
were presented with a hypothetical situation and one of five messages:
a baseline honest message (no violation), a false message (quality violation),
a message omitting crucial information (quantity violation), an evasive
message (relevance violation), or an equivocal message (manner violation).
The situation and messages were taken from the "Committed Chris" situation
used by McCornack et al. (1992). Participants read the situations and then
rated the message on a 3-item honesty scale (alpha = .86). Manipulation
check items for each violation type were also completed (quality: alpha
= .84; quantity: alpha = .76; relevance; alpha = .78; manner, alpha = .85).
All items were taken from McCornack et al. (1992), but they used 4 items.
The results of a confirmatory factors analysis were consistent with the
validity of the measures. The scales were second-order unidimensional,
replicating McCornack et al.'s (1996) re-analyses of McCornack et al. (1992)
and Jacobs et al. (1996).

The questionnaire was completed in English. As English was the medium of
instruction in their college, the respondents were thought to have had
a sufficient level of proficiency in English. Nevertheless, a glossary
of key terms was provided and additional verbal explanations were given
in order to ensure that the participants had a complete understanding of
the questionnaire. They were also given a chance to ask for further clarifications
of any points they did not understand.

RESULTS

The manipulation checks were successful. The message violating quantity
(M = 3.77) was rated as significantly less disclosive (tl25 = 2.92; p <
..004; r = .25) than the baseline message (M = 4.50). The message violating
quality (M = 3.39) was rated as less accurate (tl23 = 7.47;p < .001; r
= .56) than the baseline message (M = 4.90). Scores on the relevance violation
condition (M = 3.50) were rated significantly lower (tl2l = 5.94; p < .001;
r = .48) than the baseline condition (M = 4.80). Finally, violations of
manner (M = 3.90) were rated as less clear (tl22 = 3.42; p < .001; r =
..30) than the baseline message (M = 4.85).

The violation manipulation produced a statistically significant and large
main effect (F. = 22.54; p < .0001; eta2 = .23). Scheffe tests showed that
violations of quality and relevance were rated as significantly more deceptive
than the baseline message, while violations of quantity and manner were
not rated as any more deceptive than the baseline message. No differences
were found between the baseline message, the quantity violation, and the
manner violation. Similarly, quality violations did not differ from violations
of relation. Statistical power for r > .20 was .70 and was .95 for r >
..30. Means are presented in Table 1.

DISCUSSION

This study tested McCornack's (1992) Information Manipulation Theory (IMT)
in Hong Kong. Only violations of quality (falsification) and relevance
(evasion) were rated as more deceptive than the baseline message in Hong
Kong. Messages violating quantity (omission) and manner (equivocation)
did not differ from the completely honest message in deception ratings.
Simply put, false and evasive messages were rated as deceptive but omissions
and equivocation were not.

These results differ dramatically from those obtained in the United States.
Comparing the current results to the findings of previous studies conducted
in the U.S. (Michigan: McCornack et al., 1992; Arizona: Jacobs et al.,
1996; Hawaii: Lapinski, 1995), the Hong Kong students appear to rate each
of the message types differently than their U.S. American counterparts
(see Table 1). When statistically compared to McCornack et al.'s (1992)
original results, the means in each condition differed significantly. The
current respondents rated violations of quality (t267 = 10.67; p < .01;
r = .55), quantity t279 = 3.26; p < .01; r = .19), manner (t274 = 5.51;p
< .01; r = .32), and relevance (t271 = 3.06; p < .01; r = .18) as less
deceptive than did McCornack et al.'s (1992) subjects. Alternatively, the
honest baseline message was rated as more deceptive by the Hong Kong sample
than the U.S. sample, (t278 = -3.33; p < .01; r = .20). Simply put, what
is seen as truthful and deceptive appears to vary substantially across
cultures.

As mentioned previously, there are at least two general explanations for
the results. First, the fundamental expectations that guide conversational
understanding (i.e., Grice's Maxims) in Western cultures may not generalize
to Hong Kong. This explanation holds that violations of quantity and manner
are not seen as deceptive in Hong Kong because the Chinese do not expect
others to adhere to the maxims of clarity and disclosure. If the maxims
do not apply, one should not be expected to follow them.

[IMAGE TABLE] Captioned as: TABLE 1

An alternative explanation is that each of the maxims holds in Hong Kong,
but what counts as a violation differs between Hong Kong and the United
States. That is, the differences lie not in the fundamental assumptions
that guide conversations but in what is required to fulfill each maxim.


To explore these two explanations, message honesty ratings were correlated
with ratings of quality, quantity, relevance, and manner. As the correlations
in Table 2 show, messages that were rated as violations also tended to
be rated as dishonest (r = .43 to .67). So, for example, although the message
involving omission was not rated as any more deceptive than the baseline
message, the more messages were perceived as omitting information, the
less honest they were rated. This suggests that violations of the four
maxims may be seen as deceptive in Hong Kong, but that what counts as a
violation differs from the U.S.

To the extent that this second explanation is valid, IMT may have some
validity in non-Western cultures, or at least in Hong Kong. While differences
clearly exist in the ratings of specific messages, perceived violations
are associated with perceived deception. In other words, Hong Kong Chinese
may not have the same strict standards as U.S. Americans do for absolute
clarity and full information disclosure in the messages received.

It should be noted, however, that the correlations between violation ratings
and honesty rating were generally lower than those obtained in the U.S.
The same correlations in McCornack et al. (1992) were generally larger
(r = .57 to .79). When tested for significant differences with r to z transformations,
the correlations for the quantity (r = .67 vs. .43; z = 5.42), quality
(r = .79 vs. .67; z = 4.00), and manner (r = .62 vs. .52; z = 2.27) were
significantly larger in the McCornack et al. (1992) data than in the current
data. No differences were evident in the relevance correlations (r = .57
and .54; z = 0.69).

The dishonesty ratings of the evasive message were rather surprising and
inconsistent with our predictions. At least two explanations are plausible.
First, the relevance findings might stem from the particular situation
used in the questionnaire. In an intimate relationship, the role expectation
might require a direct address of the critical issue when a question is
raised about one's fidelity and commitment to a relationship. An irrelevant
answer tends to be interpreted as having something to hide, thus leading
to an interpretation of dishonesty. In other words, the perceptions of
deception by Hong Kong Chinese about the irrelevant message might not generalize
to other situations.

[IMAGE TABLE] Captioned as: TABLE 2

Second, the finding might not be an artifact, and Hong Kong Chinese might
have less tolerance for relevance violations than quantity or manner violations.
Two findings seem consistent with this account. Previous IMT studies (cf.,
Lapinski, 1995; McCornack et al., 1992) have found little situational variation
in deceptive message ratings. Also, in the current study, the manipulation
check indicated stronger manipulations of relevance and quality than quantity
or manner. Together, these findings might indicate that Hong Kong Chinese
are more sensitive to violations of relevance and quality than they are
to violations of quantity or manner, and these difference are not a mere
function of situational idiosyncrasies. A further study incorporating different
situations would be necessary to test this reasoning.

In general, the results of the present study indicate that there are different
cultural expectations regarding violations of conversational maxims leading
to perceptions of deceptive messages. Such differences in expectations
may lead to cross-cultural misunderstandings. Apparently, Hong Kong Chinese
have a higher threshold of tolerance for violations of conversational maxims
as compared with U.S. Americans. Hong Kong Chinese should not be seen as
more deceptive than U.S. Americans. Instead, they may use message manipulation
strategies to avoid hurting the other's feelings or to fulfill to social
obligations and expectations. Such violations may not be intended to be
covert. For example, they often leave certain things unsaid, expecting
the others to read between the lines. To U.S. Americans, such violations
of the conversational maxims would be seen as covert and thus would constitute
an act of dishonesty. Under such circumstances, the U.S. Americans would
take the partial or ambiguous messages coming from the Chinese as intentionally
deceptive. The Chinese, on the other hand, may be upset and embarrassed
by the U.S. American style of directness, as the U.S. Americans do not
seem to give due consideration to face. This may explain why the Hong Kong
Chinese rated the baseline message as less honest than did those in the
U.S.

While this study was designed from the perspective of IMT, other research
might legitimately study deception from a more culture specific framework.
For example, the current study might have been grounded in the culture
of Hong Kong Chinese. Such an approach would likely yield additional important
insights, and would be more ideally suited to the study of the nuisances
of deceptive discourse in Hong Kong.

Further research on cultural differences in deception is of course needed.
This study only examined deception in one situation, with one set of message
examples, in just one culture, and from the perspective of IMT. English
speaking Hong Kong Chinese may be more Westernized than some other Asian
populations, and a dating situation might have reduced applicability. Nevertheless,
this study provides some preliminary insights into cultural differences
in the perceptions of message honesty.

REFERENCES

Aune, R. K., Sr Waters, L. L. (1994). Cultural differences in deception:
Motivations to deceive in Samoans and North Americans. International Journal
of Intercultural Relations, 18, 159-172.

Bavelas, J. B., Black, A., Chovil, N., & Mullett, J. (1990). Equivocal
communication. Newbury Park, CA: Sage.

Bond, M. H. & Kwang-kuo, H. (1986). The social psychology of Chinese people.
In M. Bond (Ed.), The psychology of the Chinese people. Hong Kong: OUR
Bond, M. H., & Lee, P. W. H. (1981). Face saving in Chinese culture: A
discussion and experimental study of Hong Kong students. In M. Bond (Ed.),
Social life and development in Hong Kong. Hong Kong: The Chinese University
Press.

Buss, D. M., & Schmitt, D. P. (1993). Sexual strategies theory: An evolutionary
perspective on human mating. Psychological Review, 100, 204-232.

Bowers, J. W., Elliott, N. D., & Desmond, R. J. (1977). Exploiting pragmatic
rules: Devious messages. Human Communication Research, 3, 235-242.

Burgoon, J. K., Buller, D. B., Guerrero, L. K., Afifi, W. A., & Feldman,
C. M. (1996). Interpersonal deception XII: Information management dimensions
underlying deceptive and truthful messages. Communication MonoQrahs 69.


Chang, H-C, & Holt, R. (1994). A Chinese perspective on face as inter-relational
concern. In S. Ting-Tooney (Ed.), The challenge of facework: Cross-cultural
and interpersonal issues. Albany, NY: State University of New York Press.


Chiu, C-Y. (1990). Role expectation as the Principal criterion in justice
judgment among Hong Kong Chinese students. TheJournal of Psychology, 125(5),
557-565. Chiu, C-Y. (1991). Hierarchical social relations and justice judgment
among Hong Kong Chinese college students. TheJournal of Social Psychology,
131(6), 885-887. DePaulo, B. M., Kashy, D. A., Kirkendol, S. E., Wyer,
M. W., & Epstein, J. A. (1996). Lying in everyday life. Journal of Personality
and Social Psychology; 70, 979-995. Du, J. S. (1995). Performance of face-threatening
acts in Chinese complaining, giving bad news, and disagreeing. In G. Kasper
(Ed.), Pragmatics of Chinese as native and target language. Honolulu: University
of Hawaii Press. Ekman, P. (1985). Telling lies. NY: Berkley Books.

Galasinski, D. (1994). Deception. Linguist's perspective. Paper presented
at the International

Communication Association convention, Sidney, Australia. Grice, P (1989).
Studies in the way of words. Cambridge, MA: Harvard University Press. Gu,
Y. (1990). Politeness phenomena in modern Chinese. Journal of Pragmatics,
14, 237-257. Jacobs, S., Dawson, E. J, & Brashers, D. (1996). Information
manipulation theory: A replication and assessment. Communication Monographs,
63, 82.

Hofstede, G. (1980). Culture's consequences: International differences
in work related values. Newbury Park: Sage.

Hopper, R., & Bell, R. A. (1984). Broadening the deception construct. Quarterly
Journal of Speech, 70, 288-302.

Hu, H. C. (1944). The Chinese concept of 'face'. American Anthropologist,
46, 45-65. Kasper, G., & Zhang, Y. (1995). It's good to be a bit Chinese:
Foreign students' experience of Chinese pragmatics. In G. Kasper (Ed.),
Pragmatics of Chinese a native and target language. Honolulu: University
of Hawaii Press.

Lapinski, M. K. (1995). Deception and the self A cultural examination of
Information Manipulation Theory. Unpublished master's thesis. Honolulu,
HI: University of Hawaii. McCornack, S. A. (1992). Information manipulation
theory. Communication Monographs, 59, 1-16.

McCornack, S. A., & Levine, T. R. (1990). When lies are uncovered: Emotional
and relational

outcomes of discovered deception. Communication Monographs, 57, 119-138.
McCornack, S. A., Levine, T. R., Morrison, K., & Lapinski, M. (1996). Speaking
of information manipulation: A critical rejoinder. Communication Monographs,
63, 8991.

McCornack, S. A., Levine, T. R., Solowczuk, K. A., Torres, H. I., & Campbell,
D. M. (1992). When the alteration of information is viewed as deception:
An empirical test of information manipulation theory. Communication Monographs,
59, 17-29. Metts, S. (1989). An exploratory investigation of deception
in close relationships. Journal of Social and Personal Relationships, 6,
159-179.

Nishiyama, K. (1993). Japanese negotiators: Are they deceptive or misunderstood?
Paper presented at the 23rd annual convention of the Communication Association
of Japan.

O'hair, D., Cody, M. J., Wang, X. T., & Chao, E. Y. (1990). Vocal stress
and deceptions detection among Chinese. Communication Quarterly, 38, 158-169.


Saarni, C., & Lewis, M. (1993). Deceit and illusion in human affairs. In
M. Lewis & C. Saarni (Eds.), Lying and deception in everyday life. New
York: The Guilford Press. Tang, S. F Y., & Kirkbride, P. S. (1986). The
development of conflict handling skills in Hong Kong Some cross-cultural
issues. Working Paper No. 7, City Polytechnic of Hong Kong. Turner, R.
E., Edgley, C., & Olmstead, G. (1975). Information control in conversations:
Honesty is not always the best policy. KansasJournal of Sociology, 11,
69-89. Yang, K-S. (1986). Chinese personality and its change. In M. Bond
(Ed.). The psychology of the Chinese people. Hong Kong: OU.

Yum, J. O. (1988). The impact of Confucianism on interpersonal relationships
and communication patterns in East Asia. Communication Monographs, 55,
374-388.

LORRITA N. T. YEUNG, TIMOTHY R. LEVINE, and KAZUO NISHIYAMA

Lorrita N. T. Yeung (Ph.D., Macquarie University [Australia], 1994) is
an associate professor and Director of the Language Centre at Lingnan College,
Hong Kong. Timothy R. Levine (Ph.D., Michigan State University, 1992) is
an associate professor and Kazuo Nishiyama (Ph.D., University of Minnesota,
1970) is a professor in the Department of Speech at the University of Hawaii
at Manoa.

Reproduced with permission of the copyright owner.
Further reproduction or distribution is prohibited without permission.



----------------------------------------------
 International deception
Personality and Social Psychology Bulletin
Thousand Oaks
Mar 2000

Authors:                  Charles F Bond Jr

Authors:                  Adnan Omar Atoum

Volume:                   26

Issue:                    3

Pagination:               385-395

ISSN:                     01461672

Subject Terms:            Lying
                          Cross cultural studies
                          Cognition & reasoning
                          Social psychology


Abstract:

Three studies of international deception involving Americans, Jordanians
and Indians were conducted. Ancillary results reveal that people from diverse
backgrounds reach consensus in deception judgments and that motivation
can impair a liar's ability to achieve communication goals.
Copyright Sage Publications, Inc. Mar 2000

Full Text:

This article reports three studies of international deception. Americans,
Jordanians, and Indians were videotaped while lying and telling the truth,
and the resulting tapes were judged for deception by other Americans, Jordanians,
and Indians. Results show that lies can be detected across cultures. They
can be detected across cultures that share a language and cultures that
do not, by illiterates as well as university students. Contrary to a hypothesis
of ethnocentrism, perceivers show no general tendency to judge persons
from other countries as deceptive; in fact, they often judge foreigners
to be more truthful than compatriots. There is, however, some evidence
for a language-based ethnocentrism when perceivers are judging the deceptiveness
of a series of people from the same multilingual culture. Ancillary results
reveal that people from diverse backgrounds reach consensus in deception
judgments and that motivation can impair a liar's ability to achieve communication
goals.

Preception has been defined as an "act that is intended to foster in another
person a belief or understanding which the deceiver considers false" (Zuckerman,
DePaulo, & Rosenthal, 1981). judgments of deception have important consequences.
Some may even start wars (Triandis, 1994).

Psychologists have studied judgments of deception from behavior. They have
identified some nonverbal cues to deceit (Ekman, 1992) and discovered a
number of patterns in naive observers' attempts to spot lies (Zuckerman
et al., 1981). Unfortunately, most of the research on deception has been
restricted to the United States. Although a little has been learned about
deception in other countries (Aune & Waters, 1994; Cody, Lee, & Chao, 1989;
Feldman, 1979), there has been only one experimental study of international
deception to date.

Bond, Omar, Mahmoud, and Bonser (1990) investigated international deceptions
between Jordanians and Americans. Jordanian and American students were
videotaped while telling lies and truths; later, other Jordanian and American
students watched the tapes and tried to spot deceit. Results showed statistically
significant lie detection within each of the two cultures but no lie detection
between cultures.

In implying that lies cannot be detected across cultures, the Bond et al.
(1990) results would seem to have important implications. Theoretically,
these results suggest that the ability to detect deception reflects culturespecific
learning. At a practical level, they suggest that in international settings,
liars are rarely caught.

It would be premature, however, to infer from a single study that cross-cultural
lie detection is impossible. Even within a culture, lies are difficult
to detect. In the American research literature, rates of monocultural lie
detection rarely exceed 55%, when guessing would produce 50% detections
(Kraut, 1980). Thus, one could hardly expect that lies would be easy to
detect across cultures or that international lie detection abilities would
be easy to observe. The Bond et al. (1990) study may have failed to uncover
international lie detection for several reasons: They required people to
judge deception solely from visible cues, restricted their research to
students, and gave these people no motivation to lie. In principle, these
factors might influence international deception judgments, as will now
be explained.

Often, international lies are encountered in face-toface meetings, where
the liar can be seen and heard. However, in the only existing study of
international deception (Bond et al., 1990), people were forced to judge
deception from a video presentation with no sound. This unnatural perceptual
mode may have undermined judges' attempts at lie detection. Consistent
with this analysis, previous research shows that Americans are more accurate
at detecting Americans' lies if they can hear the liars in addition to
seeing them (Zuckerman et al., 1981).

International deceptions can involve individuals from diverse backgrounds.
Participants in deception research are, by contrast, homogeneous. Almost
invariably, they are students. Bond et al. (1990), for example, had university
students in one country lie to university students in another country.
Perhaps attempts at lie detection depend on the similarity of the liar
to the target of deception. Perhaps research-based conclusions about deception
judgments reflect university students' youth, wealth, and higher education
(cf. Sears, 1986). In light of these possibilities, no general conclusions
about international deception should be drawn from university students'
attempts to detect university students' lies.

The stakes in an international deception can be high (Ekman, 1992). In
most research on deception, the stakes are by contrast low. In the Bond
et al. (1990) study of international deception, for example, participants
were motivated solely by their desire to fulfill a psychology course requirement.
Perhaps people regard experimental deception tasks as a form of acting,
devoid of any consequences. Perhaps they feel none of the arousal that
highly motivated liars experience and, hence, display none of the cues
that would otherwise give them away. Consistent with this line of reasoning,
Americans are most likely to "leak" nonverbal cues to deception when highly
motivated to conceal their lies (DePaulo & Kirkendol, 1989).

The current article reports three studies of international deception. In
Experiment 1, American and Jordanian university students attempt to detect
international lies from an audiovisual presentation; in Experiment 2, illiterate
Indian farm workers attempt to detect American and Jordanian university
students' lies; and in Experiment 3,judges from three countries seek to
detect Indians' motivated lies. These studies seek to determine whether
lies can be detected across cultures.

Judgments of deception need not be accurate to have important effects.
Indeed, Triandis (1994) maintains that the Persian Gulf War resulted when
an Iraqi official mistakenly concluded that an American negotiator was
lying. Thus, the present experiments will consider not only the accuracy
of international lie detection but also international biases in judging
deceit. Previous research shows that Americans show a bias toward perceiving
other Americans as truthful (DePaulo, Stone, & Lassiter, 1985). In considering
the sorts of biases that might color international deception judgments,
we considered two possibilities. Perhaps people give the benefit of the
doubt to communicators they do not understand. If so, people might be reluctant
to judge foreigners as deceptive. Perhaps, on the other hand, people are
suspicious of outsiders. If so, ethnocentric stereotypes (Smith & Bond,
1994) might encourage them to judge foreigners as dishonest. In three experiments,
we will assess these possibilities by examining American, Jordanian, and
Indian tendencies to attribute deception to foreigners and compatriots.


EXPERIMENT 1: VISIBLE AND AUDIBLE LIES

In principle, international subterfuge might be uncovered from a variety
of cues. Eye contact, smiles, and head nodding can be seen; speech rate,
volume, and tone of voice can be heard. Although researchers have detailed
the impact of many such cues on Americans' judgments of Americans' veracity
(cf. Zuckerman et al., 1981), in the international arena, some complications
emerge. Negotiators who are ignorant of an adversary's language may attach
special significance to nonverbal cues. Yet, these negotiators must be
cognizant of crosscultural differences, lest they interpret foreign mannerisms
as evidence of deception (cf. Bond et al., 1992).

An initial study was conducted to analyze the impact on international lie
detection of visible and audible cues. American and Jordanian university
students attempted to detect one another's lies from one of three presentations:
a video presentation of the liar's face and body, an audio presentation
of the liar's speech, or an audiovisual presentation of both visible and
audible cues. Americans have difficulty detecting other Americans' lies
when they must base their judgments solely on what they can see. They are
better at detecting lies if the liar also can be heard (DePaulo et al.,
1985). Experiment 1 will determine whether similar effects obtain in American
and Jordanian university students' attempts at international lie detection.


Ordinarily, liars attempt to conceal their deceptions, hoping to gain an
advantage over an adversary (Bond, Kahler, & Paolicelli, 1985). In this
respect, liars differ from other communicators, who seek to transmit information
faithfully. In deference to the adversarial nature of deceptive interactions,
most of the liars in the present research were instructed to conceal their
deceit. To allow for a comparison with attempts at faithful information
transmission, a few liars were given a different communication goal: to
convey to others the fact that they were lying. Earlier research indicates
that Americans have some ability to convey deception to other Americans
(Zuckerman, DeFrank, Hall, Larrance, & Rosenthal, 1979). We wondered whether
cross-cultural attempts at conveying deception also might be efficacious.


METHOD

Research Participants

American and Jordanian university students participated in Experiment 1
by judging deception. The Americans were 89 female and 31 male psychology
students at Texas Christian University. The Jordanians were 30 female and
30 male psychology students at Yarmouk University in Jordan.

Videotapes

The participants judged videotapes that had been made by Bond et al. (1990).
These depicted Americans and Jordanians lying and telling the truth. On
the videotapes were 20 male and 20 female American psychology students
from Texas Christian University as well as 20 male and 20 female Jordanian
psychology students from Yarmouk University. Throughout the experimental
procedure, the Americans spoke in English; the Jordanians spoke in Arabic.


At the time the tape was made, a student sat facing a male research assistant
from the student's culture and a videotape camera was located over the
assistant's right shoulder. Students were then asked to describe a person
they knew. They were asked to describe either (a) a person they liked,
(b) a person they disliked, (c) a person they liked as if they really disliked
that person, or (d) a person they disliked as if they really liked that
person. Students were instructed either to tell the truth (if giving one
of the first two descriptions above) or to lie (if giving one of the latter
two descriptions).

After giving the initial description, the student was asked for a second
person description. Over the course of the videotaping session, each student
gave all four of the person descriptions described above, with the order
of the descriptions counterbalanced across students. For a similar procedure,
see DePaulo and Rosenthal (1979).

Students' communication goal was manipulated. On videotape, each student
told the truth and lies. While telling the truth, all of the students were
instructed to convince the research assistant that they were telling the
truth. While lying, most of the students (32 Americans and 32 Jordanians)
were instructed to conceal their lies and convince the research assistant
that their descriptions were truthful. The other students (8 Americans
and 8 Jordanians) were instructed to convey their lies and let the research
assistant know that their descriptions were false. The latter were instructed
not to say, in so many words, "I am lying" but were free to expose their
deceit in any other way.

The videotapes created a 2 (conceal vs. convey lie) x 2 (tell lie vs. tell
truth) factorial experiment. Although these videotapes had been made for
an earlier crosscultural study (Bond et al., 1990), none of the current
judges had participated in the previous research.

In all, 320 person descriptions were solicited (four from each of 80 students:
40 Americans and 40jordanians). These were edited onto four videotapes.
Each videotape depicted one description from each of the 80 students. Forty
of the descriptions were lies and 40 were truths. Of the 40 lies on each
videotape, 32 were lies that the students had tried to conceal and 8 were
lies that the student had tried to convey. To reduce the length of participants'judgment
task, the videotape depicted only the first 30 seconds of each person description.


Procedure

Visually isolated from one another in groups of five, research participants
were presented with a tape of people describing acquaintances. As each
description was presented, the participant tried to determine whether it
was the truth or a lie. Immediately after the description, participants
indicated their binary lie-or-truth judgment on a written form.

In response to one of the four videotapes described above, participants
judged the veracity of 80 person descriptions. The tapes were presented
in one of three modalities. One third of the participants judged deception
from an audiovisual presentation of the tape, one third from an audio-only
presentation, and one third from a video-only presentation. Each tape was
judged in each modality by 10 Americans and 5 Jordanians. The segments
on each tape were presented in one of two random orders.

RESULTS

Judges were asked about their language abilities. None of the 120 American
judges claimed to know Arabic. Of the 60 Jordanian judges, 59 claimed to
know English.

Lie Detection Within and Between Cultures

In the current study, each participant judged 40 lies and 40 truths. Half
of the lies and half of the truths had been told by a member of the judge's
culture; the other half had been told by a member of another culture. To
test for lie detection, we noted the percentage of lie/truth judgments
made by each of the 180 judges and compared the mean percentage correct
to the 50% that would be expected by chance. We wondered whether it is
possible to detect lies across cultures. It is. Overall, our participants'
detection accuracy was 51.25% across cultures, t(174) = 2.06,p<.05, and
54.27% within cultures, t(174) = 7.03, p < .0001.

An analysis of variance was conducted to identify factors that influence
lie detection. This was a 3 (modality: audiovisual, audio only, or video
only) x 2 (judge's culture: American vs. Jordanian) X 2 judgment status:
of target from same culture or other culture) x 2 (liar's goal: conceal
or convey lie) mixed-model ANOVA on the percentage of correct lie/truth
judgments. Results revealed that judgments were more accurate within cultures
than across cultures, F(1, 174) = 14.35, p<.001, that lies that targets
had tried to convey could be more readily discriminated from truths than
those that targets had tried to conceal, F(1, 174) = 24.43, p <.0001, and
that lie detection was less accurate when attempted from a video-only rather
than an audiovisual or an audio-only presentation; for the main effect
of modality, F(2, 174) = 16.97, p < . 001. The ANOVA also showed that the
liar's goal had its biggest effect on detection accuracy when lies were
judged in the audiovisual rather than the audio or video mode, Modality
x Goal interaction, F(2, 174) = 6.76, p < .005; that the liar's goal had
a bigger effect on the accuracy of judges from the target's own culture
rather than judges from the other culture, F(1, 174) = 4.69, p < .05; and
that as targets, Americans were more successful than Jordanians at conveying
deception. This final effect produced a two-way Judge's Culture X Judgment
Status interaction, F(1, 174) = 7.67, p < .01, and a three-way judge's
Culture X Judgment Status x Target's Goal interaction, F(1, 174) = 9.30,
p < .01. No other effects in the ANOVA were statistically significant.
Relevant means and t tests appear in Table 1. Across cultures, truths could
be discriminated from lies that a target had attempted to convey when judgments
were made from an audiovisual presentation. Within cultures, truths could
usually be discriminated from lies, except when judgments were made from
video only.

Experiment 1 provides evidence for lie detection across a language as well
as across cultures. American judges reported that they could not understand
Arabic. However, from an audiovisual presentation, these judges could discriminate
Jordanians' Arabic-language lies from truths (M correct= 53.30%), t(39)
= 2.97, p < .01.1 Indeed, Americans' judgments of Jordanians were significantly
more accurate when made from the audiovisual than from the video-only presentation
(M correct for the latter = 49.31 %), for the difference, F(1, 174) = 4.66,
p < .05. Access to Arabic speech facilitated lie detection by students
who did not know Arabic.

Judgmental Biases

In principle, ethnocentrism might encourage people to stereotype foreigners
as dishonest. Experiment 1 does not support this hypothesis. Americans
gave foreigners the benefit of the doubt in judging more Jordanians than
Americans to be truthful (Ms = 59.55% vs. 52.07%, respectively), F(1, 174)
= 36.13, p < .0001. Jordanians judged as truthful just as many Americans
as Jordanians; for the difference, F(1, 174) =.36, ns. In general, foreigners
received the benefit of the doubt only if theycould be heard. Research
participants judged as truthful 58.06% of foreigners and 53.24% of compatriots
who had been depicted in the audiovisual presentation, 61.43% of foreigners
and 49.14% of compatriots in the audio-only presentation, and 51.20% of
foreigners and 52.29% of compatriots in the video-only presentation. In
an ANOVA on percentage truth judgments, these patterns produced an interaction
between judge's culture (American vs. Jordanian) and the status of the
judgment (to own vs. other culture), F(1, 174) = 8.93, p < .005, as well
as an interaction between presentation modality and the judgment's status,
F(2, 174) = 14.60, p <.001.

DISCUSSION

Experiment 1 provides the first evidence to date of lie detection across
cultures, Although it is not easy to detect lies across cultures, neither
is cross-cultural lie detection impossible. International lie detection
seems to require an audiovisual exposure to the liar, one that had not
been available in earlier research. Experiment 1 suggests that Americans
judge foreigners to be more truthful than fellow Americans. Hearing someone
speak in an unfamiliar language may encourage judges to acknowledge their
ignorance of the speaker's culture. Then judges give a speaker the benefit
of the doubt. These Jordanians, knowing English, showed no such tendency
in judging American speakers.

Unfortunately, Experiment 1 has limitations. All of the participants of
the experiment were students of higher education, and they were judging
other students of higher education. Worldwide, university mores may encourage
tolerance of cultural differences. If so, students may be uniquely inclined
to give foreigners the benefit of the doubt. Worldwide, university students
are similar in age, economic status, and educational background. Although
Experiment 1 provides evidence of lie detection within an elite, geographically
dispersed "culture" of higher education, it need not imply that individuals
from radically different backgrounds could detect one another's lies.

EXPERIMENT 2: THE "CULTURE" OF HIGHER EDUCATION

We designed a second study to provide a more stringent test for international
lie detection. In Experiment 2, videotapes of American and Jordanian university
students were judged by Indians. Some of these Indian judges were university
students, whereas others were illiterate farm workers. If international
lie detection is confined to an elite culture of higher education, illiterate
Indian farm workers should show no ability to detect American and Jordanian
university students' lies. If only the highly educated give foreigners
the benefit of the doubt, illiterates should be more willing than university
students to judge foreigners as deceptive.

[IMAGE TABLE] Captioned as: TABLE 1:

While investigating international lie detection, Experiment 2 also provides
an assessment of the nonverbal abilities of illiterates. Illiterates comprise
roughly a third of the world's adult population (Tresserras, Canela, Alvarez,
Sentis, & Salleras, 1992) but are rarely studied by social psychologists
Jahoda, 1979). Illiterates are of theoretical interest because they can
illuminate the impact of schooling on various competencies. In principle,
one might expect illiterates to have exceptional judgmental abilities.
Perhaps illiterates are uniquely equipped to interpret nonverbal behavior
in possessing a holistic style of reasoning that education would undermine
(Rogoff, 1980). Moreover, illiterates have a special investment in face-to-face
behavior because they rely exclusive on nonwritten communications. Experiment
2 will test these possibilities.

METHOD

Research Participants

The participants were 120 residents of Maharashtra, a state in western
India. Sixty of the participants (29 females, 31 males) were English-speaking
psychology students at the University of Pune in India. The other 60 participants
(15 females, 45 males) were farm workers from Bakori village, an isolated
agricultural community 30 miles from Pune. The villagers spoke the Indian
language Marathi. As farm laborers, these villagers earned 75 cents a day.
Although the villagers had seen Indian television, few had ever met a non-Indian.


Procedure

Visually isolated in groups of five, the participants were presented with
a videotape of Americans and Jordanians describing acquaintances. As each
videotape segment was presented, participants indicated whether the person
on the tape was lying or telling the truth. University students, who were
seated in a classroom, indicated their binary lie-or-truth judgments in
writing on a response form. Farm workers, who were seated on the floor
of the village Hindu temple, indicated their judgments nonverbally by turning
a thumb up if they thought that the person on the videotape was telling
the truth or a thumb down if they thought that the person was lying.

In response to one of the four videotapes described in Experiment 1 above,
each participant judged the veracity of 80 students (40 Americans and 40
Jordanians). As before, the tapes were presented in one of three modalities:
audiovisual, audio only, or video only. The segments on each tape were
presented in one of two random orders. Each videotape was judged in each
modality by 5 Indian students and 5 Indian farm workers.

RESULTS

All 60 of the Indian students had completed bachelor's degrees and were
enrolled in a master's program. Each of these students knew English. None
of the 60 farm workers had any higher education, and 26 acknowledged that
they were illiterate. Although the other 34 farm workers claimed to be
literate, many had difficulty signing their names. Nine of the farm workers
claimed to know English but none could understand the American English
spoken by the first author of the current article. None of these 120 Indian
judges claimed to know Arabic.

Lie Detection by Students and Illiterates

Experiment 1 indicates that it is possible for Americans and Jordanians
to detect lies across cultures. As a second test for international lie
detection, we had each of 120 Indians judge 40 lies and 40 truths told
by nonIndians. Again, there was evidence of cross-cultural lie detection.
The Indian participants averaged 51.08% correct lie/truth judgments, which
is more than the 50% expected by chance, t(114) = 2.27, p < .025.

Perhaps lies can be detected internationally only by judges who are similar
to the liars. Experiment 2 does not support this hypothesis. In fact, illiterate
Indian farm workers were just as successful as Indian university students
at detecting American and Jordanian university students' lies. In a 3 (presentation
modality) X 2 (liar's goal: conceal vs. convey lie) x 2 judge's subculture:
student vs. farm worker) ANOVA on percentage correct lie/truth judgments,
the judges' subculture had no significant effects--for the main effect,
F(1, 114) =.01.

In Experiment 1, lie detection depended on the modality in which lies were
presented, with international lie detection evident only from an audiovisual
presentation. Those results were replicated here. These Indians could detect
lies across cultures only if they were judging an audiovisual presentation
(M correct lie/truth judgments = 51.90%), t(39) = 2.63, p <.025. They could
not detect non-Indian lies from either audio only or video only Ms = 50.63%
and 50.71 % correct lie/truth judgments, respectively), t(39) = .73 and
..83, both ns. In the analysis of variance, these differences produced a
main effect of modality, F(2,114) =3.53,p<.05,but no interactions involving
modality.

Experiment 1 suggested that people can convey their lies across cultures
but that they need not fear crossculture exposure of lies that they wish
to conceal. Experiment 2 yields different results. Here, detection accuracy
was just as strong for concealed lies as for conveyed lies; for the relevant
main effect in the ANOVA on percentage correct judgments, F(1, 114) = 1.84,
ns. Indeed, Indian judges could discriminate from truths lies that non-Indians
intended to conceal (M correct lie/truth judgments = 51.35%), t(114) =
2.49, p<.025, but not lies that non-Indians intended to convey (M = 50.01%),
t(114) = .01, ns.

Judgmental Biases of Students and Illiterates

In judging deception, American students give Jordanians the benefit of
the doubt, as Experiment 1 shows. We had imagined that Indian university
students would show this same judgmental bias and wondered whether illiterates
might show an opposite bias-toward ethnocentric suspicion of outsiders.
In fact, Indian university students showed some tendency to perceive Americans
andjordanians as truthful, but illiterates gave these foreigners a stronger
benefit of the doubt. Although 53.21% of the non-Indians were judged as
truthful by Indian university students, 62.72% were judged as truthfal
by Indian illiterates; for the difference, F(1, 114) = 35.58, p < .0001.
These biases were unaffected by the target's culture and by the modality
in which the foreigners were perceived; each F yields p > 10.

DISCUSSION

Individuals from starkly different backgrounds are able to detect one another's
lies. International lie detection is not confined to an elite culture of
higher education or to falsehoods that people wish to convey. The findings
of Experiment 2 suggest that illiterates are neither more nor less successful
at international lie detection than university students but that illiterates
are more inclined to give foreigners the benefit of the doubt. Although
we might be tempted to draw generalizations from these first two experiments,
some limitations of the research remain to be addressed.

As a test of the impact of communication goals on international lie detection,
the first two experiments have limitations. In these studies, judges encountered
only one fourth as many conveyed as concealed lies, and targets had no
special incentive. Human abilities to convey lies would be more fairly
assessed in a balanced research design, and the research would be more
relevant to high-stakes deceptions if participants were given more motivation
to lie.

As a study of illiterates' deception judgments, Experiment 2 has limitations.
This experiment suggests that illiterates are more willing than students
to give foreigners the benefit of the doubt, but it does not indicate the
scope of this bias. Relative to students, illiterates also might be more
trusting of compatriots. Although Experiment 2 indicates that illiterates
are no better than students at detecting foreigners' lies, it need not
imply that the two sets of judges have equal judgmental abilities. If in
their daily interactions with compatriots illiterates had acquired special
skills, these need not have helped them judge foreigners.

Together, the first two experiments suggest that language differences have
little impact on lie detection. Thus, Americans (who do not understand
Arabic) can detect Arabic-language lies and Indian farm workers (who understand
only Marathi) can detect lies told in Arabic and English. Yet, these experiments
confound language with culture. A clearer study of language differences
could be conducted within a multilingual country. There, one could compare
a perceiver's judgments of two sorts of compatriots-those who are lying
in a language that the perceiver understands and those who are lying in
another language.

EXPERIMENT 3: LANGUAGE, SUBCULTURE, AND MOTIVATION

To complement the first two experiments, we designed a third study. Having
investigated judgments of American and Jordanian liars, we videotaped Indians
telling lies and truths. Some of the Indians on the videotape spoke in
English; others spoke in the Indian language Marathi. Half of the Indians
on the videotape attempted to conceal their deceptions and half attempted
to convey that they were lying. Videotapes of these Indian lies and truths
were then judged for veracity by American students, Jordanian students,
Indian students who understood English, and Indian illiterates who did
not understand English. This final study sought to clarify the impact of
communication goals, schooling, and language differences on international
deception.

Although most of the lies told in everyday life may be relatively inconsequential
(DePaulo, Kashy, Kirkendol, Wyer, & Epstein, 1996), psychologists are fascinated
by deceptions that involve high stakes (Frank & Ekman, 1997). To accommodate
psychological interest in motivated deceptions, we gave some of the participants
of Experiment 3 financial incentives to lie. Motivation impairs Americans'
ability to lie. DePaulo and Kirkendol (1989) report that Americans' lies
can be more readily discriminated from truths if the Americans are highly
motivated to conceal a lie than if they are relatively unmotivated. Experiment
3 will determine whether in attempts to conceal deception, Indians also
suffer a motivational impairment effect. The experiment will extend earlier
efforts by studying the effect of motivation on Indians' attempts to convey
that they are lying.

METHOD

Research Participants

The participants were 120 American students from Texas Christian University
(60 female, 60 male), 60jordanian students from Yarmouk University (30
female, 30 male), and 120 Indians (30 female and 30 male students from
the University of Pune; 15 female and 45 male farm workers from Bakori
village).

Videotapes

The participants judged videotapes of Indians lying and telling the truth.
These Indian videotapes were modeled on the tapes of Americans and Jordanians
described in Experiment 1 above (cf. DePaulo & Rosenthal, 1979). The videotapes
depicted 32 female and 32 male residents of Pune, India, who had responded
to a newspaper advertisement. While being videotaped, half of the Indians
spoke in English and half spoke in the Indian language Marathi.

At the time that the tape was made, an Indian sat facing an Indian male
research assistant and a videotape camera located over the assistant's
right shoulder. Indians were then videotaped while describing (a) a person
they liked, (b) a person they disliked, (c) a person they liked as if they
really disliked that person, and (d) a person they disliked as if they
really liked that person. The order of the descriptions was counterbalanced
across participants. While telling the truth, all of the participants were
instructed to convince the research assistant that they were telling the
truth. While lying, 16 of the English-speaking Indians and 16 of the Marathispeaking
Indians were instructed to conceal their lies and convince the research
assistant that their descriptions were truthful. The others were instructed
to convey their lies and let the research assistant know that their descriptions
were false without saying, in so many words, "I am lying."

The Indians' motivation to lie was experimentally manipulated. Half of
the participants were given no incentive to achieve their communication
goal; the other half received a financial incentive. The latter could make
20 Indian rupees (Rs. 20) for each of their four person descriptions. They
made Rs. 20 each time that they convinced the research assistant that a
truthful person description was, in fact, truthful. They also made Rs.
20 for each lie if they were successful in achieving their assigned goal-either
to conceal or convey the lie. After observing each person description for
which a financial incentive had been offered, the male research assistant
(who was unaware of the order of the descriptions) guessed aloud whether
the participant was lying or telling the truth. The Rs. 20 payment was
then handed to the participant if the latter had succeeded in achieving
their goal (of conveying the truth, conveying a lie, or concealing a lie).
By giving four successful person descriptions, the participant could earn
up to Rs. 80 in 30 minutes. At the time of the study, Rs. 80 was approximately
I day's pay for an Indian university professor.

In all, 256 Indian person descriptions were solicited. These were edited
onto four videotapes. Each videotape depicted 64 descriptions: one from
each of the 64 participants-four descriptions in each of the cells of a
2 (tell lie vs. tell truth) x 2 (conceal lie vs. convey lie) x 2 (English
vs. Marathi language) x 2 (incentive vs. no incentive) factorial design.
Videotapes depicted the first 45 seconds of each person description.

Procedure

Visually isolated in groups of five, the participants were presented with
a videotape of 64 Indians describing acquaintances. As each videotape segment
was presented, participants indicated whether the person on the tape was
lying or telling the truth. All of the university students indicated their
binary lie-or-truth judgments in writing on response forms. The Indian
farm workers turned a thumb up if they thought that the person on the tape
was telling the truth and a thumb down if they thought that the person
was lying.

In response to one of the four Indian videotapes, each participant judged
the veracity of 64 Indians. The tapes were presented in one of three modalities:
audiovisual, audio only, or video only. The segments on each tape were
presented in one of two random orders. Each videotape was judged in each
modality by 10 Americans, 5jordanians, and 10 Indians (5 university students
and 5 farm workers).

RESULTS AND DISCUSSION

After judging videotapes of Indians speaking in Marathi and English, participants
reported on their language abilities. All 120 Indian judges, but none of
the American or Jordanian judges, reported that they knew Marathi. Knowledge
of English was claimed by all of the American judges, all of the Jordanian
judges, all 60 of the Indian university students, and 4 of the 60 Indian
farm workers. Of the 60 farm workers, 26 stated that they were illiterate.
Although the other 34 claimed to be literate in Marathi, many had difficulty
signing their names.

Lie Detection Within and Between Cultures

Our first two experiments indicate that American and Jordanian lies can
be detected across cultures. We wondered whether there might also be evidence
for the cross-cultural detection of Indian lies. There was. Each judge
in Experiment 3 was presented with a videotape of 32 Indians lying and
32 Indians telling the truth. An analysis revealed that non-Indians who
judged these tapes averaged 51.38% correct lie/truth judgments, which is
greater than the 50% that would be expected by chance, t(179) = 2.81, p
< .005.

American and Jordanian lies could be detected more accurately by members
of the liar's culture than by members of other cultures. This result did
not generalize to Indian lies. In fact, in detecting Indians' lies, Indians
were no more accurate than non-Indians; for the main effect of judge's
culture (Indian vs. non-Indian) in an ANOVA on percentage correct lie/truth
judgments, F(1, 294) = .0002, ns. Non-Indian judges averaged 51.38% correct
detections of Indians; Indian judges averaged 51.39%.

The earlier experiments indicate that lie detection abilities depend on
the modality in which lies are judged. In particular, American and Jordanian
lies can be detected across cultures only if they are judged from an audiovisual
presentation, whereas these lies can be detected within cultures from either
an audiovisual or audio presentation. Here, modality has a main effect
on judges' accuracy in detecting Indians' lies and truths, F(2, 294) =
6.14, p < .005, with statistically significant lie/truth discrimination
evident from the audiovisual presentation (M correct = 53.14%), t(99) =
5.11, p < .0001, but not the audio-only or video-only presentations (Ms
= 50.77% and 50.25% correct, each p > .20). This pattern was the same for
Indian and non-Indian judges-for the Modality x judge Culture interaction,
F(2, 294) =.40, ns.

The earlier studies found evidence for lie detection across languages as
well as cultures. To clarify the role of language, in the current study
we had Indians lie in two different languages-English and Marathi. Language
had no significant effects on the percentage of correct lie/truth judgments
in a 2 (English vs. Marathi language) X 2 (conceal vs. convey lie) x 2
(incentive vs. no incentive) x 2 (Indian vs. non-Indian judge) x 3 (modality)
ANOVA (p > .05 for the main effect of language and every interaction involving
language). However, when making judgments from audio only, Indian farm
workers were less successful than Indian university students in discriminating
Indian English-language lies from Indian English-language truths (M= 47.17%
vs. 52.16% correct judgments, respectively), F(1, 114) = 4.48, p < .05.
This difference may reflect the farm workers' ignorance of English and
help explain why Indians as a whole did no better than non-Indians at detecting
Indian lies.

The Liar's Goal and Motivation

Americans' attempts at concealing deception are subject to a motivational
impairment effect (DePaulo & Kirkendol, 1989). We wondered whether motivation
would impair Indians' attempts to conceal deception and whether motivation
also might have some effect on Indians' attempt to convey that they are
lying. An ANOVA on percentage correct lie/truth judgments reveals that
the impact of motivation depends on the judge's culture as well as the
liar's goal: Goal x judge's Culture x Incentive interaction, F(1, 294)
= 3.97, p < .05. Means relevant to this interaction appear in Table 2.
Follow-up analyses revealed two motivational impairment effects. In particular,
Indian judges could discriminate from truths lies that Indians had received
an incentive to conceal (M accuracy = 52.67%), t(119) = 2.27, p < .05,
as well as lies that Indians had received no incentive to convey (52.70%
accuracy), t(119) = 2.42, p<.05. They were not able to discriminate from
truths lies that Indians had received no incentive to conceal or lies that
Indians had received an incentive to convey (accuracy rates = 50.64% and
49.56%, respectively, ns). This produced a simple Liar's Goal x Incentive
interaction on the percentage of lies and truths correctly judged by Indians,
F(1, 294) = 3.96, p < .05. Thus, in judgments made by compatriots, motivation
impaired Indians' ability to conceal and convey lies.

Non-Indians' judgments were unaffected by the Indians' motivation to lie.
Non-Indian judges were, however, more accurate in discriminating from truths
lies that Indians intended to convey than ones they intended to conceal
(M accuracy rates = 52.34% vs. 50.43%, respectively), F(1, 294) = 5.36,
p < .05.

Judgmental Biases

Our studies of American and Jordanian deceptions suggest that people give
foreigners the benefit of the doubt. They also indicate that Indian illiterates
are more likely than Indian university students tojudge foreigners as truthful.
To determine whether similar results would be evident in judgments of Indians'
deceptiveness, we conducted a 2 (English vs. Marathi language) x 3 (judge
sample: Indian illiterate vs. Indian student vs. nonIndian student) x 3
(modality) ANOVA on percentage truth judgments. The ANOVA revealed no general
tendency for non-Indians to give Indians the benefit of the doubt. Instead,
it revealed a Language x Judge Sample x Modality interaction, F(4, 291)
= 5.84, p < .001, with an effect of language on illiterates' tendency to
attribute truthfulness to Indians whom they could hear. The farm workers
attributed less truthfulness to Indians whom they heard speaking English
than to Indians whom they heard speaking the farm workers' language Marathi
(judging as truthful 52.55% vs. 66.42% of such individuals, respectively),
F(1, 291) = 33.07, p <.0001. No language-based discrimination was evident
in Indian students' judgments of truthfulness. Non-Indians were more likely
to judge as truthful Indians whom they heard speaking in English than Indians
whom they heard speaking in Marathi (judging 54.19% vs. 50.83% of such
individuals as telling the truth), F(1, 219) = 5.84, p < .05. Predictably,
the language in which a person description was offered had no significant
effect on judgments made from video only (all such effects yield p > .20).


These results may reflect a language-based ethnocentrism. Both Indian farm
workers (who knew Marathi but not English) as well as non-Indians (who
knew English but not Marathi) placed more trust in Indians who spoke a
familiar language than those who spoke an unfamiliar language.

CUMULATIVE ANALYSES

Having conducted three similar experiments, we now report a few cumulative
analyses. One analysis provides an omnibus answer to a question of special
significance. From an audiovisual presentation, can people detect lies
that foreigners have attempted to conceal? Cumulative results show that
they can. Across our three experiments, 160 judges attempted audiovisual
detection of crossculturally concealed lies. In these cross-cultural judgments,
they achieved a lie/truth discrimination accuracy rate of 51.66%, which
is greater than the 50% that would be expected by chance, t(159) = 2.76,
p <.01. Lies that a foreigner attempts to conceal cannot be uncovered from
a video-only presentation (cross-cultural accuracy rate = 49.92%, ns) and
may (or may not) be uncovered from an audio-only presentation (the latter
accuracy rate 51.19%, two-tailed p = .054).

Other cumulative analyses were designed to follow up on patterns established
in earlier research. DePaulo and Rosenthal (1979) found that to Americans,
some Americans appear honest even when they are lying and others appear
dishonest even when they are telling the truth. Statistically, this "demeanor
bias" reflects a positive correlation between the percentage of judges
who infer deception from a given target's lie and the percentage of judges
who infer deception from that same target's truth.

Similar biases were evident in the judgments studied here. Here,judgments
of a given target's lies were consistent with judgments of that same target's
truths whether the judgments were made from an audiovisual, audioonly,
or video-only presentation and whether the people making the judgments
were from the target's own culture or from other cultures. For the relationship
between percentage deception judgments to a given target's lies and to
that same target's truths, partial rs that control for the target person's
culture were +.25, +.39, and +.37 for audiovisual, audio-only, and video-only
judgments made by people from the target's own culture and +.54,+.29, and
+.54 for audiovisual, audio-only, and video-only judgments made by people
from other cultures (for each of these rs, p <.001). To foreigners as well
as compatriots, some people look honest even when they are lying, whereas
others appear dishonest even when they are telling the truth.

Bond et al. (1985) found that Americans reach substantial levels of consensus
in judging Americans' deceptiveness. Analyses of the current data reveal
that there is cross-cultural consensus in deception judgments as well.
Across the three experiments reported here, participants judged lies and
truths told by 144 target persons (40 Americans, 40jordanians, and 64 Indians).
To assess agreement among judges from different cultures, we noted for
a given target the percentage of American judges, the percentage of Jordanian
judges, and the percentage of Indianjudges who inferred that the target
as lying and intercorrelated these percentages across the 144 target persons
within each of the three modalities. For results, see Table 3. As shown
in Table 3, there is statistically significant cross-cultural consensus
in judgments of deceptiveness whether the judgments are made from an audiovisual,
audio-only, or video-only presentation.

We also found evidence of judgmental consensus across languages. When judging
deception from audiovisual presentations of Jordanians speaking in Arabic,
farm workers agree with American students (r = .34, p < .05), although
none of these judges understand Arabic. When judging deception from audio-only
presentations of Indians speaking in Marathi, Jordanians agree with Americans
(r = .39, p < .05), although none of these judges understand Marathi. To
natives of diverse cultures, some people look and sound dishonest.

GENERAL DISCUSSION

These experiments provide the first evidence to date of lie detection across
cultures. Lies can be detected across cultures that share a language and
across cultures that do not. They can be detected by university students
and by illiterates, too.

[IMAGE TABLE] Captioned as: TABLE 2:

[IMAGE TABLE] Captioned as: TABLE 3:

Although Experiments 1 and 2 suggest that it is harder to detect a foreigner's
than a compatriot's lie, it is noteworthy that the compatriot liars in
the first two studies were the judges' classmates. As student peers, these
judges and liars were more similar to one another than most compatriots
(Sears, 1986) and, hence, might have been uniquely positioned to detect
one another's lies. Experiment 3 was the only study in which participants
judged lies told by nonpeer compatriots. Those lies were as well detected
across cultures as within the culture.

The current results have theoretical implications. They imply that there
are cross-cultural similarities in the way that liars act and that behavioral
concomitants of deception can be identified across cultures. Perhaps liars
throughout the world have common experiences. They may fear exposure or
have difficulty fabricating deceptions (Zuckerman et al., 1981). Perhaps
in all cultures, liars' experiences give rise to the same behaviors, or
to behaviors that convey the same impression. These function as pan-cultural
detection cues.

Visible cues are not sufficient for cross-cultural lie detection, and audible
cues facilitate lie detection even when judges do not understand the liar's
language. This explains why no cross-cultural lie detection was observed
in an earlier study of deception judgments from videoonly displays (Bond
et al., 1990). In suggesting that cross-cultural lie detection is based
on vocal rather than verbal cues, the current results are reminiscent of
earlier demonstrations that Americans can detect Americans' lies from content-filtered
speech (Zuckerman et al., 1979).

Although it is possible to detect lies across cultures, international lie
detection is not easy. Similar to earlier monocultural research (Kraut,
1980), our studies of cross-cultural deception indicate that liars are
often successful in their attempts to appear honest. This may imply that
throughout evolutionary history, deception has been more important to the
deceiver than to the target of deception (Bond et al., 1985) or that targets
of deception rarely receive immediate feedback about their mistakes (DePaulo
et al., 1985).

Facial displays of emotion can be readily recognized across cultures, as
a large research literature suggests (Ekman, 1980). At first blush, people's
strong crosscultural ability to recognize emotions may seem at odds with
their weaker cross-cultural ability to detect lies. However, there is an
underlying consistency. In the earlier literature, the facial displays
that were recognized across cultures were poses of emotions that were not
being felt (Russell, 1994). To us, the earlier literature indicates that
people are adept at feigning unfelt emotions. The present findings reveal
that they also are adept at feigning attraction for acquaintances they
dislike.

Distinct from the ability to detect deception are biases in international
judgments. Often, people regard foreigners with suspicion and mistrust
(Smith & Bond, 1994). Imagining that these ethnocentric stereotypes would
impel people to judge foreigners as deceptive, we were surprised by the
current results. People perceive foreigners as more truthful than compatriots,
especially when the target can be heard. They perceive Indians who are
speaking in an unfamiliar language as more deceptive than those who are
speaking in a language that the perceiver knows.

In our view, these judgmental biases stem from a common source-the judge's
attribution for a communication failure. Attempts at cross-language communication
can be frustrating (Ryan & Giles, 1982). When people fail to understand
a foreigner, they search for an explanation (Smith & Bond, 1994). Sometimes,
listeners attribute communication failures to their own ignorance, giving
speakers the benefit of the doubt. Sometimes, they attribute the failure
to the speaker, thereby externalizing blame. From our first two experiments,
we infer that listeners attribute communication failure to their own ignorance
when confronting a culture in which no one is speaking the listener's language.
This tendency toward self-blame is strongest among the poorly educated.
From Experiment 3, we infer that listeners externalize blame for communication
failures when they confront a culture in which some people are speaking
a familiar language and others an unfamiliar language. Then the latter
can be seen as choosing to miscommunicate.

Future research will be needed to test our interpretations, to survey lies
from other cultures, and to understand cross-cultural deceptive interactions
(Buller & Burgoon, 1996). In the meantime, some conclusions can be drawn.
It is possible to detect lies across cultures. Language and cultural differences
introduce biases into deception judgments. These biases can have international
consequences.

NOTE

1. Throughout this article, one-sample t tests are used to assess the difference
between lie/truth discrimination rates and 50%, the rate that would be
expected by chance. Each t test reported in the article includes in its
error term data from only those research participants whose lie/truth discrimination
is being assessed. For each such test, a corresponding test was conducted
in which the denominator of the t statistic was based on the pooled within-group
error term from an appropriate analysis of variance. Each time a discrimination
rate differs significantly from 50% at p < .05 by the individual-error
t test reported in the article, it also differs significantly from 50%
at p<.05 by the corresponding pooled-error t test.

REFERENCES

Anne, R. K., & Waters, L. L. (1994). Cultural differences in deception:
Motivations to deceive in Samoans and North Americans. Internationaljournal
of Intercultural Relations, 18, 159-172.

Bond, C. Rjr., Kahler, K N., & Paolicelli, L. M. (1985). The miscommunication
of deception: An adaptive perspective. Journal of Experimental Social Psychology,
21, 331-345.

Bond, C. F., Jr., Omar, A., Mahmoud, A., & Bonser, R. N. (1990). Lie detection
across cultures. Journal of nonverbal Behavior, 14,189-204. Bond, C. E
Jr., Omar, A., Pitre, U., Lashley, B. R., Skaggs, L. M., & Kirk,

C. T. (1992). Fishy-looking liars: Deception judgment from expectancy violation.
Journal of Personality and Social Psychology, 63, 969-977.

Buller, D. B., & Burgoon,J. K (1996). Interpersonal deception theory. Communication
Theory, 6, 203-242.

Cody, M. J., Lee, W. S., & Chao, E. Y (1989). Telling lies: Correlates
of deception among Chinese. In J. P. Forgas & M. J. Innes (Eds.), Recent
advances in social psychology: Proceedings of the 24th International Congress
of Psychology (Volume 1). Amsterdam: North-Holland.

DePaulo, B. M., Kashy, D. A., Kirkendol, S. E., Wyer, M. W., & Epstein,
J. A. (1996). Lying in everyday life. Journal of Personality and Social
Psychology, 70, 979-995.

DePaulo, B. M., & Kirkendol, S. E. (1989). The motivational impairment
effect in the communication of deception. In J. C. Yuille (Ed.), Credibility
assessment (pp. 51-70). Belgium: Kluwer Academic.

DePaulo, B. M., & Rosenthal, R. (1979). Telling lies: Deceiving and detecting
deceit. Journal of Personality and Social Psychology, 37, 1713-1722.

DePaulo,B.M.,Stone,J. I.,& Lassiter, G.D. (1985). Deceiving and detecting
deceit. In B. R. Schlenker (Ed.), The selfand social life (pp. 323370).
New York: McGraw-Hill.

Ekman, P. (1980). The face of man: Expressions of universal emotions in
a New Guinea village. New York: Garland STPM Press.

Ekman, P. (1992). Telling lies: Clues to deceit in the marketplace, politics,
and marriage. New York: Norton.

Feldman, R. S. (1979). Nonverbal disclosure of deception in urban Koreans.
Journal of Cross-Cultural Psychology, 10, 73-83.

Frank, M. G., & Ekman, P. (1997). The ability to detect deceit generalizes
across different types of high-stakes lies. Journal of Personality and
Social Psychology, 72, 1429-1439.

Jahoda, G. (1979). A cross-cultural perspective on experimental social
psychology. Personality and Social Psychology Bulletin, 5, 142-148.

Kraut, R. (1980). Humans as lie detectors: Some second thoughts. Journal
of Communication, 30, 209-216.

Rogoff, B. (1980). Schooling and the development of cognitive skills. In
H. C. Triandis & A. Heron (Eds.), Handbook of cross-cultural psychology
(Volume 4, pp. 233-294). Boston: Allyn & Bacon.

Russell,J. A. (1994). Is there universal recognition of emotion from facial
expression? A review of the cross-cultural studies. Psychological Bulletin,
115, 102-141.

Ryan, E. B.,& Giles, H. (1982). Attitudes towards language variation.London:
Edward Arnold.

Sears, D. O. (1986). College sophomores in the laboratory: Influences of
a narrow data base on social psychology's view of human nature. Journal
of Personality and Social Psychology, 51, 515-530.

Smith, P. B., & Bond, M. H. (1994). Socialpsychology across cultures: Analysis
and perspectives. Boston: Allyn & Bacon.

Tresserras, R., Canela, J., Alvarez, J., Sentis, J., & Salleras, L. (1992).
Infant mortality, per capita income, and adult illiteracy: An ecological
approach. American Journal of Public Health, 82, 435-438.

Triandis, H. C. (1994). Culture and social behavior. In W.J. Lonner & R.
Malpass (Ed.), Psychology and culture (pp. 169-173). Boston: Allyn & Bacon.


Zuckerman, M., DeFrank, R. S., Hall, J. A., Larrance, D. T, & Rosenthal,R.
(1979). Facial and vocal cues of deception and honesty. Journal of Experimental
Social Psychology, 15, 378-396.

Zuckerman, M., DePaulo, B., & Rosenthal, R. (1981). Verbal and nonverbal
communication of deception. In L. Berkowitz (Ed.), Advances in experimental
social psychology (Vol. 14, pp. 1-59). New York: Academic Press.

Charles E Bond, Jr.

Texas Christian University, U.S.A.

Adnan Omar Atoum

Yarmouk University, Jordan

Authors' Note: We are grateful to Swati Apte, Sharon Ekhardt, Darryl Johnson,
Michael Myers, M. N. Palsane, and Urvashi Pitre for help with this research.
The first author's contribution was supported by an Indo-American fellowship
from the Indo-U.S. Subcommission on Education and Culture and by a paid
leave of absence from Texas Christian University. Correspondence should
be addressed to Charles F. Bond, Jr., Box 298920, Department of Psychology,
Texas Christian University, Fort Worth, TX 76129, U.S.A.; e-mail: c.bond@tcu.edu.


Reproduced with permission of the copyright owner.
Further reproduction or distribution is prohibited without permission.





The effects of suspicion on the recall of cues used to make veracity judgments
Communication Reports
Salt Lake City
Winter 1998

--------------------------------------------------------------------------------

Authors:                  Murray G Millar

Authors:                  Karen U Millar

Volume:                   11

Issue:                    1

Pagination:               57-64

ISSN:                     08934215

Subject Terms:            Memory
                          Psychology
                          Communication
                          Research


Abstract:

A study examined the effects of suspicion on the recall of cues used to
make veracity judgments.  It was hypothesized that more noncontent cues
would be attended to by suspicious individuals than nonsuspicious individualds.
Copyright Western States Communication Association Winter 1998

Full Text:

The present study investigated the effects of suspicion on the recall of
cues used to make veracity judgments. It was hypothesized that more noncontent
cues would be attended to by suspicious individuals than nonsuspicious
individuals. One hundred thirteen participants viewed videotapes of persons
describing their work history truthfully or deceitfully under high and
low suspiciousness conditions. Following the videotapes, the participants
were asked to indicate whether the applicant had held the job and produced
a written explanation for their judgments. When these statements were content
analyzed for types of deception cues, the results supported the hypothesis.


MA substantial amount of research has attempted to identify the types of
behavioral and verbal cues that are produced during deception. For example,
deTurck and Miller (1990) identified six behavioral cues unique to deception-induced
arousal: message duration, pauses, nonfluencies, response latencies, adaptors,
and hand gestures. Additionally, deceivers tend to blink more (Riggio &
Friedman, 1983), speak at a higher pitch (Ekman, Friesen, O'Sullivan, &
Scherer, 1980), exhibit increased pupil dilation (O'Hair, Cody, & McLaughlin,
1981) and make more negative statements (Mehrabian, 1967). In two reviews
of this literature, Zuckerman and his colleagues have identified behavioral
and verbal cues that are reliably associated with deception (Zuckerman,
DePaulo, & Rosenthal, 1981; Zuckerman & Driver, 1985).

Attention has also been given to the behavioral and verbal cues that are
actually utilized by people to detect deception (e.g., Bond, Omar, Pitre,
Lashley, Skaggs & Kirk, 1992; deTurck, Harszlak, Bodhorn, & Texter, 1990;
Stiff & Miller, 1986). There is reason to assume that the cues used by
people to make veracity judgments are not the actual diagnostic cues associated
with deception. That is, if detectors use diagnostic cues to detect deception,
then we would expect them to be relatively accurate. However, studies that
have examined deception detection accuracy have rarely found mean detection
accuracies significantly above what would be expected from random guessing
(DePaulo, Stone, & Lassiter, 1985; Kraut, 1978; Kraut & Poe, 1980; Zuckerman
et al.,1981; Zuckerman & Driver,1985). Fiedler and Walka (1993) have proposed
that most detectors rarely receive prompt feedback on the objective truth,
which prevents any efficient learning. Further, when people are trained
to use diagnostic cues, their detection accuracy improves. For example,
when deTurck et al. (1990) trained detectors to use reliable deception
cues, a significant increase in detection accuracy was reported.

The studies that have examined cues utilized in veracity judgments indicated
a reliance on the verbal content of the message (e.g., Fielder & Walka,
1993). Fiedler and Walka (1993) suggested that naive observers rely on
heuristics (c.f. Kraut, 1978; McCornack & Parks, 1986; Stiff & Miller,
1986). Two commonly used cues are uniqueness of the event reported and
falsifiability of the message. That is, if the event reported is unexpected
or contains a large number of factual (vs. emotional) statements, the communicator
is more likely to be perceived as lying. Further, McCornack and his colleagues
found that other elements of the message (e.g., message clarity, relevance,
quality, and quantity) influenced the perceived deceptiveness of the message
(McCornack & Parks, 1986; McCornack & Levine, 1990; McCornack, Levine,
Solowczuk, Torres, & Campbell, 1992). Specifically, McCornack et al. (1992)
demonstrated that manipulations of quality (e.g., not providing highly
sensitive information) were perceived as the "most deceptive" messages.


However, it is possible that elements of the deception situation may cause
people to attend to other types of cues (Ebesu & Miller, 1994; Millar &
Millar, 1995; Stiff, Miller, Sleight, Mongeau, Garlick, & Rogan, 1989).
One variable that may cause people to shift to other types of cues is situationally
aroused suspicion. In low-suspicion situations, relying on verbal content
cues may reflect the strong tendency for people to limit the amount of
their cognitive effort when making judgments about others. That is, there
is considerable evidence that people tend to be cognitive misers when making
inferences about others (e.g., Gilbert, 1989; Gilbert, McNulty, Giuliano,
& Benson, 1992; Langer, Blank, & Chanowitz, 1978; Quattrone, 1982). Consequently,
people in low-suspicion situations may attend to the most available and
easily retrieved information to make veracity judgments. Although availability
of information does not necessarily imply that the person will utilize
the information to make veracity judgments, there is considerable evidence
suggesting that availability is related to utilization. For example, Fazio
and his colleagues have developed an extensive line of research that demonstrated
that attitudes available in memory are utilized to direct behavior (Fazio,
1989; Fazio, Chen, McDonel, & Chairman, 1982). In most spoken interactions,
the goal is to understand and respond to the content of the message. Therefore,
the content of the message will be more available and may be utilized to
make veracity judgments (e.g., does the information make sense, is there
sufficient detail).

Alternatively, in moderately high suspicion situations, persons may be
motivated to expend more cognitive effort. That is, a number of researchers
have proposed that moderate levels of suspicion operate to increase the
amount of cognitive effort used by people to make veracity judgments (Burgoon,
Buller, Ebesu, & Rockwell, 1994; Stiff, Kim, & Ramesh, 1992; McCornack
& Levine, 1990). Consistent with this proposal, Hilton and his colleagues
have repeatedly found that suspicious observers were less likely to rely
on heuristics when making judgments about other people and were more likely
to engage in more deliberative atttributional thinking and the examination
of subtle cues (Fein, Hilton, & Miller, 1990; Hilton, Fein, & Miller, 1993).
Consequently, people in suspicious situations may be more likely to attend
to some less-available aspects of the interaction. That is, suspicious
people may also attend to noncontent cues (e.g., gaze aversion). In the
present study, it was hypothesized that

Hi: While making veracity judgments, people who are suspicious will use
more noncontent

cues than will people who are not suspicious.

H2: While making veracity judgments, people who are suspicious will use
less content cues than will people who are not suspicious.

METHOD

Participants

The sample consisted of 113 participants (64 female and 49 male) who were
recruited to participate in the study. Twenty-seven participants were recruited
from a large urban community and 86 from the undergraduate population at
a large southwestern university. Participants were recruited on a voluntary
basis with class credit offered for undergraduates. The ages of the participants
ranged from 17 years to 58 years with the average age of the sample being
26 years. The subjects participated in the experimental conditions in groups
of three to four people and the sessions were conducted by male and female
experimenters. Subjects were randomly assigned to each of the two experimental
conditions.

Materials

Videotapes were constructed using 24 (14 female and 10 male) undergraduates,
recruited from a large southwestern university, who participated on a voluntary
basis. At the beginning of each session, participants were required to
list in order their five most recent jobs. To enhance motivation to lie
successfully, the communicators were advised that the researchers were
interested in deceptive ability because the ability to lie successfully
is related to a number of important skills and traits such as intelligence
(see DePaulo, Kirkendol, Tang, & O'Brien, 1988). Then the communicators
were informed that they would be videotaped twice: once while they were
being truthful and once while they were being deceitful. In the truth session,
the communicators were asked to state truthfully what their most recent
job was, and to respond to a series of questions about this job. For example,
applicants were asked: what was the best part of your last job?; what did
you think of your boss?; describe your coworkers (see Bond, Kahler, & Paolicelli
(1985) for a similar procedure). In the deceitful session, the communicators
were instructed to lie about what their most recent job was (communicators
were restricted to jobs that were plausible) and invent answers to the
questions about their jobs. None of the participants, when asked to lie,
described jobs held (according to their earlier statements) in the last
five years. The order of the truthful and deceitful video sessions was
randomized. After completing the videotaping sessions, the communicators
were debriefed and released.

Procedure

Participants were recruited to a study investigating job interview impression
management. They were informed that they would be viewing videotapes made
during a job interview of a number of people who are describing their most
recent jobs. Participants were asked to watch each videotape very closely
because they would be required to make a number of judgments about the
interviewee's performance. Each participant viewed a videotape that consisted
of eight different communicators describing their most recent jobs. On
each tape presented to the participants, one-half of the descriptions were
truthful and one-half were deceitful. The order of the truthful and deceitful
communications was randomized on each tape.

Suspicion manipulation. Before beginning the videotape, the experimenter
provided the participants with some background information about the eight
interviews being presented. Then directions were given to manipulate suspicion.
Participants who were randomly assigned to the high suspicion condition
were told that some of the people being interviewed may not be completely
truthful (see Burgoon et al., 1994 for similar manipulation). Participants
assigned to the low suspicion condition were told nothing about the potential
veracity of the interviewee.

Measures. Immediately following the background information, participants
were seated four to five feet from a television monitor that displayed
the videotapes of the eight interviews. After each interview the tape was
stopped and the participants were asked to evaluate the performance of
the job applicant on a number of scales (see McCornack & Levine, 1990 for
discussion of the importance of using filler items). For example, the participants
were asked to indicate by checking "yes" or "no" whether the applicant
had adequate social skills. After each yes/no response the participants
were asked to indicate how confident they were about this judgment on a
scale with endpoints of 0% to 100%. Embedded in these items was one question
that asked the participants to indicate whether the applicant had held
the job he/she was describing in the interview. In addition, space was
provided for the participants to explain why they thought the applicant
had or had not held the job. These written responses provided the material
for the content analysis.

When the participants had finished judging all eight candidates on the
videotapes, the participants indicated how suspicious they would be of
a person in the same situation as presented in the videos on a scale with
endpoints of "1" meaning very suspicious and "9" meaning not suspicious
and "9" meaning not suspiciozcs.

The participants were the questioned concerning their understanding of
the experimental hypotheses, debriefed, and released.

Results

Manipulation Check. To determine whether participants in the low- and high-suspicion
conditions perceived the situation differently, responses to the suspiciousness
questions were analyzed in a one-factor (high vs. low suspicion) analysis
of variance (ANOVA). As expected, the participants in the high suspicion
condition reported being significantly more suspicious (M = 3.94) than
participants in the low suspicion condition (M = 6.42), FCl, 111) = 68.38,p
< .001, eta2 = .38.

Suspiciousness and Cue Use. Each of the statements in the participants'
open-ended descriptions of why they thought the applicant had or had not
held the job were coded as referring to a content cue or a noncontent cue.
Content cues refer to specific aspects of the verbal content of the message;
they focus on what was said by the communicator. For example, content cues
include judgments about whether the information presented by the communicator
was correct and whether the communicator able to provide details about
the job. Noncontent cues refer to all other visual and vocal deception
cues. For example, noncontent cues include judgments about whether the
communicator spoke quickly, shifted postures, or hesitated before responding
(see Stiff et al., 1989 for an examination of visual and vocal cues). Two
research assistants who were unaware of the experimental hypothesis coded
the statements independently. Inter-rater reliability was strong (r= .85).


The content cues and the noncontent cues listed by each participant were
summed across the eight interviews. These sums were analyzed in a 2 (high
vs. low suspicion) x 2 (content vs. noncontent cues) ANOVA with repeated
measures assumed on the last factor. The main effect for cue type was significant,
F(1, 111) = 175.46, p < .001, eta2 = .39. In general, participants used
more content cues than noncontent cues to make their veracity judgments.
As expected, a significant interaction between cue type and suspicion emerged,
F(1,111) = 4.65, p = .03, eta2 = .04. Simple effects analysis revealed
that, consistent with HI, more noncontent cues were used when participants
were high in suspicion (M = 2.89) than when they were low in suspicion
(M = 2.48) F111) = 7.12,p = .01, eta2 = .06. Similarly, as H2 predicts,
less content cues were used when participants were high in suspicion (M
= 6.41) than when they were low in suspicion (M = 7.15), F(1, 111) = 6.43,
p = .01, eta2 = .05.

DISCUSSION

The present research indicated that people attempting to assess veracity
recall used significantly more content cues than noncontent cues. This
finding is consistent with previous research demonstrating the preference
for global heuristics and content-related cues by lie detectors ( Fiedler
& Walka, 1993; McCornack & Parks, 1986; Stiff & Miller, 1986). The present
research also indicated that suspicion influences the number of noncontent
cues recalled. That is, participants that are suspicious recalled using
more noncontent cues than did participants that were less suspicious. We
suggest that this pattern occurred because individuals, when not suspicious,
respond to what is being said in the message. When suspicious, however,
participants exert more cognitive effort and also attend to noncontent
cues.

The results of the present study may provide a clue to the relationship
between suspicion and detection accuracy in relational partners found by
McCornack and Levine (1990). These researchers, as well as others, have
found that in relationships, partners exhibit a "truth bias" that interferes
with deception detection (Buller & Aune, 1987; McCornack & Parks, 1986).
However, McCornack and Levine (1990) also demonstrated that moderate levels
of suspicion are associated with increased accuracy. Perhaps, like the
interview setting, suspicious partners begin to attend to more diagnostic
noncontent cues in addition to the content of the message.

Although the results are consistent with our expectations the present study
does have limitations. Participants were asked to make self-reports on
cues they used to detect deception. A large body of research has indicated
that people often have difficulty describing their cognitive processes
(Nisbett & Wilson, 1977). It may be that cues the participants recalled
were not all the cues they actually used to make their veracity judgments.
That is, it is possible that some deception cues may be used unconsciously
or may simply be difficult to recall. However, the present study has demonstrated
an enhanced ability to recall noncontent cues when participants are suspicious.
The relationship between the cues that are recalled and the cues that are
actually used is an issue for future research.

REFERENCES

Bond, C., Kahler, K., & Paolicelli, L. (1985). The miscommunication of
deception: An adaptive perspective. Journal of Experimental Social Psychology,
21, 331-345. Bond, C. F.,Jr., Omar, A., Pitre, U., Lashley, B. R., Skaggs,
L. M., & Kirk, C. T. (1992). Fishy-looking liars: Deception judgment from
expectancy violation. Journal of Personality and Social Psychology, 63,
969-977.

Buller, D. B., & Aune, R. K. (1987). Nonverbal cues to deception among
intimates, friends, and

strangers. Journal of Nonverbal Behavior, 11, 269-290.

Burgoon, J. K., Buller, D. B., Ebesu, A. S., & Rockwell, P. (1994). Interpersonal
deception: V. Accuracy in deception detection. Communication , Monographs,
61, 303-325. deTurck, M. A., Harszlak, J. J., Bodhorn, D. T., & Texter,
L. A. (1990). The effects of training social perceivers to detect deception
from behavioral cues. Communication Quarterly, 38, 1-11.

deTurck, M. A. & Miller, G. R. (1990). Training observers to detect deception:
Effects of

self-monitoring and rehearsal. Human Communication Research, 16, 603-620.
DePaulo, B. M., Kirkendol, S., Tang, J., & O'Brien, T. (1988). The motivational
impairment effect in the communication of deception: Replications and extensions.
Journal of Nonverbal Behavior, 12,177-202.

DePaulo, B. M., Stone, J. I., & Lassiter, G. D. (1985). Deceiving and detecting
deceit. In B. R. Schlenker (Ed.), The self and social life (pp. 323-370).
New York: McGraw-Hill Ebesu, A. S., & Miller, M. D. (1994). Verbal and
nonverbal behaviors as a function of deception

type. Journal of Language and Social Psychology, 13, 418-442. Ekman, P.,
Friesen, W. V., O'Sullivan, M., & Scherer, K. R. (1980). Relative importance
of face, body, and speech in judgments of personality and affect. Journal
of Personality and Social Psychology, 38, 270277.

Fazio, R. H. (1989). On the power and functionality of attitudes: The role
of attitude accessibility. In A. Prantkanis, S. Bricklayer, & A. Greenwald
(Eds.), Attitude structure and function (pp. 153-179). Hillsdale, NJ: Erlbaum.


Fazio, R. H., Chen, J., McDonel, E., & Chairman, S. (1982). Attitude accessibility,
attitudebehavior consistency, and the strength of the object-evaluation
association. Journal of Experimental Social Psychology, 18, 339-353.

Fein, S., Hilton J. L., & Miller, D. T. (1990). Suspicion of ulterior motivation
and the correspondence bias. Journal of Personality and Social Psychology,
58, 753-764. Fiedler, K., & Walka, I. (1993). Training lie detectors to
use nonverbal cues instead of global

heuristics. Human Communication Research, 20, 199-223. Gilbert, D. T. (1989).
Thinking lightly about others: Automatic components of the social inference
process. In J. S. Uleman &J. A. Bargh (Eds.), llnintended thought (pp.
189-210). New York: Guilford.

Gilbert, D. T., McNulty, S. E., Giuliano, T. A., & Benson, J. E. (1992).
Blurry words and fuzzy deeds: The attribution of outcome behavior. Journal
of Personality and Social Psychology, 62,18-25.

Hilton, J. L., Fein, S., Sr Miller, D. T. (1993). Suspicion and dispositional
inference. Personality and Social Psychology Bulletin, 19, 501-512.

Kraut, R. (1978). Verbal and nonverbal cues in the perception of lying.
Journal of Personality

and Social Psychology, 36, 380391.

Kraut, R., & Poe, D. (1980). Behavioral roots of person perception: The
deception judgments of customs inspectors and laymen. Journal of Personality
and Social Psychology, 39, 784-798.

Langer, E., Blank, A., & Chanowitz, B. (1978). The mindlessness of ostensibly
thoughtful action. Journal of Personality and Social Psychology, 36, 635642.


Mehrabian, A. (1967). Orientation behaviors and nonverbal attitude communication.
Journal of

Communication, 17, 324-332.

McCornack, S. A., Levine, T. R., Solowczuk, K., Torres, H. I., & Campbell,
D. M. (1992). When the alteration of information is viewed as deception:
An empirical test of information manipulation theory. Communication Monographs,
59,17-29. McCornack, S. A., & Levine, T. (1990). When lovers become leery:
The relationship between suspicion and accuracy in detecting deception.
Communication Monographs, 57, 219230.

McCornack, S. A., & Parks, M. R. (1986). Deception detection and relationship
development: The other side of trust. In M. L. McLaughlin (Ed.), Communication
Yearbook 9 (pp. 337-389). Beverly Hills, CA: Sage.

Millar, M. G., & Millar, K. U. (1995). Detection of deception in familiar
and unfamiliar persons: Effects of information restriction. Journal of
Nonverbal Behavior, 19, 6984. Nisbett, R. E., & Wilson, TD. (1977). Telling
more than we can know: Verbal reports on mental





The effects of cognitive capacity and suspicion on truth bias
Communication Research
Beverly Hills
Oct 1997

--------------------------------------------------------------------------------

Authors:                  Murray G Millar

Authors:                  Karen U Millar

Volume:                   24

Issue:                    5

Pagination:               556-570

ISSN:                     00936502

Subject Terms:            Cognition & reasoning
                          Communication
                          Perceptions


Abstract:

Millar and Millar investigated the effects of cognitive capacity and suspicion
on veracity judgments.
Copyright Sage Publications, Inc. Oct 1997

Full Text:

This study investigated the effects of cognitive capacity and suspicion
on veracity judgments. It was hypothesized that under low suspicion conditions,
truth bias would be more pronounced when participants had low cognitive
capacity than when participants had high cognitive capacity. One hundred
and seven participants viewed presentations of people either truthfully
or deceptively describing a series of pictures. Prior to the presentations,
a short description designed to increase suspicion was read to half the
participants. Participants viewed half of the presentations while working
on arithmetic problems (low capacity) and the other half while not working
on arithmetic problems (high capacity). Following each presentation, the
participants were required to evaluate the communicator's performance on
a number of scales and indicate whether the communicator was actually describing
the picture. The results partially supported the hypothesis.

Most studies that have examined the ability of people to detect deception
have found that the detection accuracy of naive observers rarely exceeds
60% and usually is near levels that would be expected by chance (DePaulo,
Stone, & Lassiter, 1985; DePaulo, 1988; DePaulo & DePaulo, 1989; Ekman,
1985; Ekman & O'Sullivan, 1991; Kraut, 1980; Zuckerman, DePaulo, & Rosenthal,
1981; Zuckerman & Driver, 1985). A variety of reasons have been offered
for the low detection accuracies reported in these studies. For example,
low detection accuracy may have been caused by the absence of baseline
information on the deceiver's truthful behavior (e.g., Brandt, Miller,
& Hocking, 1982; O'Sullivan, Ekman, & Friesen, 1988) or the absence of
probes by the detector (e.g., Buller, Strzyzewski, & Comstock, 1991; Stiff
& Miller, 1986). However, correcting these problems has usually produced
only modest gains in detection accuracy.

Another reason people may have difficulty detecting deception is their
tendency to assume that people are telling the truth in most interpersonal
interactions (McCornack & Parks, 1986). That is, the assumption of truthfulness
or truth bias may prevent people from searching for deception cues, thus
reducing detection accuracy. A large number of studies have indicated that
truth bias is a powerful and widespread effect occurring in interactions
between familiar people (e.g., McCornack & Levine, 1990; McCornack & Parks,
1986; Millar & Millar, 1995; Stiff, Kim, & Ramesh, 1992) and nonfamiliar
people (e.g., Buller, Strzyzewski, & Hunsaker, 1991; DePaulo et al., 1985;
Kalbfleisch, 1992; Riggio, Tucker, Throckmorton, 1987).

Two explanations have been offered for the widespread presence of truth
bias. First, a functional explanation suggested that truth bias facilitates
communication and maintains relationships. For example, Kraut and Higgins
(1984) proposed that the assumption of truthfulness in a conversational
partner is a fundamental part of most conversations, allowing participants
to go beyond what literally is said (cf. Clark & Clark, 1977; Grice, 1975).
Similarly, McCornack and Parks (1986) suggested that trust or a truth bias
is an integral part of maintaining intimacy in close relationships.

A second explanation proposes that truth bias may be the result of a cognitive
heuristic or simple decision rule (Buller et al., 1991; Stiff et al., 1992).
That is, when evaluating the veracity of communications, people use the
simple decision rule or heuristic that most messages are truthful.l The
proposal that people use this simple decision rule is consistent with a
vast amount of evidence indicating that social perceivers employ a large
variety of both general and idiosyncratic heuristics (Tversky & Kahneman,
1974, 1983). There is also evidence that people employ a number of heuristics
(e.g., infrequency, falsifiability) when they are motivated to attempt
to detect deception (Fiedler & Walka, 1993).

Heuristic processing of information allows for the evaluation of complex
stimuli with a limited amount of cognitive effort (Tversky & Kahneman,
1974, 1983). The heuristic proposed by Stiff et al. (1992) would limit
the amount of cognitive effort used in making veracity judgments. That
is, the simple decision rule that others are generally telling the truth
(truth bias) requires far less cognitive effort than scrutinizing each
message for deception cues. Limiting effort is important when making veracity
judgments because people do not have enough cognitive capacity to make
veracity judgments based on a careful scrutiny of each communication. Furthermore,
there is considerable evidence that people tend to be cognitive misers
when making inferences about other people (e.g., Gilbert,1989; Gilbert,
McNulty, Giuliano, & Benson, 1992; Langer, Blank, & Chanowitz, 1978; Quattrone,
1982).

Although interpreting truth bias as resulting from a cognitive heuristic
is consistent with current social cognitive approaches to person perception,
there have been few attempts to provide evidence for this conceptualization.
The present research endeavored to provide support for a heuristic explanation
of truth bias by manipulating the amount of cognitive capacity available
to people making veracity judgments. If a truth bias heuristic operates
to reduce cognitive demands, then the amount of cognitive capacity available
to make an inference should be related to the use of the heuristic. That
is, as people have less cognitive capacity, they should be more motivated
to reduce processing demands through the use of heuristics (Chaiken, Liberman,
& Eagly, 1989). Consequently, when people have less cognitive capacity
available to make a veracity judgment, truth bias should become more pronounced.


In addition to manipulating cognitive capacity, suspicion was manipulated
in the current research. Suspicion was manipulated in an effort to link
possible changes in veracity judgments to the use of the truth bias heuristic.
Research on veracity judgments has repeatedly demonstrated that increases
in suspicion lead to more perceptions of deception (e.g., McCornack & Levine,
1990; Stiff et al., 1992; Toris & DePaulo, 1985). It seems as if suspicion
motivates people to initiate more mindful processing (Buller, Strzyzewski,
& Comstock, 1991; Burgoon, Buller, Ebesu, & Rockwell, 1994; Hilton, Fein,
& Miller, 1993). Stiff et al. (1992) have concluded that suspiciousness
was enough to offset the use of the truth bias heuristic even in well-developed
relationships. If suspicion motivates more mindful processing, then suspicion
may limit heuristic processing when people have sufficient cognitive capacity.


It was expected that when participants are not suspicious, reductions in
cognitive capacity should cause more use of the truth bias heuristic. That
is, as capacity is reduced, more heuristic processing would be employed
to reduce cognitive demands. Consequently, participants low in suspicion
should make more truth judgments when they have low cognitive capacity
than when they have high cognitive capacity. Alternatively, when participants
are suspicious, this effect should reverse. Suspicious participants who
have sufficient cognitive capacity should use less of the truth bias heuristic.
That is, suspicion initiates more mindful processing and less reliance
on heuristics. However, when suspicious participants have less cognitive
capacity, they may again need to rely on heuristic processing and adopt
a lie bias (Levine & McCornack, 1991). Consequently, participants high
in suspicion should make more truth judgments when they have high cognitive
capacity than when they have low cognitive capacity. Overall, suspicion
should be associated with fewer truth judgments as well as interact with
cognitive capacity.

In the present study, participants were required to view four communicators
who were describing pictures truthfully and deceitfully Prior to the presentations,
half of the participants read a short description of the situation that
was designed to increase suspicion about the communicators. The other half
of the participants did not read this description. Then the participants
viewed two presentations while being distracted (low capacity) and two
of the presentations with no distraction (high capacity). After viewing
each presentation, the participants were required to evaluate the communicator's
performance on a number of scales. Embedded in these items was one question
that asked the participants to indicate whether the communicator was actually
describing what was in the picture, that is, the participants were asked
to judge the veracity of the communicator.

Method

Participants

One hundred and seven (62 female and 45 male) participants recruited both
from a large urban community and from undergraduates at a large southwestern
university participated in the study. Participants were recruited using
sign-up sheets placed on campus bulletin boards. Participants received
no monetary compensation, but class credit was offered to undergraduates.
The ages of the participants ranged from 17 years to 80 years of age, with
an average age in the sample of 33 years. The subjects participated individually
in experimental sessions that were conducted by male and female experimenters
in standard classrooms. Participants were randomly assigned to each of
the two between-subjects experimental conditions: low suspicion or high
suspicion.

Materials

Three videotapes were constructed in which four communicators described
two pictures in a truthful manner and two pictures in a deceitful manner.
The order of the truthful and deceitful communications was randomized in
each tape. The communicators were 22 (12 females and 10 males) undergraduates
who were recruited from a large southwestern university. At the beginning
of the session, to enhance motivation to lie successfully, the communicators
were advised that the researchers were interested in deceptive ability
because the ability to lie successfully is related to a number of important
skills and traits, such as intelligence (see DePaulo, Kirkendol, Tang,
& O'Brien, 1988). Then the communicators were presented with a series of
pictures depicting pleasant scenes (e.g., a mother and son playing on a
beach) and unpleasant scenes (e.g., a train accident). For half the trials,
the communicater was instructed to be truthful, and for the other half
of the trials, the communicator was instructed to be deceitful. The order
of the different types of trials was randomized. If the communicator was
instructed to be truthful, he or she was required to describe (a) what
kind of overall feeling the picture created, (b) what was happening in
the picture, and (c) the people or objects in the picture. If the communicator
was instructed to lie, he or she was required to describe (a) the opposite
feeling to what the picture created, (b) something that was not happening
in the picture, and (c) people or objects that were not in the picture
(see Ekman & O'Sullivan, 1991, for similar procedure).

Procedure

Participants were recruited to a study that was ostensibly investigating
how situational factors influence the evaluation of public presentations.
Participants were informed that they would be required to view people making
presentations describing pictures. The participants were also informed
that they would have to view the presentation while experiencing a variety
of different situational factors.

Suspicion manipulation. Before beginning the presentations, the experimenter
provided the participants with some background information about the four
communications. Participants who were randomly assigned to the high suspicion
condition were told that some of the people being interviewed might be
presenting false information (see Levine & McCornack, 1991; McCornack &
Levine, 1990; Stiff et al., 1992; for similar manipulations). Participants
assigned to the low suspicion condition were told nothing about the potential
veracity of the communicator.

Cognitive capacity. Immediately following the background information, participants
began viewing the four presentations. On two of the four trials, an attempt
was made to limit the amount of cognitive capacity available to the detectors
while viewing the presentation. On these trials, the detector was required
to solve simple arithmetic problems while viewing the communicator's description
of the picture. Detectors were asked to solve as many problems as they
could and were told that the math problems would be used in a later portion
of the study. Problems were presented to the detector for the entire length
of the communication. On the other two of the four trials, the detectors
were not required to solve arithmetic problems (see Festinger & Maccoby,
1964; Osterhouse & Brock, 1970; Petty, Wells, & Brock, 1976; Przybyla &
Byrne, 1984; for similar procedures). All six possible orders of the two
high capacity and two low capacity trials were used.

Measures. After each presentation, the detector was asked to evaluate the
presenter's performance on a number of scales. For example, the detectors
were asked to indicate by checking yes or no whether the communicator had
spoken clearly or whether the communicator liked what he or she had been
describing. After each yes or no response, the detectors were asked to
indicate how confident they were about this judgment on a scale with endpoints
of 0% to 100%. Embedded in these items was a question that asked the detectors
to indicate whether the communicator had described what was actually in
the picture, that is, the participants were asked indirectly to indicate
whether they believed the person was lying.

When the detectors had finished answering these questions, they were asked
to write down as much of the communicator's description of the picture
as they could recall. Also, participants were required to complete a thoughtlisting
procedure to get an index of what they were thinking about while watching
the videotape. In this procedure, detectors were provided with a form containing
eight boxes and asked to write down the first thought they had while listening
to the presentation in the first box, the second thought in the second
box, and so on. After completing the listing of their thoughts, the detectors
were asked to place a + beside the box if the thought was favorable toward
the presentation, a- if the thought was negative toward the presentation,
or a 0 if the thought was irrelevant or neutral toward the presentation
(see Greenwald, 1968; Wu & Shaffer, 1987; for descriptions of this procedure).


When the detectors had responded to each of the four presentations, they
were asked to indicate their age and gender. Also, they were asked to indicate
how suspicious they were of the communicators on a 9-point scale with endpoints
of 1 = not suspicious and 9 = very suspicious. Finally, the participants
were questioned concerning their understanding of the experimental hypotheses
and debriefed.

Results

Manipulation check. To assess the effectiveness of the suspicion manipulation,
the participants were asked to indicate how suspicious they were of the
communicators on a 9-point scale. When scores from this scale were analyzed
in a one-factor (high vs. low suspicion) analysis of variance (ANOVA),
participants in the high suspicion conditions indicated more suspicion
(M = 6.07, SD = 1.93) than participants in the low suspicion conditions,
M = 1.96, SD = 1.18, F(1, 105) = 174.81, p < .001, eta^sup2^ = .62. To
assess the effectiveness of the cognitive capacity manipulation, the number
of problems solved was summed for the low capacity trials. In the low capacity
conditions, the participants performed on average 8.52 problems. When the
number of problems solved in the low capacity conditions was analyzed in
a one factor (high vs. low suspicion) ANOVA, the level of suspicion did
not produce any significant differences.

Veracity judgments. A measure indicating the number of judgments of deception
was constructed by summing the number of times each participant indicated
that a communicator had not described what was in the picture, that is,
the number of times he or she indicated that the communicator was being
deceptive. Separate scores were created for the truthful communications
with low and high capacity and for the deceitful communications with low
and high capacity. These scores were analyzed in a 2 (high vs. low suspicion)
x 2 (high vs. low capacity trial) x 2 (truthful vs. deceitful communication)
ANOVA with repeated measures assumed on the last two factors. As expected,
a significant Suspicion x Capacity interaction was obtained, F(1, 105)
= 5.36, p = .02, eta^sup2^ = .05.2 When participants were less suspicious,
they made fewer judgments of deception when they had low capacity than
when they had high capacity, F (1, 52) = 5.14, p = .03, eta^sup2^ = .09.
By contrast when participants were more suspicious, there was no significant
difference between the two capacity groups, F = 1.00, (see Table 1). In
addition to an interaction, a strong main effect for suspicion was found,
F(1, 105) = 11.53, p = .001, eta^sup2 ^= .09. Participants made more judgments
of deception in the high suspicion conditions than in the low suspicion
conditions.

Detection accuracy. For each participant, the number of times he or she
correctly judged whether the applicant was being truthful or deceitful
was analyzed in the standard ANOVA. As expected, a significant main effect
for type of communication (truthful vs. deceptive) was obtained, F(1,105)
= 328.43, p < .0001, eta^sup2^= .76. Truthful statements were more accurately
detected (M = 1.70, SD = .55) than deceptive statements (M = .26, SD =
..52). In addition, a significant interaction between type of communication
and level of suspicion was obtained, F(1, 105) = 9.07, p = .002, eta^sup2^
= .08. Participants were more accurate with deceptive communications when
they were high in suspicion (M = .37, SD = .62) than when they were low
in suspicion (M = .16, SD = .37), F(1, 105) = 4.43, p = 04, eta^sup2^ =
..05. Alternatively, participants were more accurate with truthful communications
when they were low in suspicion (M = 1.85, SD = .41) than when they were
high in suspicion (M = 1.55, SD = .63), F(1, 105) = 8.33,p = .005, eta^sup2^
= .07.

Confidence. For each participant, his or her confidence ratings of the
deceptive judgments were also analyzed in the standard ANOVA, and the only
effect to reach significance was a main effect for capacity, F(1, 105)
= 13.23, p < 001, eta^sup2^ = .11. When participants had low capacity,
they were less confident of their judgments (M = 74.44, SD = 20.36) than
when they had high capacity (M = 80.37, SD = 21.45). Also, the relationship
between confidence in the judgment and accuracy of the judgment was examined.
When the participants' overall confidence ratings were correlated to their
overall accuracy ratings, no significant relationship was found (r = -.13,
p = .16). When separate correlations between accuracy and confidence were
calculated in each of the experimental conditions, the correlation to reach
significance was in the high cognitive capacity and low suspicion condition
(r = -.34, p = .01).

Recall accuracy. The statements produced by the participants during the
message recall procedure were scored for accuracy by two raters. Overall,
there was a high degree of consistency between the raters (r = .91), and
discrepancies were solved by discussion. An accuracy index was created
by subtracting the number of incorrect statements (statements that did
not reflect statements in the presentation) from the number of correct
statements (statements that did reflect statements in the presentation).
When this index was analyzed in the standard ANOVA, a main effect for capacity
was obtained, F(1, 105) = 91.85, p < .001, eta^sup2^ = .46. As we expected,
participants were better able to recall the messages when they had high
capacity (M = 6.71, SD = 2.37) than when they had low capacity (M = 4.52,
SD = 1.73). Also, a significant interaction between suspicion and capacity
was obtained, F(1, 105) = 8.22, p = .005, eta^sup2^ = .07. Participants
who had high capacity recalled the message more accurately when they were
high in suspicion than when they were low in suspicion, F(1, 105) = 6.10,
p = .01, eta^sup2^ = .05. When participants had low capacity, this effect
disappeared, F < 1, (see Table 2).

Cognitive responses. The participants' cognitive reactions to the presentations
were assessed by requiring them to write down their responses to the presentation
and rate these responses as positive, negative, or irrelevant. For each
participant, the number of positive, negative, and irrelevant responses
was analyzed in three separate standard ANOVAs. When the number of positive
responses was analyzed, only a main effect for capacity was significant,
F(1,100) = 32.71,p < .001, eta^sup26 = .24.3 High capacity was associated
with more positive responses (M = 3.62, SD = 2.34) than low capacity (M
= 2.22, SD = 2.03). When negative responses were analyzed, a main effect
for capacity was again obtained, F(1, 100) = 10.39, p = .002, eta^sup2^
= .09. High capacity was associated with fewer negative responses (M =
2.25, SD = 1.94) than low capacity (M = 3.12, SD = 2.17). In addition,
this analysis also produced a main effect for suspicion, F(1, 100) = 4.75,
p = .03, = .04, witheta^sup2^ high suspicion producing more negative responses
(M = 2.99, SD = 1.59) than low suspicion (M = 2.35, SD = 1.49). Finally,
when responses that were irrelevant to the message were analyzed, no significant
effects were obtained.

Discussion

[IMAGE TABLE] Captioned as: Table 1

[IMAGE TABLE] Captioned as: Table 2

The results provided partial support for the hypothesis. When detectors
were in the low suspicion conditions, they made more judgments of truth
with low capacity than with high capacity, that is, there was more truth
bias when detectors had reduced cognitive capacity However, when detectors
were in the high suspicion conditions, there was no significant difference
between high and low capacity conditions. The pattern of results from the
recall data provides support for our hypothesis that suspicion motivates
people who have sufficient cognitive capacity to engage in mindful processing
of the presentation. That is, participants who were made suspicious and
who had high cognitive capacity recalled the message better than participants
in the other conditions. Overall, these findings suggest that truth bias,
at least in part, functions as a heuristic designed to decrease the amount
of cognitive effort needed to make judgments about others' veracity.

Although the results from the veracity and recall measures were consistent
with the hypothesis, the failure to find a significant effect for cognitive
capacity in the high suspicion conditions is puzzling. If the suspicion
manipulation predisposed people to adopt a lie bias (Levine & McCornack,
1991), then it might be expected that reductions in cognitive capacity
would cause more reliance on a lie bias. That is, lie bias might operate
like truth bias to reduce cognitive demands. Perhaps, in the present study,
the amount of suspicion created was not high enough to induce a lie bias.
Recall that participants in the high suspicion conditions on average did
not rate themselves as being extremely suspicious (M = 6.07). It may be
that higher levels of suspicion would have produced a stronger tendency
toward the lie bias, and people would have been more inclined to adopt
the lie bias to deal with reduced cognitive capacity. Of course, under
very high levels of suspicion, people may simply adopt lie bias regardless
of cognitive load. Future research should examine the effects of higher
suspicion levels.

When the accuracy of the veracity judgments was examined, an expected main
effect for the veracity of the communication was found. Overall, as we
would expect if people have a truth bias, participants were more accurate
in detecting truthful messages than deceptive messages. Consistent with
past research, increases in suspicion were not associated with an overall
increase in detection accuracy (e.g., Buller, Strzyzewski, & Comstock,
1991; McCornack & Levine, 1990; Stiff et al., 1992; Toris & DePaulo, 1985).
Instead, suspicious participants were more accurate with deceptive communications,
and nonsuspicious participants were more accurate with truthful communications.


The examination of the detectors' confidence in their judgments indicated
that overall, the participants' confidence in their veracity judgments
was not related to the accuracy of their judgments. This is consistent
with some findings in the literature (e.g., Vrij, 1994). However, when
this relationship was examined separately in each condition, a negative
relation between accuracy and confidence emerged in the low suspicion and
high capacity condition. As participants were more confident in their veracity
judgments, their judgments became less accurate. The tendency to overestimate
detection ability has also been found in a number of studies (e.g., Brandt,
Hocking, & Miller, 1980; deTurck, Harszlak, Bodhorn, & Texter, 1990).

Finally, the analyses of the detectors' cognitive responses produced a
predictable pattern of results. Detectors in the low capacity conditions
produced fewer positive responses and more negative responses than participants
in the high capacity conditions. Performing arithmetic problems while attempting
to listen to a presentation is a less pleasant and satisfying experience
than just listening to the presentation. Also, detectors who were more
suspicious found the experience more negative than detectors who were less
suspicious.

Despite providing some support for the hypothesis, the present study has
several limitations. First, the effect of the problem-solving manipulation
on the detectors' cognitive capacity was not directly measured. Cognitive
capacity was not directly measured because it is difficult to construct
a measure that would not influence cognitive capacity or interact with
the experimental procedures. Although cognitive capacity was not directly
assessed, the results present a pattern of evidence consistent with the
notion that the problemsolving procedure manipulated the amount of cognitive
capacity. Not only did the detectors do more arithmetic problems in low
capacity conditions but they also were less confident in their judgments
and were less able to recall the presentation. Together, these measures
provide a reasonably good indication that cognitive capacity was manipulated
as intended.

Second, the lie detection task used in the study was noninteractive, that
is, detectors and presenters were unable to converse with one another.
In many real life communication situations, there is an interaction between
the sender and detector, where each person can influence the behavior of
the other (see Buller & Burgoon's, in press, interpersonal deception theory
for a discussion of the importance of interactive paradigms). A noninteractive
procedure was used in the present research because in an interactive paradigm
it would have been difficult to control the level of suspicion and cognitive
capacity; that is, the type and number of interactions between detector
and communicator would have influenced both suspicion and cognitive load.


Finally, the development of the truth bias heuristic is not addressed by
the present research. Perhaps the evolution of the truth bias heuristic
is related to other general heuristics such as availability. It may be
that originally, people encode most communications as truthful because
of difficulties involved in processing false information (Gilbert, 1993;
Gilbert, Krull, dc Malone, 1990) and the cognitive and practical difficulties
involved in confirming deceptions (Wegner, Coulton, & Wenzlaff, 1985).
That is, in most social situations, people do not receive feedback regarding
the truthfulness of a communication, making it difficult to confirm a deception.
If these many instances of truth judgements are easy to bring to mind,
then this may lead people to assume that the likelihood of truthful communication
is very high (cf. Stiff et al., 1992). This is a question for future research.


Notes

1. It is possible that people simply assume communications are truthful
without any processing (heuristic or systematic) of the communication or
situation. However, we do not believe that the truth bias effects found
in the literature could occur without any processing taking place. If people
simply applied the rule that people are truthful without any processing
of the communication or situation, then all communication should be judged
as truthful. The literature clearly indicates that people are influenced
by elements of the communication and situation (e.g., McCornack & Parks,
1986; Millar & Millar, 1995).

2. Several further analyses were conducted on all of the dependent measures
to examine the impact of a number of control variables. The order of truth
and deceptive communications, and the order of the high and low capacity
manipulation, were added separately as factors to each of the analyses.
The order variables were not involved in any significant effects. Also,
to examine whether the participants' responses changed from the beginning
to the end of the session, trials was added as a factor to each analysis.
The trials variable was not involved in any significant effects.

Finally, a set of analyses were conducted to examine whether the positive
skew associated with using count data influenced the results. All the analyses
on count data reported in this article were performed with the data subjected
to a square transformation (see Kirk, 1986, and Winer, 1971, for discussion
of transformations). These analyses produced the same pattern of significant
effects. The means and analyses reported in the text have been presented
in their original form.

3. Five of the participants failed te complete the cognitive responses
measure.

References

Brandt, D. R., Hocking, J., & Miller, G. (1980). The truth-deception attribution:
Effects of familiarity on the ability of observers to detect deception.
Human Communication Research, 6, 99-110.

Brandt, D. R., Miller, G., & Hocking, J. (1982). Familiarity and lie detection:
A replication and extension. Western Journal of Speech Communication, 46,
276-290.

Buller, D. B., & Burgoon, J. K. (in press). Interpersonal deception theory.
Communication Theory.

Buller, D. B., Strzyzewski, K. D., & Comstock, J. (1991). Interpersonal
deception: I. Deceivers' reactions to receivers' suspicions and probing.
Communication Monographs, 58, 1-24.

Buller, D. B., Strzyzewski, K. D., & Hunsaker, F. (1991). Interpersonal
deception: II. The inferiority of conversational participants as deception
detectors. Communication Monographs, 58, 25-40. Burgoon, J., Buller, D.,
Ebesu, A., & Rockwell, P. (1994). Interpersonal deception: V. Accuracy
in deception detection. Communication Monographs, 61, 303-325.

Chaiken, S., Liberman, A., & Eagly, A. (1989). Heuristic and systematic
processing within and beyond the persuasion context. In J. S. Uleman &
J. A. Bargh (Eds.), Unintended thought (pp. 212-252). New York: Guilford.


Clark, H., & Clark, E. (1977). Psychology and language. New York: Harcourt
Brace Jovanovich.

De Paulo, B. M., Kirkendol, S., Tang, J., & O'Brien, T. (1988). The motivational
impairment effect in the communication of deception: Replications and extensions.
Journal of Nonverbal Behavior, 12, 177-202. DePaulo, B. M., Stone, J. I.,
& Lassiter, G. D. (1985). Deceiving and detecting deceit. In B. R. Schlenker
(Ed.), The self and social life (pp. 323-370). New York: McGraw-Hill.

DePaulo, P J. (1988). Research on deception in marketing communications:
Its relevance to the study of nonverbal behavior. Journal of Nonverbal
Behavior, 12, 253-273.

DePaulo, P. J., & DePaulo, B. M. (1989). Can deception by salespersons
and customers be detected through nonverbal behavioral cues? Journal of
Applied Social Psychology, 19,1552-1577.

deTurck, M. A., & Harszlak, J., Bodhorn, D., & Texter, L. (1990). Deception
and arousal: Isolating the behavioral correlates of deception. Human Communication
Research, 12, 181-201.

Ekman, P (1985). Telling lies: Cues to deceit in the marketplace, marriage,


and politics. New York: W W Norton.

Ekman, P, & O'Sullivan, M. (1991). Who can catch a liar?American Psychologist,
46, 913-920.

Festinger, L., & Maccoby, N. (1964). On resistance to persuasive communica

tions. Journal of Abnormal and Social Psychology, 68, 359-366. Fiedler,
K., & Walka, I. (1993). Training lie detectors to use nonverbal cues instead
of global heuristics. Human Communication Research, 20, 199-223. Gilbert,
D. (1989). Thinking lightly about others: Automatic components of the social
inference process. In J. S. Uleman & J. A. Baught (Eds.), Unintended thought
(pp. 189-210). New York: Guilford. Gilbert, D. (1993). The assent of man:
Mental representation and control of belief. In D. Wegner & J. Pennebaker
(Eds.), Handbook of mental control (pp. 57-87). Englewood Cliffs, NJ: Prentice
Hall. Gilbert, D., Krull, D. S., & Malone, P. (1990). Unbelieving the unbelievable:
Some problems in the rejection offalse information. Journal of Personality
and Social Psychology, 59, 601-613.

Gilbert, D., McNulty, S. E., Giuliano, T., & Benson, J. (1992). Blurry
words and fuzzy deeds: The attribution of outcome behavior. Journal of
Personality and Social Psychology, 62,18-25.

Greenwald, A. G. (1968). Cognitive learning, cognitive response to persuasion,
and attitude change. In A. G. Greenwald, T. C. Brock, & T. M. Ostrom (Eds.),
Psychological foundations of attitude (pp. 148-170). New York: Academic
Press.

Grice, H. P. (1975). Logic and conversation. In P Cole & J. Morgan (Eds.),


Syntax and semantics (Vol. 3, pp. 41-58). New York: Seminar Press. Hilton,
J., Fein, S., & Miller, D. (1993). Suspicion and dispositional inference.
Personality and Social Psychology Bulletin, 19, 501-512. Kalbfleisch, P.
(1992). Deceit, distrust, and the social milieu: Application of deception
research in a troubled world. Journal of Applied Communication Research,






The behavioral correlates of sanctioned and unsanctioned deceptive communication
Journal of Nonverbal Behavior
New York
Fall 1998

--------------------------------------------------------------------------------

Authors:                  Thomas H Feeley

Authors:                  Mark A deTurck

Volume:                   22

Issue:                    3

Pagination:               189-204

ISSN:                     01915886

Subject Terms:            Behavior
                          Lying
                          Cheating
                          Honesty


Abstract:

Ninety-three students were randomly assigned to one of three veracity conditions:
(1) truthful, (2) unsanctioned-deceptive, or (3) sanctioned-deceptive.
 Participants in the truthful condition were honest when reporting strategies
used when attempting to unscramble a series of anagrams.
Copyright Human Sciences Press, Inc. Fall 1998

Full Text:

ABSTRACT: Ninety-three students were randomly assigned to one of three
veracity conditions: (1) truthful, (2) unsanctioned-deceptive, or (3) sanctioned-deceptive.
Participants in the truthful condition were honest when reporting strategies
used when attempting to unscramble a series of anagrams. Students in the
sanctioneddeceptive and unsanctioned-deceptive condition were implicated
to cheat (by looking at the answers) on the anagram task by a research
confederate. Students in the sanctioned condition were asked by an experimenter
to conceal their cheating by lying to the interviewer about their "high
score" on the anagram task whereas students in the unsanctioned condition
were not given any instructions about how to answer the interviewer's questions
regarding their anagram-solving strategies. All interviews were videotaped
and verbal and nonverbal behaviors were analyzed by four student-coders.
Results indicated that unsanctioned deceivers, when compared to sanctioned
deceivers, made fewer speech errors and speech hesitations, gazed less
at their targets, and used fewer other references.

Introduction

People tell lies for a variety of reasons. Some lie to protect themselves
from punishment or embarrassment (e.g., Stiff, Corman, Krizek, & Snider,
1994); some lie to gain a monetary or social reward under false pretense
(e.g., Feeley & deTurck, 1995; Stiff & Miller, 1986); some lie to project
a false image of themselves; others lie to protect another's feelings (e.g.,
Bell & DePaulo, 1996; DePaulo & Bell, 1996); and still others lie just
for the sake of lying; some find it exciting and challenging. Ekman (1991)
calls this duping delight.

Scholars of deception have long examined the behaviors associated with
deceptive communication (for meta-analytic reviews see DePaulo, Stone,
& Lassiter, 1985; Kraut, 1980; Zuckerman, DePaulo, & Rosenthal, 1981; Zuckerman
& Driver, 1985). Results from over fifty studies have yielded some promising
results. When compared to truth tellers, deceivers tend to commit more
speech errors and speech hesitations, speak for a shorter length of time,
use more adaptors, and convey less immediacy. Authors of experiments analyzing
the behavioral correlates of deception have used many different scenarios
to study deceptive communication. For example, an early study by DePaulo
and Rosenthal (1979) is not atypical of many of the paradigms employed
to solicit deception. Senders in the DePaulo and Rosenthal study were instructed
to honestly and dishonestly describe people they like and people they dislike.
A later experiment by deTurck and Miller (1985) used an entirely different
paradigm. Participants in the deTurck and Miller study were induced by
a research confederate to cheat (or not to cheat in the truthful condition)
on a dot-estimation task. After the participants estimated the number of
dots on ten cards they were asked by an experimenter to describe the strategies
they used while estimating the number of dots on each card. Participants
who were implicated were forced to either 'fess-up' or lie about their
remarkable performance on the dot-estimation task. These two paradigms
are popular experimental designs used to examine deceptive behavior. It
is interesting to note that participants in each of the two paradigms may
experience completely different motivations when they tell lies. The question
then becomes: do the dynamics of deception differ across differing motivational
conditions? Will deceivers behave differently when they are motivated to
lie on their own volition or when they are motivated to lie to satisfy
the requirements of the experiment(er)? The present research will address
these questions. Miller and Stiff (1993) considered lies of the former
type to be unsanctioned lies whereby senders decided on their own whether
to lie or tell the truth. Lies sanctioned by the experimenter for purposes
of research are labeled sanctioned lies. The following discussion will
distinguish motivational features of each type of deception.

Unsanctioned Deception

In deciding whether or not to deceive or tell the truth, communicators
must balance the consequences of detection with the perceived benefits
of telling the truth (Feeley, 1997a; Stiff et al., 1994; Stiff & Miller,
1986). When deception is sanctioned by the experimenter, the sender does
not make the decision to lie or tell the truth-the decision is made for
the sender. Unsanctioned lies seek to capture the necessary elements that
characterize heightened arousal and increased cognitive effort: fear of
punishment, accountability, and motivation to deceive effectively. For
example, a study by Stiff et al. (1994) offered individuals an opportunity
to cheat on a class quiz and many students chose the option to cheat. After
the quiz, students were interviewed and asked specific questions about
their honesty on the quiz. Participants who chose to lie about the cheating
did so on their own accord, were motivated to conceal their illicit behavior,
and were solely responsible for their communicative behavior (Stiff et
al., 1994). Similar paradigms have been used to increase accountability
and deceiver motivation (deTurck & Miller, 1985; Dulaney, 1982; Exline,
Thibaut, Hickey, & Gumpert, 1971; Feeley, deTurck, & Young, 1995; Stiff
& Miller,1986); however, most studies in deception have employed the sanctioned
lie manipulation.

Sanctioned Deception

When one thinks of lying, the types of deception that spring to mind are
usually stories of sordid defendants on trial for murder or a shifty husband
facing accusations of infidelity. While these types of deception are certainly
exciting, arousing, and interesting, they seem to be the exception rather
than the norm in everyday social interaction. Most lies require little
planning, are relatively uninvolving and have minimal consequences if uncovered
(DePaulo, Kashy, Kirkendol, Wyer, & Epstein, 1996; Kashy & DePaulo, 1996).
Recent research by DePaulo et al. (1996) found participants to have mild
reactions to recent lies they have told. Using a diary method, participants
reported that their lies were not very serious, they did not plan them
much and did not worry much about getting caught. If most lies require
little planning, require minimal effort, and provoke little fear in the
sender of being caught, then perhaps laboratory studies using sanctioned
lies create more experimental realism than some scholars would contend
(e.g., Feeley et al., 1995).

Proponents of the sanctioned lie hedge their position on the ecological
validity of the lies used during research experiments. That is, the lies
told in the laboratory mirror most lies told in everyday communication:
exaggerated expressions of liking and false reports of attitudes and likes
about an issue or another person (cf. Burgoon & Buller, 1994; DePaulo &
Rosenthal, 1979; Levine & McCornack, 1992).

The critical difference between sanctioned and unsanctioned deception manipulations
rests in each paradigm's ability to motivate individual's to lie successfully.
Lies told in each of the two paradigms, it is argued here, involve different
motivations to lie. For example, participants in most unsanctioned paradigms
are motivated to lie to gain a reward (e.g., to win a contest or to acquire
a higher quiz grade), or to escape punishment (e.g., get caught cheating
by an experimenter), whereas participants in most sanctioned lie paradigms
are motivated to escape detection for entirely different reasons. For example,
participants in a study by DePaulo, Lanier, and Davis (1983) were told
that the ability to lie successfully was extremely important and often
associated with professional success. Several studies (e.g., Buller & Aune,
1987; Burgoon & Buller, 1994) failed to offer any motivation to senders'
to deceive other than for research purposes. Thus, it appears the motivation
to deceive would be higher (or at least different) for unsanctioned liars
than for their sanctioned counterparts. This contention, however, has yet
to be supported by any empirical evidence to date. A goal of the present
study to examine the motivational differences between the two deception
conditions. More specifically, the present research examines the behavioral
correlates of sanctioned and unsanctioned lies. Unsanctioned liars may
exhibit a different behavioral profile than sanctioned liars.

It should be noted that Zuckerman and Driver's (1985) meta-analytic review
examined behavioral differences between highly motivated and less highly
motivated deceivers. They considered senders highly motivated if promised
either a monetary reward for doing well (e.g, deTurck & Miller, 1985) on
the deception task or told that the ability to lie successfully was associated
with some type of skill (e.g., DePaulo et al., 1983)-all other studies
were considered low motivations to deceive. Their results found that when
compared to less highly motivated senders, highly motivated senders exhibit
less eye gaze, fewer blinks, fewer head movements, fewer postural shifts,
a shorter response length, a slower speech rate, greater pitch, and more
negative statements. However, the interpretation of these results should
be only speculative. These results were collapsed across several different
motivational paradigms and almost all of the manipulations employed sanctioned
lies.

The Current Study

The current experiment is an attempt to examine the behavioral differences
between truth tellers, unsanctioned deceivers, and sanctioned deceivers.
Earlier it was suggested that the different motivations for communicating
may significantly influence the behavioral profile of the communicator.
Sanctioned liars would most likely be motivated to control their self-presentation
in an effort to dupe the interviewer or to appear cooperative with the
experimental requirements. Unsanctioned liars, who are not given permission
to lie, may be motivated to lie to gain a reward (two compact disks), or
to escape punishment for cheating on an experimental task. It would be
interesting to examine if the two differing motivations elicit different
behavioral profiles. If they do, students of deception may need to re-examine
the behavioral profiles of deceptive communicators with careful consideration
given to deceiver motivation. Initial efforts toward this consideration
have been advanced earlier by Zuckerman, DePaulo, and Rosenthal (1981)
in their meta-analysis.

Method

Participants and Procedure

Two hundred twelve students from an introductory communication class at
a large eastern university participated in partial fulfillment of a course
requirement. Participants signed up for a study advertised as, "a study
in examining conversational management techniques in an interviewing situation."
Students were required to sign up for in one hour time blocks in groups
of four. It was required that students sign up with strangers.' Along with
a research confederate, the group of four students were given a prebriefing
statement that explained the purpose of the study. Participants were told
that the purpose of the study was to "examine interviewing behavior during
abstract problem-solving." The abstract problem was solving anagrams or
unscrambled words. Participants were then told they would be divided into
a group of three interviewees (EEs) and a group of two interviewers (ERs).
The goal of the EE was to unscramble as many of the anagrams as possible.
The goal of the ER was to interview and evaluate the EE's conversational
management skills during the interview. After reading the prebriefing statement,
participants were asked to sign a consent form. After randomly dividing
the groups, each group was escorted to adjacent research room for further
instructions.2

Interviewee Manipulation

Interviewees were seated and given a prebriefing form that described their
role in the experiment. EEs were reminded that the each person in the group
which solves the most anagrams would be given a reward of two compact disks
of their choice from a local music store. EEs were then required to complete
a two-minute trial task consisting of twenty anagrams. The experimenter
left the room during the two-minute trial task. After the time elapsed,
the EEs were given feedback regarding their performance on the trial task.
This was done by the experimenter who consulted a folder located on a desk
next to the table where the EEs were seated. After questions were addressed,
the five-minute, 35 anagram experimental task began. The experimenter then
left the room while the EEs began the anagram task. At this point in the
experiment the veracity manipulation was introduced. The veracity manipulation
divided EEs into three conditions: 1 ) truthful, 2) unsanctioned lie, and
3) sanctioned lie. Confederates in the truthful condition were instructed
to assist on only one or two of the answers and no more. In the unsanctioned
and sanctioned lie conditions, the confederate pretended to grow tired
after one minute of trying to unscramble the words. The confederate then
proceeded to open the experimenter's folder where the answers to the practice
task were located. After opening the folder, the confederate proceeded
to share some of the answers with the group (10-12). In instances where
the group was reticent to use the answers, the confederate was instructed
to write down the answers him/herself.

After the five minutes elapsed, the experimenter re-entered the EE research
room and explained that EEs would now be interviewed individually and asked
questions about the strategies they used to achieve their anagram score.
Students in the unsanctioned condition were immediately escorted to the
interview room. Students in the sanctioned condition were not immediately
escorted to the research room. Instead, students were debriefed about the
cheating manipulation by the experimenter. Specifically, EEs in the sanctioned
lie condition were told that "a goal of this study is to examine some of
the strategies used by individuals when telling lies." Participants were
then asked to lie to the ER about why they performed so well on the abstract
task. Any EEs who felt uncomfortable about lying to the ER were excused
from the study and offered full research credit for their participation
(as promised in the prebriefing).3

To sum up, EEs in the truthful condition were not exposed to any cheating
and subsequently were telling the truth when asked about their strategies
while unscrambling the anagrams. EEs in the unsanctioned lie condition
were exposed to cheating and must choose to lie or tell the truth regarding
their good performance on the anagram task. EEs in the sanctioned lie condition,
however, were 'caught' by the experimenter and asked to lie for the purposes
of the experiment. Interviewer Instructions

While the EEs completed the anagram task, ERs were given a prebriefing
form that described their role in the experiment. ERs were instructed to
become familiar with the eight interview questions while the EEs completed
the anagram task. ERs were also told they would be asked to assess the
EE on several dimensions after the interview was completed. ERs were told
that in many interview situations individuals are often dishonest in reporting
their abilities on certain problem-solving tasks. Furthermore, ERs were
told that fifty percent of the EEs in the experiment were given some of
the answers to the anagram task and thus would be lying or embellishing
while describing the strategies and methods they reported about solving
the anagrams.'

Interviews

After the ERs and EEs completed their separate tasks, they were reunited
in the main research room for interviewing. Interviews were conducted in
pairs and the two EEs and one ER who were not interviewing were asked to
wait in a room outside of the interviewing room. Interviews lasted, on
the average, four to five minutes and videotaped with the participants'
permission.

ERs asked EEs four baseline questions (What is your name?, What is your
major and what do you like best about your major?, Where is your hometown?,
and What do you like best about the University?) and four questions about
the strategies used to unscramble the anagrams (What strategies did you
use to unscramble the anagrams?, How do you explain your score on the anagram
task?, Did anything else happen in the research room to help your ability
to perform well on the abstract task?, How did the other group members
help on the problem-solving task?). After each interview, ERs and EEs completed
a questionnaire that asked them to assess their behavior and the other's
behavior on several dimensions.5 It is important to note that there were
eight instances (four female and four male EEs) in which EEs in the unsanctioned
lie condition did not choose to lie about why they did so well on the task.
These sixteen students (8 ERs and 8 EEs) were excused from the study. This
dropout rate is a nontrivial number; when one considers that no students
in the sanctioned lie condition and truth condition elected to discontinue
the study. This may affect the interpretation of the results (i.e., random
assignment, strong inference). It could be suggested that the students
in the unsanctioned lie condition represent only those students who elected
to lie. This differential drop-out rate may explain why the majority of
research paradigms in deception use the sanctioned paradigm.

After ERs and EEs completed their separate self-report measures, students
were debriefed thoroughly regarding the purpose of the experiment. Participants
were given a debriefing statement that explained the purpose of the study,
the goals of the study, and cited previous research in deception, should
participants be interested in reading further on the subject. After reading
the debriefing and having their concerns addressed, participants completed
a consent form that requested permission to use the selfreport data and
videotaped interview for research and educational purposes.

Coding the Videotapes

Four students served as coders and were paid for the assistance. Coders
worked independently and were asked to code the participants along fourteen
behaviors. Each coder was given one-half of the participants to code. Thus,
each individual on the videotapes was coded by two different student-coders.
Three verbal behaviors were analyzed (number of selfreferences, number
of other-references, and number of total words) along with five vocal behaviors
(number of speech errors, number of speech hesitations, number of pauses,
speech length, and response latency) and six visual behaviors (length of
adaptors, eye gaze length, number of foot gestures, number of hand gestures,
number of postural shifts, and amount of smiling). Coders were given a
training session which defined and demonstrated the fourteen cues, defined
by Zuckerman and Driver (1985). The cues were defined using definitions
offered by Zuckerman and Driver (1985). Listed below you will find the
definition for each cue. Inter-coder reliabilities using Pearson correlations
for each cue were: adaptors .61; message duration .87; eye gaze .73; foot
gestures .73; hand gestures .93; pauses .97; postural shifts .80; response
latency .66; speech errors .86; speech hesitations .88; self-references
..88; number of words .93; other references .80. Speech errors, speech hesitations,
self-references, number of words, and other references were coded from
written transcripts of the videotapes. All counts and amounts of behaviors
were divided by interview length to control for amount of time speaking.
Speech rates were computed by dividing the speech length by the number
of total words. Verbal Behaviors

1. Self-references: measured by frequency of references to self.

2. Other-references: measured by frequency of references to other (e.g.,
he, her, she, him).

3. Number of words: measured by total number of words. Vocal Behaviors


4. Speech length: total amount of time spent speaking.

5. Speech hesitations: measured by frequency of filled pauses (e.g., ers,
uhms, ahs).

6. Response latency: measured by amount of time between the end of a question
and the beginning of an answer.

7. Speech rate: the total number of words divided by the duration of the
message.

8. Pause: the total number of silent breaks in speech.

9. Speech errors: measured by frequency of nonfluencies, grammatical errors,
word and/or sentence repetition, sentence change, sentence incompletion,
or slips of the tongue.

Visual Behaviors

10. Adapters: amount of time spent using self-manipulations (e.g., rubbing
hand on leg).

11. Eye gaze: measured by duration of time spent looking directly at the
interviewer.

12. Foot gestures: measured by total amount of time spent moving feet.


13. Hand gestures: measured by total amount of time spent moving hands.


14. Postural shifts: total number of postural shifts.

Results

All analyses used 95 EEs and gender breakdown was equivalent across experimental
conditions, x^sup 2^ = 8.42, df = 12, p = .80. Ninety-two males and ninety-eight
females participated. In some instances data was not codable for subjects'
behavior and reduced the degrees of freedom in the statistical tests (listwise
deletion was used) (e.g., unclear audio, camera shot cut off hands, etc.).


Manipulation Check

To test the deception manipulation, two one-way ANOVAs were performed with
veracity serving as the between-subjects factor and reported honesty and
number of correct answers serving as the criterion variables. Results indicate
a significant difference with respect to EE reported honesty, F (2, 92)
= 119.32, p < .001, eta^sup 2^ = .72. Scheffe post-hoc analysis of the
means found significant differences between all three veridical conditions.
Participants in the truthful condition (M = 7.57) reported more honesty
(on an 8-point scale) than participants in the unsanctioned lie condition
(M = 5.81), who reported more honesty than sanctioned liars (M = 2.94).
It is interesting to note that sanctioned liars and unsanctioned liars
reported different levels of honesty on the anagram task. It is likely
that since unsanctioned liars voluntarily lied during the interview, they
decided to maintain their deception throughout the questionnaire. A significant
difference was also reported between veracity conditions on the number
of correct answers on the anagram task, F (2, 92) = 198.83, p < .001, eta^sup
2^ = .81. Participants in the unsanctioned lie condition (M = 19.23) and
sanctioned lie condition (M = 19.74) unscrambled significantly more anagrams
than did students in the truthful condition (M = 10.49). Taken together,
these results indicate that the deception manipulation was successful.


Behavioral Differences by Veracity Condition

A goal of the present study was to test for behavioral differences between
sanctioned deceivers, unsanctioned deceivers, and truthtellers. To test
for differences between each deceptive paradigm, and to test for differences
between each deception manipulation and a truthful sample of communicators,
separate one-way ANOVA tests were conducted with the verbal and nonverbal
behaviors servings as the criterion variables and veracity (unsanctioned/sanctioned/truthful)
serving as the between-subjects independent factor.

Table 1 lists the means and standard deviations by veracity condition and
F test values and effect sizes. Significant differences were found with
speech errors, F(2, 89) = 19.32, p < .001, eta^sup 2^ = .30, speech hesitations,
F(2, 89) = 3.39, p < .05, eta^sup 2^ = .07, and other references, F (2,
89) = 3.05, p < .05, eta^sup 2^ = .06.

Results of LSD post-hoc tests revealed only one significant behavioral
difference between truthful communicators and unsanctioned deceptive communicators.
Truthful senders committed more speech errors than unsanctioned deceivers.
Regarding differences between truth tellers and sanctioned deceivers, truth
tellers committed fewer speech errors and spoke for a greater length of
time than sanctioned deceivers.

To test for verbal and nonverbal difference between sanctioned and unsanctioned
deceivers and truth tellers, contrast weights were assigned to compare
means between each condition. Table 2 outlines the contrast differences
for each behavior and the significance of each difference. For the first
column, contrast weights of + 1 and -1 were assigned to truthful participants
and sanctioned lie participants respectively. In the second column, contrast
differences between truthful (+1 weight) and unsanctioned lies (-1 weight)
were analyzed and finally in the third column truthful behaviors (+ 2 weight)
were compared to both sanctioned (-1 weight) and unsanctioned (-1 weight)
behaviors.

Discussion

The results of the previous experiment indicate the type of manipulation
used to elicit deceptive communication in research makes a significant
difference. This study compared sanctioned deception, that is, lying to
satisfy the requirements of the experimenter, to unsanctioned deceptionlying
without explicit permission from the experimenter. Unsanctioned deceivers
used less adaptors, eye gaze, self references, other references and committed
fewer speech errors and speech hesitations than sanctioned deceivers. At
the same time, unsanctioned deceivers shifted more while answering questions.


These results beg the question: Why should unsanctioned and sanctioned
liars behave differently? To address this question, we offer two possible
explanations: the motivational impairment effect and communication appraisal.
It was hinted at earlier that an individual's motivation to deceive may
differ as a function of sanctioning. That is, the motivation to escape
detection may be greater with unsanctioned liars than with sanctioned liars.
In this study, unsanctioned liars were ostensibly lying for one or two
possible reasons: (1 ) to perform well on the anagram task to win the two
compact disk prizes offered by the experimenter or (2) to avoid being caught
cheating by the experimenter. Compare these motivations to the motivation
of the sanctioned liar who lied primarily to satisfy the requirements of
the study. DePaulo and colleagues (e.g., DePaulo & Kirkendol, 1989; DePaulo
et al., 1983) have found that increased motivation to lie has deleterious
effects on deception success. Stated differently, lies told by motivated
liars are more readily detected than lies told by unmotivated liars when
receivers are exposed to the sender's nonverbal behavior (i.e., audio,
audiovisual, visual). DePaulo et al. (1983) dubbed this the motivational
impairment effect.

[IMAGE TABLE] Captioned as: TABLE 1

[IMAGE TABLE] Captioned as: TABLE 2

Were Unsanctioned Liars Motivationally Impaired?

To explain the effects of motivation on deception detection success, DePaulo
et al. (1985) advanced the behavioral control explanation. The behavioral
control explanation states that highly motivated senders attempt to control
their nonverbal presentation in an effort to appear truthful. Highly motivated
senders, however, may be less successful at controlling their nonverbal
behaviors (i.e., visual and vocal behaviors) and more successful at controlling
their verbal behaviors. This would explain why motivated deceivers experience
more deceptive success when receivers are exposed to only the verbal channel
of communication (DePaulo, Kirkendol, Tang, & O'Brien, 1988). Unsanctioned
liars in the present study may have attempted to control their entire nonverbal
presentation but were only partially successful. That is, unsanctioned
liars, when compared to sanctioned liars, were successful controlling the
verbal channel (less speech errors, less speech hesitations, less self
and other references) but less successful at controlling their nonverbal
presentation (more postural shifts, less eye gaze). Unfortunately, the
study presented herein failed to account for deceiver motivation or deceiver
arousal, thus, the contention that motivation to deceive is higher for
unsanctioned liars than for sanctioned liars is speculative until further
data proves otherwise.

Are Sanctioned Liars More Motivated?

Perhaps we were mistaken to conceptualize unsanctioned liars as more motivated
than sanctioned liars. It may be that sanctioned liars were more motivated
to lie than unsanctioned liars. Participants in the sanctioned lie condition
may have viewed the opportunity to lie without consequence as a challenge
and therefore found it arousing and exciting. Research by Tomaka, Blascovich,
Kelsey, and Leitten (1993) has shown that individuals often appraise a
situation in one of two ways: as threatening or as challenging. Threat
appraisals are those in which the perception of danger exceeds the perception
of abilities or resources to cope with the stressor. By contrast, challenge
appraisals are those in which the perception of danger does not exceed
the perception of resources or abilities to cope (Tomaka et al., 1993).
The unsanctioned liar may have seen the situation as threatening and the
sanctioned liar may have seen the situation as challenging. Research has
shown that these appraisals often determine stress levels, behavior, and
success on a task (Tomaka, Blascovich, & Kelsey, 1992; Tomaka et al., 1993;
Young & deTurck, 1996).

Future Research

The next logical step to build on the present findings would be to replicate
the original DePaulo et al. (1983) experiment with one exception: senders'
lies should be sanctioned or unsanctioned and senders' motivation should
be manipulated. Thus, sanctioning would be crossed with sender motivation.
It would also be interesting to investigate how students in each deceptive
paradigm appraise the communication situation using Tomaka et al.'s (1993)
appraisal measure. If, in fact, unsanctioned liars are more highly motivated
than sanctioned liars, detection rates should be higher when receivers
are exposed to the nonverbal channel (audio, audiovisual, visual) than
when receivers are exposed only to the verbal channel (written).

Notes

1. Participants in the experiment were limited to strangers to avoid a
possible confound with respect to relational familiarity. Research suggests
that the dynamics of deception may be differ as a function of relational
familiarity (cf. Comadena, 1982; Levine & McCornack, 1992; McCornack &
Parks, 1986).

2. This procedure is similar to a technique first introduced by Exline
et al., 1971 whereby participants were implicated for cheating on a dot-estimation
task (see also deTurck & Miller, 1985; Dulaney, 1982; Feeley & deTurck,
1995; Stiff & Miller, 1986).

3. No participants left the experiment, in fact, most participants seemed
excited about the opportunity to lie.

4. This was done in an effort to raise suspicion on the part of detectors.
As it was, ERs still evaluated EEs to be telling the truth 83% of the time.
The results of lie detection accuracy rates by veracity condition can be
found in Feeley (1997b).

5. Results from EEs' self-reports indicated no differences in involvement,
nervousness, impression management, competence, and fluency across experimental
conditions; for an extended discussion see Feeley (1997b).

References

Bell, K.L., & DePualo, B.M. (1996). Liking lying. Basic and Applied Social
Psychology, 18, 243-266.

Buller, D. B., & Aune, R. K. (1987). Nonverbal cues to deception among
intimates, friends, and strangers. Journal of Nonverbal Behavior, 11, 269-290.
Burgoon, J. K., & Buller, D. B. (1994). Interpersonal deception: III. Effects
of deceit on perceived communication and nonverbal behavior dynamics. Journal
of Nonverbal Behavior, 18, 155-184.

Cohen, J. (1969). Statistical power analysis for the behavioral sciences.
New York: Academic Press.

Comadena, M. E. (1982). Accuracy in detecting deception: Intimate and friendship
relationships. In M. Burgoon (Ed.), Communication Yearbook 6 (pp. 446-472)
Beverly Hills, CA: Sage.

DePaulo, B. M., & Bell, K. L. (1996). Truth and investment: Lies are told
to those who care. Journal of Personality and Social Psychology, 71.

DePaulo, B. M., Kashy, D. A., Kirkendol, S. E., Wyer, M. M., & Epstein,
J. E. (1996). Lying in

everyday life. Journal of Personality and Social Psychology, 70, 979-995.
DePaulo, B. M. & Kirkendol, S. E. (1989). The motivational impairment effect
in the communication of deception. In J. Yuille (Ed.), Credibility assessment
(pp. 51- 70). Norwell, MA: Kluwer.

DePaulo, B. M., Kirkendol, S. E., Tang, J., & O'Brien, T. P. (1988). The
motivational impairment effect in the communication of deception: Replications
and extensions. Journal of Nonverbal Behavior, 12, 177-202.

DePaulo, B. M., Lanier, K., & Davis, T. (1983). Detecting the deceit of
the motivated liar. Journal of Personality and Social Psychology, 45, 1096-1103.
DePaulo, B. M., & Rosenthal, R. (1979). Telling lies. Journal of Personality
and Social Psychol

ogy, 37, 1713-1722.

DePaulo, B. M., Stone, J. I., & Lassiter, G. D. (1985). Deceiving and detecting
deceit. In B.R. Schlenker (Ed.), The self and social life (pp. 323-370).
New York: McGraw-Hill. deTurck, M. A., & Miller, G. R. (1985). Deception
and arousal: Isolating the behavioral corre

lates of deception. Human Communication Research, 12, 181 201. Dulaney,
E. F. (1982). Changes in language behavior as a function of veracity. Human
Communication Research, 9, 75-82.

Ekman, P. (1991). Telling lies. New York: W. W. Norton.

Exline, R. E., Thibaut, J., Hickey, C. B., & Gumpert, P. (1970). Visual
interaction in relation to Machiavellianism and an unethical act. In R.
Christie & F. L. Geis (Fds.), Studies in Machiavellianism (pp. 53-75).
New York: Academic Press. Feeley, T. H. (1997a, November). Choosing deceptive
communication. Paper presented to National Communication Association, Chicago,
IL.

Feeley, T. H. (1997b). Exploring sanctioned and unsanctioned lies in interpersonal
deception.

Communication Research Reports, 13, 164-173.

Feeley, T. H. & deTurck, M. A. (1985). Global cue usage in behavioral lie
detection. Communication Quarterly, 43, 420-430.

Feeley, T. H., deTurck, M. A., & Young, M. J. (1995). Behavioral familiarity
in lie detection.

Communication Research Reports, 12, 160-169.

Fiedler, K., & Walka, I. (1993). Training lie detectors to use nonverbal
cues instead of global heuristics. Human Communication Research, 20, 199-223.


Kraut, R. (1980). Humans as lie detectors: Some second thoughts. Journal
of Communication, 30, 209-216.

Levine, T. R., & McCornack, S. A. (1992). Linking love and lies: A formal
test of the McCornack and Parks model of deception detection. Journal of
Social and Personal Relationships, 9, 143-154.

McCornack, S. A., & Parks, M. R. (1986). Deception detection and the other
side of trust. In M. L. Mclaughlin (Ed.), Communication Yearbook 9 (pp.
377-389). Beverly Hills, CA: Sage.

Miller, G. R., & Stiff, J. B. (1993). Deceptive communication. Newbury
Park, CA: Sage. Stiff, J. B., Corman, S. R., Krizek, R., & Snider, E. (1994).
Individual differences and changes in

nonverbal behavior: Unmasking the changing faces of deception. Communication
Research, 21, 555-581.

Stiff, J. B., & Miller, G. R. (1986). "Come to think of it . . .": Interactive
probes, deceptive communication, and deception detection. Human Communication
Research, 12, 339357.

Tomaka, J., Blascovich, J., & Kelsey, R. M. (1992). Effects of self-deception,
social desirability, and repressive coping on psychophysiological reactivity
to stress. Personality and Social Psychology Bulletin, 18, 616-624.

Tomaka, J., Blascovich, J., Kelsey, R. M., & Leitten, C. L. (1993). Subjective,
physiological, and behavioral effects of threat and challenge appraisal.
Journal of Personality and Social Psychology, 65, 1-13.

Young, M. J., & deTurck, M. A. (1996, November). The effects of stress
and coping on deceptive communication. Paper presented to Speech Communication
Association Convention, San Diego, CA.

Zuckerman, M., DePaulo, B. M., & Rosenthal, R. (1981). Verbal and nonverbal
communication of deception. In L. Berkowitz (Ed.), Advances in experimental
social psychology (Vol. 14, pp. 1-59). New York: Academic Press.

Zuckerman, M., & Driver, R. (1985). Telling Lies: Verbal and nonverbal
correlates of deception. In A. W. Siegman & S. Feldstein (Eds.), Nonverbal
communication: An integrated perspective (pp. 129-147). Hillsdale, NJ:
Lawrence Erlbaum.

Thomas H. Feeley, Department of Communication, State University of New
York at Geneseo. Mark A. deTurck, Department of Communication, State University
of New York at Buffalo.

This paper is based on the first author's doctoral dissertation directed
by the second author. This research was supported by a grant from the Mark
Diamond Research Foundation and a grant from the Geneseo Foundation both
to the first author. The authors would like to thank Aldert Vrij, Bella
DePaulo, Steve McCornack, and Ron Riggio for comments on an earlier version
of this manuscript.

Address correspondence to Thomas H. Feeley, Department of Communication,
State University of New York at Geneseo, Geneseo, NY 14454; E-mail: feeley@uno.cc.geneseo.edu.


Reproduced with permission of the copyright owner.






Individual differences in hand movements during deception
Journal of Nonverbal Behavior
New York
Summer 1997

--------------------------------------------------------------------------------

Authors:                  Aldert Vrij

Authors:                  Lucy Akehurst

Authors:                  Paul Morris

Volume:                   21

Issue:                    2

Pagination:               87-102

ISSN:                     01915886

Subject Terms:            Nonverbal communication
                          Behavior
                          Personality
                          Cognition & reasoning
                          Hands


Abstract:

The influence of two personality traits on making hand movements during
deception was examined.  It was hypothesized that individuals with high
public self-consciousness and individuals skilled in controlling their
behavior would make fewer hand movements during deception.
Copyright Human Sciences Press, Inc. Summer 1997

Full Text:

ABSTRACT: This article addresses the influence of 2 personality traits
on making hand movements during deception, namely public self-consciousness
and ability to control behavior. It was hypothesized that especially individuals
with high public self-consciousness and individuals who are skilled in
controlling their behavior would make fewer hand movements during deception
compared to truth-telling. A total of 56 participants were interviewed
twice; in one interview they told the truth and in the other interview
they lied. Before the interviews the participants completed a personality
inventory to measure their levels of public self-consciousness and ability
to control their behavior. The results supported the hypotheses. Some implications
of these findings are discussed.

Recent studies concerning the relationship between hand movements and deception
suggest that a decrease in such movements indicates deception (Davis &
Hadiks, 1995; Ekman, 1988, 1989; Hofer, Kohnken, Hanewinkel, & Bruhn, 1992;
Vrij, 1995; Vrij, Semin, & Bull, 1996). These findings are in direct contrast
to beliefs among observers that liars tend to increase their movements
during deception (Akehurst, Kohnken, Vrij, & Bull, 1996; DePaulo, Stone,
& Lassiter, 1985; Vrij & Semin, 1996; Zuckerman, DePaulo, & Rosenthal,
1981).

The control and cognitive load frameworks have been offered to explain
differences in hand movements during deception and truth-telling. According
to the control framework, liars try to control their body language to avoid
giving off possible nonverbal cues to deception and to make a credible
(reliable) impression (DePaulo,1988, 1992; DePaulo & Kirkendol, 1989; Ekman,
1989; Kohnken, 1990). Liars, for instance, believe that movements will
make them appear suspicious. Therefore, they will move very deliberately
and tend to avoid those movements which are not strictly essential, thus
resulting in an unusual degree of rigidity and inhibition. In agreement
with this explanation, Vrij et al. (1996) found that decreases in subtle
non-functional hand movements during deception were associated with the
extent to which the participants had tried to control their behavior.

The cognitive load framework (Burgoon, Kelley, Newton, & KeelyDyreson,
1989; Ekman & Friesen, 1972; Kohnken, 1989) emphasizes that deception is
a cognitively complex task. It assumes that it is more cognitively difficult
to fabricate a plausible and convincing lie consistent with everything
the observer knows or might find out than to tell the truth. There is evidence
to suggest that people engaged in cognitively complex tasks make fewer
hand and arm movements; the cognitive load results in a neglect of body
language, reducing overall animation (Ekman & Friesen, 1972).

Unfortunately, the majority of deception studies use a between-subjects
design in order to investigate behavioral differences between deceivers
and truth tellers, that is, some people tell the truth and some people
lie and the behavioral differences between the two groups are determined.
The fact that research reveals that deception is associated with a decrease
in hand movements does not imply that all people show a decrease in these
movements during deception. It may well be the case that some people show
an increase in hand movements during deception, or show no differences
at all between truth-telling and lying. This raises the question: How many
people show a decrease in hand movements during deception and how many
people show an increase in hand movements during deception? A between-subjects
research paradigm cannot answer this question, because in such a paradigm
the subject is either lying or telling the truth. Only a within-subjects
design, in which a participant takes part in both the honest and deceptive
interviews, can offer an answer to this question. If different behavioral
patterns occur, that is, if some people show a decrease in hand movements
during deception whereas other people show an increase in hand movements
during deception, this then raises the question: Which individual differences
account for these different behavioral styles during deception? Both of
the above questions will be addressed in the present experiment.

The first question (how many people show a decrease in hand movements during
deception) is so far, to our knowledge, only addressed by Ekman, O'Sullivan,
Friesen, & Scherer (1991), who used a within-subjects design in their experiment.
They found that 39% of participants showed fewer hand movements during
deception than truth-telling, 26% showed more hand movements during deception
than truth-telling, and 35% showed no substantial differences in hand movements
between both interviews.1 In order to gain further insight into this issue
we merged and reanalyzed the datafiles of three experiments using within-subjects
designs (Vrij, 1995, 1996; Vrij et al., 1996), in which we examined differences
in hand movements between truth tellers and liars (see the Method section
for a description of the procedure used in these studies). A total of 218
participants (all college students) participated in the three experiments,
and of these 52% showed a decrease in hand movements during deception,
30% an increase in hand movements during deception, and 18% showed no difference
in hand movements.2 These data suggest that although more people decrease
than increase their hand movements during deception, a substantial number
of people exhibit an increase in hand movements during deception or display
no behavioral differences at all between truthtelling and deception. These
findings again make relevant the question: Which individual differences
account for these different behavioral patterns?

Deception is a complex social activity. Several authors (Ekman, 1989; Knapp,
Hart, & Dennis, 1974; Kohnken, 1989; Kraut & Poe, 1980; Riggio & Friedman,
1983) emphasize that deception evokes stress. Stress can be a result of
guilt (feeling guilty over doing things that one is not allowed to do),
or a result of fear of being caught. Therefore, deceivers will show a tendency
to display nervous behavior (including making hand movements) and observers
will expect deceivers to show such behavior and will therefore look for
cues of nervous behavior. As a result, a liar, in order not to get caught,
has to suppress and avoid showing signs of nervous behavior. Hence, two
aspects are important, namely (1) being aware of how to make a credible
impression, and (2) possessing the skills necessary to make a credible
impression.

We may assume that people high in public self-consciousness will know how
to make a credible impression. Public self-consciousness refers to the
ability of persons to become aware of another's perspective and to act
from that perspective (Fenigstein, Scheier, & Buss, 1975). Since observers
believe that an increase in hand movements indicates deception (Akehurst
et al.,1996; Vrij & Semin,1996), we may expect that individuals high in
public self-consciousness will realize this and it is therefore hypothesized
that individuals with high public self-consciousness will make fewer hand
movements during deception (Hypothesis 1). This hypothesis is based upon
the control approach. Also, on the basis of the cognitive load approach,
it may be predicted that people high in public self-consciousness especially
will make fewer movements during deception, because being concerned about
making a good impression on somebody else requires a lot of cognitive attention
(Patterson, 1995), which then necessitates a decrease in such attention
to one's own movements.

People will differ in how successful they are in controlling their behavior.
We therefore hypothesized that individuals who are skilled in controlling
their behavior will be more successful in suppressing their hand movements
during deception than people who are not so skilled in controlling their
behavior (Hypothesis 2).3

Several studies have been conducted examining individual differences displayed
during deception to date (Exline, Thibaut, Hickey, & Gumbert, 1970; Knapp,
Hart, & Dennis, 1974; O'Hair, Cody, & McLaughlin, 1981; Riggio & Friedman,
1983; Siegman & Reynolds, 1983); most of these studies dealt with Machiavellianism
and none of them focused on hand movements. At least two studies examined
a related issue, namely individual differences in deception ability (Riggio,
Tucker, & Throckmorton, 1988; Vrij, 1993). In both studies observers (undergraduate
students in Riggio's et al. study, police detectives in Vrij's study) watched
videotapes of people who had been instructed either to tell the truth or
to lie. After each, the observers indicated whether the person was lying.
In both studies the people who appeared on the tape were administered a
number of standardized social skill instruments. Vrij found that people
high in public self-consciousness made a more credible impression on observers
than people low in public self-consciousness. Riggio et al. found that
people high in controlling their behavior were more successful deceivers
than people low in controlling their behavior. Both studies did not provide
information about the behavior shown by the people on the tape and thus
an explanation of the findings has to be speculative. An explanation in
line with the present hypotheses is that people high in public self-consciousness
and people high in controlling their behavior refrained from making hand
movements during deception and therefore made a credible impression on
observers.

Method

Subjects

A total of 56 participants (college students) participated in the study
(20% male). The average age was 24 years (SD = 7 years).

Procedure

The experiment was conducted at the University of Portsmouth in Portsmouth
(United Kingdom). Participants were asked to take part in a study investigating
their ability to deceive. Before the experiment started they were requested
to fill out a questionnaire concerning background characteristics (gender
and age) and personality traits (public self-consciousness and ability
to control their behavior). The experimental setting was an interview similar
to the ones used in previous studies (Vrij, 1995, 1996; Vrij et al., 1996).
Participants were given the following instructions: "We are doing an experiment
to investigate people's ability to deceive. In a minute, you will be interviewed
twice about the possession of a small set of headphones. You will actually
have the set of headphones in your possession during one interview while
during the other interview you will not have the set of headphones in your
possession. Both times you have to deny the possession of the set of headphones."


The order of lying versus telling the truth was counterbalanced. A total
of 27 subjects received the set of headphones before the first interview
with the request to hide them carefully, while the other 29 subjects received
the set of headphones just before the second interview. The latter group
had seen the set of headphones prior to the first interview. After the
instructions the experimenter took the subject to the interview room. The
interviewer (a male lecturer in the Psychology Department of the University
of Portsmouth who was unaware of the hypotheses tested in the experiment)
asked the subject to take a seat and started the first interview. All interviews
were standardized in that the following six questions were asked: (1) "Do
you have the set of headphones in your possession?" (2) "Are you telling
the truth?" (3) "Tell me exactly what you have in your possession." (4)
"You forgot to mention the set of headphones, didn't you?" (5) "Are you
telling me that you don't have the headphones in your possession?", (6)
"Are you absolutely sure that you are telling me the truth?" After the
first interview the subject left the interview room for a short period
of time either to return the set of headphones to the experimenter (if
the subject was in possession of the set of headphones during the first
interview) or to receive the set of headphones (if the subject was not
in possession of the set of headphones during the first interview). Next
the subject re-entered the interview room for the second interview. The
second interview was identical to the first one. Both interviews were videotaped.
On the videotapes the subjects' whole bodies were visible. The average
length of the honest interviews was 26 seconds (SD = 4 seconds), and the
average length of the deceptive interviews was 27 seconds (SD = 5 seconds).


Independent Variables

The independent variables were (A) the type of interview (lying vs. telling
the truth), and (B) the order in which the honest and deceptive interviews
were carried out, that is, lying-telling the truth or telling the truthlying.


Personality Traits

Public self-consciousness was measured with the following six questions
using true/false rating scales (Fenigstein et al., 1975): (1) I'm concerned
about what other people think of me, (2) I usually worry about making a
good impression, (3) I'm concerned about the way I present myself, (4)
I'm self-conscious about the way I look, (5) I'm usually aware of my appearance,
and (6) One of the last things I do before I leave my house is look in
the mirror. The six items were clustered into one scale, "public self-consciousness"
(Cronbach's alpha = .73). The ability to control behavior was measured
with 16 items using true/false rating scales, derived from Briggs, Cheek,
and Buss' (1980) "other-directedness" (items 1 to 6) and "acting" (items
7 to 10) scales and Riggio's (1986) "emotional control" (items 11 to 13)
and "social control" (items 14 to 16) scales: (1) In different situations
and with different people, I often act like very different persons, (2)
In order to get along and to be liked, I tend to be what people expect
me to be rather than anything else, (3) I'm not always the person I appear
to be, (4) Sometimes I put on a show to impress or entertain people, (5)
Even if I am not enjoying myself, I often pretend to be having a good time,
(6) I may deceive people by being friendly when I really dislike them,
(7) I would probably make a good actor, (8) I have considered being an
entertainer, (9) I can make impromptu speeches on topics about which I
have almost no information, (10) I can look anyone in the eye and tell
a lie with a straight face (as long as it is a white lie), (11) I am able
to conceal my true feelings from just about anyone, (12) I am very good
at maintaining a calm exterior, even when upset, (13) When I am really
not enjoying myself at a social function, I can still make myself look
as if I am having a good time, (14) I find it very easy to play different
roles at different times, (15) When in a group of friends I am often spokesperson
for the group, (16) I can fit in with all types of people, young and old,
rich and poor. The sixteen items were clustered into one scale, "ability
to control behavior" (Cronbach's alpha = .72).

Firstly, we calculated the correlation between both scales, which was not
significant r(56) = .05, ns. Secondly, we dichotomized both scales. Those
who had a score lower than the mean score (M = 4.53) on the public self-consciousness
(PSC) scale were allocated to the "low PSC" group (40% of the subjects),
and the others were allocated to the "high PSC" group (60% of the subjects).
Similarly, those who had a score lower than the mean score (M = 7.95) on
the ability to control behavior (ACB) scale were allocated to the "low
ACB" group (50% of the subjects), and the others were allocated to the
"high ACB" group (50% of the subjects). In total, 12 subjects were allocated
to the "low PSC-low ACB" cell, 16 subjects to the "high PSC-low ACB" cell,
11 subjects to the "low PSC-high ACB" cell and 17 subjects to the "high
PSC-high ACB" cell.

Dependent Variable

The dependent variable was the hand movements displayed by the subjects.
These movements were scored in detail by two independent coders using the
videotapes. Scoring was conducted by utilizing a coding scheme previously
used in other studies (Vrij, 1995, 1996; Vrij et al., 1996). The coders
scored the frequency of hand/finger movements (also referred to as hand
movements): a hand/finger movement is a movement of a hand or finger without
the arm being moved; every single hand/finger movement was scored, with
simultaneous movements of more fingers being scored as one movement; continuing
movements (rubbing one's hands together and fidgeting) were scored every
two seconds. A reliability measure was calculated for the two coders, r
= .98, p < .001. These movements, which have been called "subtle hand and
finger movements" (Vrij, 1995), differ somewhat from illustrators or adaptors
(Ekman & Friesen, 1969). Illustrators are gestures which accompany speech
and accent. They were not counted, because these movements include arm
movements. For similar reasons, adaptors which include arm movements (scratching
the head, face, wrists and so on) were not counted either. We did count,
however, adaptors which do not include arm movements.

The reported behavioral scores were based on the average scores of the
two coders. The frequency of hand movements reported below was calculated
on a per minute basis to correct for length of the interview.4

We monitored whether the deception interview evoked stress by asking the
participants, after both interviews, to indicate how nervous they felt
during (a) the truth- telling interview, and (b) the deception interview.
Answers were given on 7-point rating scales, ranging from (1) not at all
nervous to (7) very nervous. A mixed-model analysis of variance (ANOVA)
with condition (truth-telling vs. lying) and order (truth-telling-lying
vs. lying-truth-telling) as the independent factors, and stress as the
dependent variable, showed a significant effect for condition. The participants
said they felt more stressed during the deception interview (M = 4.79)
than during the truth-telling interview (M = 3.43), F(1, 54) = 61.83, p
< .01. Results

A mixed-model ANOVA with condition (truth-telling vs. lying) and order
(truth-telling-lying vs. lying-truth-telling) as the independent factors
and hand movements as the dependent variable indicated significant main
effects for condition, F(1, 54) = 5.16, p < .04, and order, F(1, 54) =
4.67, p < .05, and a significant condition X order interaction effect,
F(1, 54) = 4.89, p < .05. The mean scores regarding the condition factor
revealed that, as expected, the subjects showed fewer hand movements during
the deceptive interviews (M = 4.63) than during the honest interviews (M
= 7.62). The mean scores concerning the order effect showed that the subjects
exhibited the most hand movements when deception preceded truthtelling
(M = 8.41 vs. M = 3.99). The mean scores and contrast effects regarding
the interaction effect revealed that the subjects showed fewer hand movements
during the deceptive interviews (M = 5.36) than during the honest interviews
(M = 11.47) when deception preceded truth-telling, F(1, 26) = 6.29, p <
..05, whereas the difference in hand movements during the deceptive interviews
(M = 3.95) and during the honest interviews (M = 4.03) was not significant
when truth-telling preceded deception, F(1, 26) = .00, ns. We found a similar
interaction effect in previous studies (Vrij, 1995, 1996). The following
explanation may be plausible. The experimental design implied that the
students already knew during the second interview which questions would
be asked. As a result of this knowledge, the students may have been better
able to behave frankly, resulting in unforced hand and finger movement.
This reasoning, however, is not completely satisfactory because it implies
that the students would feel less tension during the second interview as
well, which was the case in our previous experiments but not in this experiment.


We also calculated how many subjects showed a decrease or increase during
deception by subtracting the number of hand movements made on a per-minute
basis during the honest interview from the number of hand movements made
on a per-minute basis during the deception interview. The results showed
that 54% of participants made fewer movements during the deception interview
than during the honest interview, 25% made more movements during the deception
interview than during the honest interview, and 21 % showed the same number
of movements during both interviews. People who showed the same number
of movements during both interviews did not make any hand movement at all
during either interview. This categorization (a decrease in hand movements
during deception, an increase in hand movements during deception, and no
difference in hand movements between interviews) is used in further analyses.


In order to test Hypotheses 1 and 2 a log-linear analysis was conducted
with hand movements (decrease, increase, no difference), public self-consciousness
(high vs. low), and ability to control behavior (high vs. low) as variables.5
The model showed two significant effects, namely a Hand Movements X Public
Self-Consciousness effect, X2(2, N=56) = 12.29, p < .01, and a Hand Movements
X Ability to Control Behavior effect, X2(2, N=56) = 6.40, p < .05. The
final model included these two effects (goodness-of-fit: X2(3, 56) = 3.37,
p = .33). Tables 1 and 2 give the frequency distributions concerning both
effects.

Table 1 reveals that a majority of people with high public self-consciousness
made fewer hand movements during deception than truth-telling. This supports
Hypothesis 1. Table 1 further shows that many people with high public self-consciousness
managed to make the same number of hand movements during deception and
truth-telling (they did not make any hand movements at all during both
interviews). Moreover, Table 1 shows that only a minority of people with
high public self-consciousness made more hand movements during deception
than truth-telling. The results for people with low public self-consciousness
are less clear. Table 1 shows that 48% of them made more movements during
deception, whereas 44% made fewer movements during deception.

Table 2 shows that people with high ability to control their behavior also
made fewer hand movements during deception than truth-telling. This supports
Hypothesis 2. Moreover, Table 2 shows that only a minority of people with
high ability to control their behavior made more hand movements during
deception than truth-telling. Again, the results for people with low ability
to control their behavior are less clear. Table 2 shows that 39% of them
made more movements during deception, whereas 43% made fewer movements
during deception.

The Hand Movements x Public Self-Consciousness x Ability to Control Behavior
effect was not significant, X2(2, N=56) = 2.44, p = .29, meaning that the
two personality factors had an additive effect on hand movement patterns.
Nevertheless it is interesting to look at the magnitudes of the differences
when both personality factors are taken into account. Table 3 is therefore
shown for exploratory purposes.

Firstly, Table 3 makes clear that showing the same number of hand movements
during both interviews (i.e., refraining from making any hand movements
at all during both interviews) is a behavioral pattern typically pursued
by people with high public self-consciousness, and not so much by those
with an ability to control their movements. Secondly, there was a difference
in behavioral patterns between people who are low in both public self-consciousness
and ability to control behavior and people who are high in both public
self-consciousness and ability to control behavior. For instance, 75% of
the first group and only 6% of the latter group showed an increase in hand
movements during deception. The difference in frequency distribution between
these two groups was highly significant, X2(2, N=29) = 18.28, p < .01.
The frequency distributions of the "low PSC-high ACB" and "high PSC-low
ACB" groups did not differ significantly, X2(2, N=27) = .65, ns.

Discussion

To our knowledge no prior study of how behavior differs in honest and deceptive
interactions has examined individual differences in hand movements during
deception. This article addressed the influence of two personality traits
on deceptive behavior, namely public self-consciousness and ability to
control behavior. The findings revealed that both traits were important
in deceptive behavior. People who were high in public self-consciousness
made fewer hand movements during deception compared to truth-telling, and
people who had a high ability to control their behavior made fewer hand
movements during deception compared to truth-telling.

One explanation for why people high in public self-consciousness make fewer
hand movements during deception compared to truth-telling than their counterparts
is that they experience much cognitive load during deception due to the
fact that they not only have to deceive the other but also have to think
how to make a credible impression. An alternative explanation is that they
realize that observers will pay attention to these movements in order to
catch a lie. By refraining from making these movements they try to make
the lie detection task for observers a more difficult one (the findings
show that many of them try to avoid making any hand movements at all).
This suggestion has an important implication. It is often suggested that
a decrease in hand movements during deception happens involuntarily (both
the control and the cognitive approach assume that the decrease in hand
movements occurs involuntarily). It may, however, be the case that some
people show such a decrease on a voluntary basis in order to fool the observer.
Our data do not enable us to determine which of these explanations is the
most plausible one to explain the outcomes. Therefore, further research
is needed concerning this issue. Anticipating such a study, one may argue
that the cognitive load explanation seems to be the least plausible solution,
as Kashy and DePaulo's (1996) study about individual differences in lie-telling
in everyday situations revealed that people high in public self-consciousness
do frequently tell lies. We may assume that this experience will make telling
lies easier and generally not very demanding for them.

[IMAGE TABLE] Captioned as: TABLE 1

[IMAGE TABLE] Captioned as: TABLE 2

[IMAGE TABLE] Captioned as: TABLE 3

A plausible explanation as to why relatively large numbers of people with
a low ability to control their behavior showed more movements during deception
than during truth-telling than their counterparts is that they did not
successfully manage to suppress their nervous movements during deception.
In order to validate this explanation, we calculated correlations between
making hand movements during deception and self-ratings of stress during
deception for both people with a low and for people with a high ability
to control their behavior. The findings showed, as suggested, a positive
correlation between making hand movements during deception and being stressed
during deception for people with a low ability to control their behavior
r(28) = .32, p < .05, whereas such a relationship did not exist for people
with a high ability to control their behavior, r(28) = .06, ns.

The findings revealed that 25% of the participants who were low in both
public self-consciousness and ability to control behavior showed a decrease
in hand movements during deception. This findings is perhaps surprising,
as theoretically these people did not seem to bother about how to make
a credible impression and didn't believe that they were able to control
their behavior. We might therefore expect an increase in these participants'
hand movements during deception (due to the fact that they were nervous
during deception and therefore may have displayed nervous behavior). Despite
the previous paragraph we believe that these people accidently showed a
decrease in hand movements, perhaps due to cognitive load. It could be
that they found it difficult to fabricate a lie, and therefore used to
measure ability to control behavior. It may be argued that should also
someone who scores low on the "ability to control behavior" scale not only
perceives himself/herself as bad do not controlling behavior, but should
also experience much cognitive load during deception, because we may assume
that these people do not frequently deceive and do not perceive deception
to be a routine approach to resolving dilemmas presented in communicative
contexts.6 In line with this reasoning is Kashy and DePaulo's (1996) recent
finding that people low in "other-directedness" (part of our ability to
control behavior scale) do not frequently deceive other people.

We are aware that our findings are obtained in an experimental situation
which may well differ from real-life situations. What is the ecological
validity of our findings, or in other words, what do these results say
about daily life situations? It has been frequently argued that behavioral
differences between truth-tellers and liars can only be expected if liars
are stressed (as a result of guilt or fear) and/or if some mental effort
is required to formulate the lie. If both stress and mental effort are
absent, behavioral cues to deception will probably not occur (DePaulo,
Kashy, Kirkendol, Wyer, & Epstein, 1996; Ekman, 1985; Vrij, 1997). DePaulo
et al. (1996) recently showed that the majority of lies told in daily life
situations are little lies, that is, they are unarousing and do not require
much mental effort. As a result, they will probably not be associated with
behavioral cues. Our research paradigm differs from this situation in the
sense that our subjects experienced some stress when they were lying. Of
course, this happens in daily life situations as well; not all everyday
lies are unarousing, as some lies do result in some level of stress, while
other lies are even more serious and may therefore result in high levels
of stress. We believe that our research paradigm provides insight into
how people behave in situations in which some stress is involved, such
as situations in which the liar feels some delight about being able to
fool somebody else or situations in which discovering the lie will result
in some (minor) negative consequences (unflattering situation for the liar,
hurting somebody elses' feelings and therefore creating a somewhat awkward
situation in the interaction with the other, some form of minor punishment,
and so on).

Recently, researchers have started to criticize the most popular research
paradigm used to study deceptive communication. They have argued that more
attention is needed concerning differences in deceptive behavior between
lies with high or low stakes for being caught (Ekman et al., 1991), and
differences in deceptive behavior between different types of lies, that
is, falsification, equivocation, and concealment (Buller, Burgoon, White,
& Ebesu, 1994). It has also been criticized that the role of the interviewer
is usually too limited, that is, a more active role for the interviewer,
in which he/she is allowed to interact with the potential deceiver, is
needed (Kalbfleisch, 1994; Patterson, 1995). Considering our findings we
would suggest that more attention is also needed regarding individual differences
among deceivers in deception research.

NOTES

1. A "substantial" difference means a difference between the two interviews
of more than twice the standard error of measurement.

2. Unlike Ekman et al. (1991) we classified all differences (even the smallest
ones) as differences. These findings and Ekman et al.'s (1991 ) findings
show almost the same ratio (6: 4)

of people who decrease or increase their movements during deception when
the "no difference" categories are disregarded.

3. The "ability to control behavior" construct is related to the "self-monitoring"
construct (Snyder, 1974), according to the definition of self-monitoring
given by Briggs, Cheek, and Buss (1980, p. 6): "The prototypic high self-monitoring
individual . . . is particularly sensitive to the expression and self-presentation
of relevant others in social situations and uses these cues as guidelines
for self-monitoring (that is, regulating and controlling) his or her own
verbal and nonverbal self-presentation." According to Briggs et al. (1980),
the SelfMonitoring Scale is composed of three factors, namely Acting, Extraversion,
and OtherDirectedness. We included Acting and Other-Directedness (Briggs
et al., 1980) in our ability to control construct. Instead of Extraversion
(the third factor of the self-monitoring scale) we included Riggio's (1986)
"control" measures to complete the ability to control behavior scale. We
did this because we believed, after reading the control and extraversion
items, that the control items were more related to what we were looking
for than the extraversion items. See the Method section for a full description
of the ability to control behavior scale we used.

4. Aggregating behaviors across the entire interaction might be considered
a potential limitation because differences in nonverbal behavior during
the course of the interaction cannot be analyzed. To gain insight into
these differences requires a segmentation analysis. We realize that we
lose some information by dichotomizing the personality variables and trichotomizing
the hand movements variable. However, we believe that this is the best
way to test our hypothesis. At first sight, a perhaps neater version to
test the hypothesis would be to use the difference between truth and lie
for each person as the dependent variable. The problem with such an analysis
is, however, that it does not take into account the number of people that
showed an increase or decrease in hand movements during deception (the
relevant information in this context) but the extent of increase or decrease
in hand movements these people showed during deception, which is a different
and in this context irrelevant issue.

6. Thanks to an anonymous reviewer for this useful comment.

REFERENCES

Akehurst, L., Kohnken, G., Vrij, A., & Bull, R. (1996). Lay persons' and
police officers' beliefs regarding deceptive behaviour. Applied Cognitive
Psychology, 10, 461-471. Buller, D. B., Burgoon, J. K., White, C. H., &
Ebesu, A. S. (1994). Interpersonal deception VII. Behavioral profiles of
falsification, equivocation, and concealment. Journal of Language and Social
Psychology, 13, 366-395.

Burgoon, J. K., Kelley, D. L., Newton, D. A., & Keely-Dyreson, M. P. (1989).
The nature of arousal and nonverbal indices. Human Communication Research,
16, 217-255. Briggs, S. R., Cheek, J. M., & Buss, A. H. (1980). An analysis
of the Self-Monitoring Scale. Journal of Personality and Social Psychology,
38, 679-686. Davis, M., & Hadiks, D. (1995). Demeanor and credibility.
Semiotica, 106, 5-54. DePaulo, B. M. (1988). Nonverbal aspects of deception.
Journal of Nonverbal Behavior, 12, 153-162.

DePaulo, B. M. (1992). Nonverbal behavior and self-presentation. Psychological
Bulletin, 111, 233-243.

DePaulo, B. M., Kashy, D. A., Kirkendol, S. E., Wyer, M. M., & Epstein,
J. A. (1996). Lying in everyday life. Journal of Personality and Social
Psychology, 70, 979-995. DePaulo, B. M., & Kirkendol, S. E. (1989). The
motivational impairment effect in the communication of deception. In J.
C. Yuille (Ed.), Credibility assessment (pp. 51-70). Dordrecht: Kluwer
Academic Publishers.

DePaulo, B. M., Stone, J. L., & Lassiter, G. D. (1985). Deceiving and detecting
deceit. In B. R. Schenkler (Ed.), The self and social life (pp. 323-370).
New York: McGraw-Hill. Ekman, P (1985). Telling lies. New York. W. W. Norton.


Ekman, P (1988). Lying and nonverbal behavior: Theoretical issues and new
findings. Journal of Nonverbal Behavior, 12, 163-176.

Ekman, P. (1989). Why lies fail and what behaviors betray a lie. In J.
C. Yuille (Ed.), Credibility Assessment (pp. 71-82). Dordrecht: Kluwer
Academic Publishers. Ekman, P., & Friesen, W. V. (1969). The repertoire
of nonverbal behavior: Categories, origins,

usage, and coding. Semiotica, 1, 49-98.

Ekman, P., & Friesen, W. V. (1972). Hand movements. Journal of Communication,
22, 353374.

Ekman, P, O'Sullivan, M., Friesen, W. V., & Scherer, K. R. (1991). Face,
voice, and body in detecting deceit. Journal of Nonverbal Behavior, 15,
125-135. Exline, R., Thibaut, J., Hickey, C., & Gumpert, P (1970). Visual
interaction in relation to Machiavellianism and an unethical act. In P.
Christie & F. Geis (Eds.), Studies in Machiavellianism (pp. 53-75). New
York, NJ: Academic Press.

Fenigstein, A., Scheier, M. F., & Buss, A. H. (1975). Public and private
self-consciousness: Assessment and theory. Journal of Consulting and Clinical
Psychology, 43, 522-527. Hofer, E., Kohnken, G., Hanewinkel, R., & Bruhn,
Ch. (1992). Diagnostik und attribution von glaubwurdigkeit. Kiel: final
report to the Deutsche Forschungsgemeinschaft, KO ssz/4-2. Kalbfleisch,
P J. (1994). The language of detecting deceit. Journal of Language and
Social Psychology, 13, 469-496.

Kashy, D. A., & DePaulo, B. M. (1996). Who lies? Journal of Personality
and Social Psychol

ogy, 70, 1037-1052.

Knapp, M. L., Hart, R. P., & Dennis, H. S. (1974). An exploration of deception
as a communi

cation construct. Human Communication Research, 1, 15-29. Kohnken, G. (1989).
Behavioral correlates of statement credibility: Theories, paradigms and
results. In H. Wegener, F. Losel, & J. Haisch (Eds.), Criminal behavior
and the justice system: Psychological perspectives (pp. 271-289). New York:
Springer-Verlag. Kohnken, G. (1990). Glaubwurdigkeit: Untersuchungen zu
einem psychologischen konstrukt.

Munchen: Psychologie Verlags Union.

Kraut, R. E., & Poe, D. (1980). On the line: The deception judgments of
customs inspectors and laymen. Journal of Personality and Social Psychology,
36, 380-391. O'Hair, H. D., Cody, M. J., & McLaughlin, M. L. (1981). Prepared
lies, spontaneous lies, Machiavellianism, and nonverbal communication.
Human Communication Research, 7, 325-339.

Patterson, M. L. (1995). A parallel process model of nonverbal communication.
journal of Nonverbal Behavior, 19, 3-31.

Riggio, R. E. (1986). Assessment of basic social skills. Journal of Personality
and Social Psychology, 51, 649-660.

Riggio, R. E., & Friedman, H. S. (1983). Individual differences and cues
to deception. Journal of Personality and Social Psychology, 45, 899-915.


Riggio, R. E., Tucker, J., & Throckmorton, B. (1988). Social skills and
deception ability. Personality and Social Psychology Bulletin, 13, 568-577.


Siegman, A. W., & Reynolds, M. A. (1983). Self-monitoring and speech in
feigned and un

feigned lying. Journal of Personality and Social Psychology, 45, 115-128.
Snyder, M. (1974). Self-monitoring of expressive behavior. Journal of Personality
and Social Psychology, 30, 526-537.

Vrij, A. (1993). Credibility judgments of detectives: The impact of nonverbal
behavior, social skills, and physical characteristics. Journal of Social
Psychology, 133, 601-611. Vrij, A. (1995). Behavioral correlates of deception
in a simulated police interview. Journal of Psychology: Interdisciplinary
and Applied, 129, 15-29.

Vrij, A. (1996). Misverstanden tussen politie en verdachten in een gesimuleerd
politieverhoor. Nederlands Tijdschrift voor de Psychologie, 51, 137-146.


Vrij, A. (1997). Nonverbal communication and credibility. In A. Memon,
A. Vrij, & R. Bull. Legal psychology: Accuracy and perceived credibility
of suspects, victims and witnesses (work tide). New York: McGraw Hill,
in press.

Vrij, A., & Semin, G. R. (1996). Lie experts' beliefs about nonverbal indicators
of deception. Journal of Nonverbal Behavior, 20, 65-81.

Vrij, A., Semin, G. R., & Bull, R. (1996). Insight into behavior during
deception. Human








Honest or deceitful?:  A study of persons' mental models for judging veracity
Human Communication Research
Thousand Oaks
Dec 1997

--------------------------------------------------------------------------------

Authors:                  John S Seiter

Volume:                   24

Issue:                    2

Pagination:               216-259

ISSN:                     03603989

Subject Terms:            Social research
                          Cognition & reasoning
                          Models
                          Honesty


Abstract:

Seiter focused on understanding the detailed and dynamic mental models
that people develop for judging veracity.  He hypothesized that individual
differences in such mental models would predict participants' attributions
and confidence in making attributions.
Copyright Sage Publications, Inc. Dec 1997

Full Text:

This study focused on understanding the detailed and dynamic mental models
that people develop for judging veracity. It was hypothesized that individual
differences in such mental models, assessed by using Thagard's (1989) ECHO
computer simulation program, would predict participants' attributions and
confidence in making attributions. After watching videotapes of targets,
120 participants rated targets' veracity and their own confidence in making
attributions. Half of these participants also provided "on-line" information
that, in turn, was entered into ECHO. The two groups did not differ in
their judgments of veracity, but the on-line group was significantly more
confident. Results from ECHO and network analysis indicated not only that
participants' mental models for detecting deception are detailed, changing,
and idiosyncratic, varying in their structure and degree of coherence,
but also that a number of previously unidentified cognitive structures
are used for detecting deception. Results that confirm the hypotheses are
also presented.

How do we decide that we are being lied to? Although considerable research
has examined the degree to which factors such as suspiciousness (e.g.,
Burgoon, Buller, Ebesu, & Rockwell,1994; Toris & DePaulo, 1985), cognitive
biases (e.g., Zuckerman, Fischer, Osmun, Winkler, & Wolfson, 1987), and
familiarity with a deceptive source (e.g., Brandt, Miller, & Hocking, 1980;
Burgoon et al., 1994; Comadena, 1982; McCornack & Parks, 1985; O'Sullivan,
Ekman, & Friesen, 1988; Seiter & Wiseman, 1995) mediate the accuracy with
which individuals can detect deception, only a handful of studies have
dealt specifically with the ways in which people judge deception, regardless
of how accurate the judgments are (e.g., Bond, Kahler, & Paolicelli, 1985;
Gordon, Baxter, Rozelle, & Druckman,1987; Hale & Stiff,1990; Kraut,1978;
Kraut & Poe,1980; Maier & Janzen, 1967; Riggio & Friedman,1983; Riggio,
Tucker, & Widaman, 1987; Stiff, Hale, Garlick, & Rogan, 1990; Stiff & Miller,
1984, 1986; Streeter, Krauss, Geller, Olson, & Apple,1977; Zuckerman, Koestner,
& Driver, 1981). In most instances, such research has examined the verbal
and nonverbal cues that people rely on when making attributions about another
individual's veracity.

Although the results of these studies show some general consistencies (i.e.,
most indicate that people expect liars to exhibit less eye contact, more
postural shifts, slow speech, longer response latencies, and nervous behaviors),
they present a somewhat inconsistent picture of many of the behaviors that
people look for when detecting deception. For example, whereas Riggio and
Friedman (1983) found that liars were perceived as smiling less, Stiff
and Miller (1986) and Gordon et al. (1987) found that perceptions of deception
were positively associated with increased smiling. Moreover, whereas Stiff
and Miller (1986), Gordon et al. (1987), and Kraut and Poe (1980) reported
that increases in self-touching (i.e., selfadaptors and self-grooming)
were associated with attributions of deception, Riggio and Friedman (1983)
found just the opposite (i.e., self-touching was not associated with perceptions
of deception).

It is clear, then, that the results of studies examining cues that are
used to detect deception are not without their inconsistencies. Of course,
these inconsistencies could stem from any number of factors such as the
tendency of researchers to make faulty assumptions about factors (e.g.,
arousal) underlying the enactment of deception (see Fiedler & Walka, 1993),
the use of different methodologies, different subject pools, and so forth.
On the other hand, these inconsistencies may suggest that these cues are
not "absolute" indicators of deception. Rather, their interpretation may
depend on the particular configurational array within which they are embedded.


Whatever the case, it is clear that previous research tells us very little
about the process by which judgments of veracity are made. How do people
integrate a vast array of nonverbal and verbal behavioral information,
past knowledge, and inference to conclude that another individual is lying
or telling the truth? How might people make sense of cues that may be multiple
and possibly contradictory? How might their perceptions change as they
receive more information? For instance, in the courtroom, a juror might
initially perceive a defendant as an honest person of high integrity. However,
after considering the specificity of testimony against the defendant, a
juror might change his or her mind, concluding that the defendant is a
liar. Ultimately, however, the juror's mind could be changed again by a
number of factors such as the intensity of the defendant's denials. In
short, because of a number of different information sources, conclusions
about a person's veracity may change over time, resulting in different
judgments at different times.

This study argues that recent work by Thagard (1989) and Miller and Read
(1991), which provides a "connectionist," cognitive science conceptual
framework and methodology, offers a promising approach for addressing such
issues. Using this connectionist approach, the general purpose of this
study is to explore, more dynamically and holistically, individuals' rich
and detailed "mental models" or constructions of events that lead to inferences
about the deceptiveness or truthfulness of a communicator. As will be discussed
in more detail later, the work of Thagard (1989) and Miller and Read (1991)
provides a model of the ways in which individuals develop coherent mental
representations of others to explain or make attributions about others`
behavior. As such, it will be argued that such a model might be useful
for understanding and predicting the process and outcomes of deception
detection. Specifically, guided by this model, this study both qualitatively
and quantitatively examines (a) how individuals construct mental representations
of others' veracity and (b) the degree to which the coherence of an individual's
mental representation of others is related to outcomes such as the attributions
he or she makes, and his or her confidence in making such attributions.
The following section discusses prior literature to illustrate how a cognitive
framework can be useful for understanding such issues.

REVIEW OF LITERATURE

A Cognitive Framework for Understanding Deception Detection

Although little research has addressed the ways in which judgments of deception
are related to cognitions, two notable exceptions are represented in the
works of Fiedler and Walka (1993) and Buller and Walther (1989). Fiedler
and Walka argued that, in general, people lack the knowledge needed to
use nonverbal cues that discriminate honest from deceptive communication
so that, when detecting deception, they rely on general heuristics that
may not be reliable for making such discriminations. Similarly, relying
on the work of Rumelhart (1984), Buller and Walther (1989) argued that
two types of cognitive processes can mediate the act of deception detection:
"bottom-up" and "top-down" processing. Bottom-up processing occurs when
cognitive schemata (i.e., hierarchicall;, organized mental structures containing
knowledge about the social and physical world that guide perception, interpretation,
storage, and recall of information) at low levels of abstraction are activated
by a particular datum and then activate higher level schemata to account
for that datum (Buller & Walther, 1989). In contrast, top-down processing
occurs when a higher order schema is activated first, and, in turn, lower
level schemata are activated to search for more data (Buller & Walther,
1989; Rumelhart, 1984).

According to Buller and Walther (1989), previous research in the area of
deception detection can be seen as taking either a bottom-up or topdown
approach. For example, research that examines the specific nonverbal behaviors
that people use to detect deception has taken a bottom-up approach because
it assumes that specific behavioral changes (e.g., reduced gaze, longer
response latencies, and more speech errors) activate attributions of deceit
from those attempting to detect deceit. On the other hand, a considerable
amount of deception research has taken a top-down approach, including research
that has examined subjects' stereotypes about deceptive behavior (see above).
According to Buller and Walther (1989), these stereotypes amount to a "deception
script" containing the behavioral sequences that people expect deceivers
to encode. Communicators, they argue, enter some conversations with these
scripts "preactivated" and the deception scripts, in turn, guide attention
to subsequent behavior in a top-down fashion.'

Whether deception detection is based on a top-down or bottom-up process
(or both), it appears clear that we need to better understand the detailed
connections between the "behaviors" that perceivers observe and the conceptual
structures that perceivers activate in evaluating the deceptiveness of
a target individual. On the basis of the work on connectionist modeling
in social cognition and cognitive science (Kintsch, 1988; Rumelhart & McClelland,
1986), Thagard (1989) and Miller and Read (1991) provide a useful, detailed
framework for elaborating and exploring such connections among behavior
and inferences. The general purpose of Thagard (1989) and Miller and Read's
(1991) work is to provide an account of the process by which individuals
form coherent mental representations of others to explain, or make attributions
about, others' behavior. In developing these coherent mental models or
pictures of others, Miller and Read (1991) note that people rely on cognitive
knowledge structures (which include goals, plans, scripts, roles, and themes)
to make inferences from others' behavior and to integrate these inferences
into a meaningful picture.

According to Miller and Read (1991), this process of developing a coherent
model includes two steps. First, input (e.g., the observation of behavior,
communication, and events) leads to the activation of mental concepts that
are related to the input. For example, if we see a man talking to a woman
in a bar, it might activate a goal-based concept such as "he's trying to
pick someone up," or if we see someone with a certain skin color, it might
activate stereotypes about race (Miller & Read, 1991). In other words,
concepts at a more general and higher level of abstraction (he has the
goal of picking someone up) are activated because they can be used to explain
input at a lower level of abstraction (man in bar/talking to woman). Miller
and Read (1991) argued that there are three ways in which any two concepts
may be linked. First, when connected by a positive or excitatory link,
activation of one concept increases the activation of the other concept.
Second, when connected by a negative or inhibitory link, activation of
one concept decreases the activation of another. And third, there may be
no link between two concepts and therefore no activation or deactivation
between them.

From the discussion so far, it is clear that activated concepts may be
large in number and possibly contradictory. Indeed, Kintsch (1988) and
Read (1992) noted that concepts may initially be activated indiscriminantly,
with little attention being paid to their consistency with other activated
concepts. For instance, a man talking to a woman in a bar may be explained
by a goal-based concept such as "he's trying to pick someone up," but also
with the contradictory knowledge that the man is happily married to someone
else but friendly. Thus, many of the activated concepts may be likened
to hypotheses or inferences competing to explain input (see McClelland,
Rumelhart, & Hinton, 1986). Those concepts that receive the most activation
are used to make sense of the social world.

How, then, do some concepts receive more activation than others? According
to Miller and Read (1991), once an initial loose network of connected concepts
is developed, the second step of building a coherent mental model involves
spreading activation throughout the network of concepts. In this step,
activation is simultaneously dispersed through the links and concepts in
the mental network, resulting in differing levels of activation for each
of the concepts. The theoretical rules by which the coherence of a network
is determined were specified by Thagard (1989) in his theory of explanatory
coherence. The principles are as follows: First, the principle of symmetry
argues that if one concept (P) coheres with another (Q), then the latter
(Q) also coheres with the former (P). Thus, in terms of the cognitive links
discussed earlier, symmetry produces a symmetric excitatory link if two
concepts cohere and a symmetric inhibitory link if two concepts incohere.
Second, the principle of breadth refers to the fact that a concept (e.g.,
"he cheated on the exam") receives more activation than its competitors
(e.g., "he did not cheat") if it explains more facts (e.g., he received
a good grade, he did not know the subject material later, he was looking
at other students' exams) than its competitors. The third principle involves
being explained and argues that explanations receive more activation if
they are explained by further explanations (e.g., "he cheated" is explained
by "he did not study,' and "he did not study," and "he did not study" is
explained by "he is lazy") than if they are not. Fourth, the principle
of simplicity argues that a concept receives more activation when it requires
fewer assumptions (e.g., "he received a good grade, did not know the subject
material, and looked at other students' exams because he cheated" requires
fewer assumptions than "he received a good grade, did not know the subject
material, and looked at other students' exams because he was lucky, nervous,
and forgetful"). Fifth, the principle of analogy argues that a concept
is more coherent if its support is analogous to support for a similar concept
(e.g., "he cheated on other exams"). Sixth, the principle of data priority
argues that a concept describing results of an observation receives more
activation than a concept whose sole justification is what it explains.
Finally, the principle of contradiction argues that when two concepts contradict
(e.g., "he cheated" versus "he did not cheat"), they do not cohere.

In addition to specifying these principles, Thagard's (1989) theory assumes
that concepts are not activated in isolation. Instead, Thagard's theory
argues that the principles are applied at the same time in such a way that
activation is simultaneously dispersed through the links and concepts in
a network. As Read (1992) noted,

This is an example of a parallel constraint satisfaction process that is
a fundamental part of recent work on connectionist modeling or parallel
distributed processing. Such a process evaluates, in parallel and simultaneously,
the extent to which concepts in the network are consistent with and supported
by other concepts in the network. This is in contrast to a serial process
where each concept would be evaluated, one at a time. (p. 12)

Thus, the framework provided by Thagard (1989) and Miller and Read (1991)
assumes that at any given time, there may be multiple interpretations of
the same behaviors. The theory shows how strong concepts in a network win
out over weaker ones. In general, a concept is highly activated if it is
connected with a greater number of excitatory links, whereas a concept
has a low level of activation if it is connected with a greater number
of inhibitory links or is connected to relatively few concepts. Concepts
that are not supported by other concepts in a network fade away. On the
other hand, concepts that are supported are strengthened and, as assumed
by the theory, are used to make sense of social interactions and other
people. In other words, we use highly activated concepts to make attributions
about other people. The next section discusses the ways in which this framework
informs the study of deception. RESEARCH QUESTIONS AND HYPOTHESES It was
noted earlier that previous deception research has not addressed issues
such as the ways in which people construct meaningful attributions about
a person's veracity on the basis of multiple communication sources and
factors. It is argued here, however, that the work of Thagard (1989) and
Miller and Read (1991) provides a model to address such issues. Specifically,
in addition to allowing an integration of former research on deception,
an explanatory coherence model is capable of addressing the ways in which
people attempt to make sense of the information that they receive. It suggests
that the process of deception detection can be understood through a qualitative
examination of the mental models that individuals construct when observing
the potentially deceptive communication of others. For example, an examination
of these networks might promote an understanding of the ways in which people
integrate information (e.g., nonverbal and verbal cues, knowledge, inferences)
to reach a conclusion about someone's veracity, the ways in which people's
decisions about someone's veracity may change, and the ways in which people
come to terms with contradictory information such as the simultaneous observation
of behaviors that are indicative of both truthfulness and deceptiveness.
Thus, although it is suspected that the mental models that individuals
construct will be highly idiosyncratic, one research question (RQ) is proposed
as follows:

RQ1: What is the nature of the mental models that people construct when
attempting to reach conclusions about other individuals' veracity?

In addition, the theory of explanatory coherence makes testable assumptions
about the relationship between concepts that receive activation in a person's
cognitive network and attributions that are made as a result of such activation.
Specifically, the theory assumes that concepts that receive high levels
of activation (in accordance with the principles set forth by Thagard)
are used to make sense of the social world. With regard to this study,
when judging veracity, individuals must choose between two competing models:
Is the person lying or is the person telling the truth? If the theory of
explanatory coherence is tenable, the concepts receiving the most activation
(when Thagard's principles are applied) should correspond to attributions
that are made in "real life." In other words, coherence theory would predict
that an actual attribution of truthfulness or deceit would depend on which
of these concepts received the highest level of activation in a person's
cognitive network. Thus, the following hypothesis is tested in the present
study: H1: Subjects will perceive an individual as truthful when their
mental models of "truthfulness" (i.e., "the person I am observing is being
honest") receive the highest level of activation in their cognitive networks,
whereas subjects will perceive an individual as deceptive when their mental
models of "dishonesty" receive the highest level of activation.

Furthermore, it is argued that coherence theory provides a means of predicting
how confident individuals should be concerning the attributions they make
about deceptiveness. Specifically, concepts with high levels of activation
are those that are connected to, or supported by, a greater amount of input
or data. Thus, these concepts should be the ones that are related not only
to the final attribution that an individual makes but also to the confidence
with which the individual makes the attribution. Stated differently, the
second hypothesis of this study is as follows: H2: Subjects with a high
level of activation for a concept (e.g., "deceitfulness") should be more
confident in making an attribution (e.g., of deceitfulness) than a person
with a low level of activation for the concept.

METHOD

Overview

One-hundred and twenty participants watched one of four videotapes in which
a stimulus target communicated truthfully or deceptively. As the tapes
proceeded, half of the participants were asked to make inferences "on-line"
regarding the stimulus target's behaviors. After viewing segments of the
tape, these participants made inferences about the target's veracity and
rated how confident they were in making such inferences. These participants'
inferences were subsequently analyzed using Thagard's (1989) ECHO program,
a computer simulation that implements the seven theoretical principles
of explanatory coherence discussed earlier. This analysis provided measures
of participants' mental models and the degree to which deception or truthfulness
were highly activated concepts (i.e., inferences) within these models.
The basic design, therefore, was correlational, assessing the relationship
between the activation level of concepts in individuals' mental models
and the confidence with which attributions of deceptiveness were made.


The remaining 60 participants were also asked to judge the target's veracity
and their own confidence but did so after viewing the stimulus tapes all
the way through. Thus, these participants acted as a control group, helping
to determine whether stopping the tape affected confidence or veracity
ratings.

Development of the Stimulus Tapes

Stimulus Targets

The videotaped stimulus targets were one White man, age 29, and one White
woman, age 31. Both were known personally to the researcher and both volunteered
to participate in the study

Lists of Interview Questions

An examination of previous research (e.g., Cupach & Metts, 1986; Harvey,
Agostinelli, & Weber, 1989; Harvey, Wells, & Alvarez, 1987) indicated that
people's accounts for the dissolution of their close relationships would
be a realistic and rich source of communication messages for this study's
stimulus materials. An interview list that consisted of six preliminary
and six probing questions concerning previous relationships and why they
failed was developed and is shown in Appendix A.

Room and Videotaping Equipment

Two chairs, placed approximately 6 ft (1.8 m) apart, faced one another
in the middle of a large living room. A video camera that filmed in color
was set up behind one chair so that it focused on the top of the stimulus
targets' heads and just below their knees when they were seated on the
other chair.

Videotaping Procedures

Each of the stimulus targets participated in one "honest interview" and
one "deceptive interview." Before the first, targets were told, We will
be asking you several questions about a past close relationship you were
in that ended. Please answer all of the questions to the best of your ability
and be completely truthful.

Before the deceptive interview, the targets were told, We will be asking
you several questions about a past close relationship that ended but want
you to be completely deceptive. That is, make up a person and lie about
your relationship with him/her. Please answer all of the questions to the
best of your ability.

After the targets were given the above instructions, they were allowed
5 minutes to prepare for the interview and then led to the chair. Next,
a White male interviewer asked the target if he or she had any questions,
started the video equipment, took a seat in the chair facing the stimulus
target, and proceeded to ask the target the interview questions. The list
of interview questions was meant to be a guide for the interview, providing
topically similar tapes across the targets and conditions. When possible,
the interviewer followed up each question, probing for detailed information
about reasons for the target's answers, thoughts, and feelings. Following
the first interview, the target was given 10 minutes to relax and then
prepared for the second interview. When each interview was completed, the
targets were asked if they were indeed truthful in the "honest interview"
and deceptive in the "deceptive interview." Both targets indicated that
they had been honest and deceptive when instructed to do so.

Viewing of Videotapes

Participants

The participants in this study were 120 students (48 males, 72 females;
mean age = 21.9) enrolled in interpersonal and organizational communication
classes at a large southwestern university. All received course credit
for volunteering to participate in the study.

Apparatus

The four videotaped stimulus target interviews were viewed by the participants
on a large color television set. The approximate times of the taped interviews
were as follows: honest male student, 6 minutes and 45 seconds; deceptive
male student, 10 minutes and 15 seconds; honest female student, 7 minutes;
deceptive female student, 9 minutes.

Procedures

Half of the participants (25 male students, 35 female students) watched
one of the four stimulus target videotapes. Upon arriving at the laboratory
one at a time, each participant was randomly assigned to view one of the
tapes. When all four tapes had been viewed, random assignment started again.
All these participants were told that they would be watching a videotape
of someone who was either lying or telling the truth. Moreover, they were
told that the goal of the experiment was to determine what they were thinking
about the person on the tape while they were watching the tape. Participants
were told that they would be asked to distinguish between the inferences
they made and the facts or data (if any) on which the inferences were based.


Following the preliminary instructions, the researcher (a White man) played
one videotaped interview for each participant, stopping the videotape at
predetermined times to ask the participants questions about their perceptions
of the videotaped stimulus targets. Stopping times were predetermined on
the basis of the questions asked during the making of the stimulus tapes.
Specifically, because six general interview questions were asked of stimulus
targets, the tapes were stopped six times (i.e., once after each question
was answered) during the course of the interview.

Once the tapes were stopped, the participants were asked questions regarding
their impressions of the stimulus targets. After stopping the tape each
time, participants were asked one standard question: "What is happening
here?" Typically, participants responded by making inferences about the
targets' veracity (e.g., "I think she is lying"). Because each participant's
mental model developed differently, further questions were not standardized
or asked in a standard order but were based on participants' responses.
All follow-up questions were similar, however, in that they were designed
to probe for other inferences or for data that participants had observed
to support inferences. Examples of such questions included the following:
"Why do you think this is happening?"; "What makes you think that?"; "How
is this person acting?"; and "Were you thinking anything else?" To ensure
that participants were not led to observe some behaviors and not others,
and make some inferences and not others, the interviewer limited follow-up
questions to those that explored what concepts (data and inferences) were
being activated by participants and to clarifications regarding how the
concepts were related. Moreover, as each question was answered, the interviewer
attempted to give verbally and nonverbally neutral responses while mapping
participants' mental models.

As noted earlier, Thagard's (1989) theoretical principles are based on
relationships between concepts (i.e., data and inferences) in a cognitive
network. Thus, to analyze each participant's mental model according to
these principles, it was necessary to record the ways in which each participant
perceived concepts in their mental models to be related to one another.
With this in mind, a mapping procedure was developed using a large poster
board with numerous rows and columns. Each time the participant made an
inference and identified it as such, the inference was written along the
far left side of a row and the top of a column. Furthermore, the data that
participants believed were explained by the inference were written along
the far left side of a row and at the top of a column. This resulted in
a matrix with all of the concepts (i.e., data and inferences) generated
by participants appearing in both the rows and the columns.

Information in the columns indicated concepts that participants used to
explain other concepts, and information in the rows indicated concepts
that were explained by other concepts. When the participant made an inference
(e.g., the stimulus target is lying) and believed that a particular datum
was explained by that inference (e.g., the stimulus target seems anxious),
a mark indicating this relationship was placed where the row and column
for the inference and datum met. When the participant indicated two contradictory
inferences (e.g., the stimulus target is lying and is telling the truth),
this was also indicated where the row and column for the inferences met.
Participants also indicated when they believed inferences or data were
contradictory (an example of this procedure is shown in Appendix B). Thus,
determinations regarding inferences, data, and relationships between concepts
were made by participants.

In addition to mapping participants' mental models, it was also necessary
to assess whether they believed the stimulus target on the tape was behaving
honestly or deceptively, and how confident they believed their attributions
were. Participants' perceptions of stimulus subjects' veracity were assessed
throughout the interview (each time the tape was stopped) by asking them
(a) to indicate on a dichotomous checklist whether they thought the subjects
were lying or telling the truth; and (b) to indicate on a 9-point, Likert-type
instrument whether they thought the target was definitely deceptire or
definitely truthful. Participants' confidence in making the dichotomous
attribution was assessed with a 9-point, Likert-type rating scale (1 =
extremely confident and 9 = not at all confident about their attribution).2
The remaining 60 participants (23 male students, 37 female students) also
watched the videotapes but did so in groups of 15 without being interviewed
and without having the tape stopped. After being randomly assigned to one
tape and being given the same instructions as the interviewed participants,
each group watched a tape and rated the stimulus target using the same
scales described above. After rating the taped interviews, all participants
were thanked and debriefed.

Data Analysis

The data collected from the interviewed participants were analyzed using
a computer program developed by Thagard (1989). The program, called ECHO,
is a simulation that simultaneously sets in motion the seven principles
that Thagard (1989) argued could be used to assess the coherence of a network
(see above). How are the principles implemented by ECHO? First, as the
principle of symmetry suggests, the program establishes an excitatory link
between concepts that cohere and an inhibitory link between concepts that
"incohere." These links are symmetric, so the activation weight from Concept
1 to Concept 2 is the same as the activation weight from Concept 2 to Concept
1. Second, the principle of breadth is implemented in ECHO by summing the
activation of all the concepts that are explained. Third, the principle
of being explained is implemented by the program because concepts send
activations to concepts they explain. Fourth, the principle of simplicity
is implemented when activation sent by a concept is divided by the number
of concepts that do the explaining. Thus, a single explanatory concept
will receive more activation than will multiple explanatory concepts. Fifth,
the principle of data priority is implemented by linking concepts that
represent observed data to a special evidence unit that always has an activation
of 1.0 to indicate that evidence has a special degree of strength and is
more immune to deactivation than are other concepts. Sixth, the principle
of analogy is implemented when ECHO produces excitatory links between the
explanatory concepts contained in the analogy. Seventh, the principle of
contradiction is implemented by establishing an inhibitory link between
two contradictory concepts so that concepts become deactivated when they
compete with more strongly activated concepts.

The use of the program for this study involved specifying which concepts
participants used as evidence or data, which they used as inferences, and
which inferences they used to explain which data. The matrix of data presented
in Appendix B has been translated into List Processing (LISP) structures
in Appendix C to illustrate how the concepts (i.e., data and inferences)
are entered into the program. Appendix D diagrams the matrix of data to
illustrate the process more concretely. First, inferences or data in the
network (i.e., the stimulus target is truthful, the stimulus target is
deceptive, the stimulus target is anxious, the stimulus target's answers
are specific, the stimulus target's answers are vague, the stimulus target
stuttered) are coded as El, E2, E3, E4, E5, and E6, respectively. Second,
it is necessary to specify which inferences explain which data/inferences.
For example, E2 (the stimulus target is deceptive) and E3 (the stimulus
target is anxious) explain E5 (the stimulus target's answers are vague),
and El (the stimulus target is truthful) explains E4 (the stimulus target's
answers are specific). In other words, the fact that the stimulus target
is lying and is nervous explains why the target's answers are vague, and
the fact that the stimulus target is truthful explains why the target's
answers are specific. Next, it is necessary to specify contradictions in
the network. For example, El (stimulus target is truthful) contradicts
E2 (stimulus target is deceptive). Finally, concepts that are considered
data must be specified, for they are linked to a special evidence unit
that is given an activation of 1.0 to indicate that evidence has a higher
degree of credibility. In this example, concepts E4, E5, and E6 are considered
data units.

Following the specification of links in each participant's network, the
ECHO program was run. The program works by applying a parallel constraint
satisfaction process in which activation is simultaneously spread throughout
the network to converge on a pattern of activation that best fits the constraints
imposed by the concepts and links in a network. The program runs in cycles,
each corresponding to an action in the simulated social interaction. As
new concepts are added to the network, the strengths between links are
modified in parallel.4 The modification of concepts continues until all
activation asymptotes. Ultimately, activation weights (ranging from 1 for
excitatory links to -1 for inhibitory links) are assigned automatically
to the concepts in a network. As the program runs, some concepts become
more activated than others. A concept that receives a high level of activation
will cause concepts that compete with it to become deactivated. For example,
in the diagram in Appendix D, the "truthful" concept might receive a lower
level of activation than the "deceptive" concept. Those concepts that receive
the most activation can be interpreted as being the concepts that a person's
model coheres around (Miller & Read, 1991). Thus, with regard to this study,
if the concept "he is lying" receives the most activation, it can be assumed
to be more acceptable than the concept "he is telling the truth." This
type of assessment was made for each of the participants in the present
study.

Ultimately, therefore, the use of Thagard's (1989) program yielded scores
indicative of the level of activation for the concepts "the target is lying"
and "the target is telling the truth." These scores were used in subsequent
analysis in two ways. First, the most activated concept was compared with
participants' final attributions (i.e., "the target is lying" or "the target
is telling the truth") to determine whether this type of analysis is useful
for predicting the actual attributions that people render." Second, by
subtracting each participant's activation level for truthfulness from their
activation level for deceptiveness, a single score resulted that was correlated
with participants' confidence scores (i.e., how confident they were when
making an attribution regarding targets' veracity). The next section presents
the results of these analyses.

RESULTS

The Nature of Mental Models

Overview

Coherence analysis provides a useful method for understanding the mental
models that individuals construct and use to evaluate others' veracity.
It not only enables us to examine the richness and detail of these networks
but also demonstrates the ways in which concepts within these networks
are integrated and become more or less salient for the communicator. Specifically,
coherence analysis indicates quantitatively the degree to which concepts
such as "the communicator is lying" or "the communicator is telling the
truth" become either activated or deactivated in a network. In addition,
such analysis indicates why some concepts become more activated than others.
As will be discussed shortly, the content (i.e., specific observed behaviors,
knowledge and inferences) in participants' mental models varied considerably
from participant to participant. Thus, at one level of analysis, participants'
mental models for judging veracity were highly idiosyncratic. However,
at a more general level, most of Thagard's (1989) theoretical principles
of explanatory coherence were operating in all of the participants' mental
models (see below). Two of these participants' mental models were selected
for discussion in this section because they represent most or all of Thagard's
principles that were used by the rest of the sample.

Figure 1, based on output from Thagard's (1989) ECHO program, depicts the
mental model of a participant who thought that "R," the female stimulus
target, was lying about a previous relationship she had with a man named
Jack. As can be seen in the figure, concepts used in the network include
information about observed verbal and nonverbal behavior, knowledge, and
inference. How did the participant integrate and use such information to
reach a conclusion about R's veracity? Moreover, why was it the case that
the participant's concept for deceptiveness (E2 "R is lying") received
a higher level of activation (.91) than did the participant's concept for
truthfulness (El "R is telling the truth") (-.82)? First, recall that in
the theory of explanatory coherence (and the ECHO program), concepts that
explain more facts receive more activation. In this case, the concept for
deceptiveness explains more data than the concept for truthfulness does;
whereas "R is lying" directly accounts for eight pieces of data (indicated
in the figure by single-headed arrows from E2 to E3, E6, E7, E9, ElO, E14,
E15, and E16), "R is telling the truth" accounts for only two (E20 and
E26). Thus, "R is lying" has more explanatory breadth.

Second, Thagard's theory specifies that concepts receive more activation
if they are explained by higher order concepts. In Figure 1, it can be
seen that although the statement "R is lying" is explained by E17 (which,
in turn, is explained by the strong subnetwork of E18 and E19 and E21 through
E24) and E12 and E13, the statement "R is telling the truth" gets activation
from no higher level concepts. Thus, "R is lying" is a stronger concept
in the network than "R is telling the truth."

Finally, the theory specifies that concepts that contradict more or stronger
concepts become deactivated. Thus, in this network, because the statement
"R is telling the truth" contradicts the stronger activated "R is lying"
(as indicated in Figure 1 by a double-headed arrow), it receives less activation.
In short, "R is telling the truth" receives little activation because it
explains less, is explained by less, and contradicts the highly activated
concept "R is lying."

Figure 2 depicts a mental model in which the opposite is true: The participant
believed that "S," the male stimulus target, was telling the truth about
having a relationship with a woman named Barb. As can be seen in the figure,
although the content of the concepts is much different from that in the
previous network, Thagard's (1989) theoretical principles are applied in
the same way, illustrating how multiple sources of information can be integrated
and used in parallel to judge the stimulus target's veracity. In this case,
concept El, "S is telling the truth," directly explains 15 facts (E5, E6,
E9, ElO, Ell, E12, E16, E18, E21, E27, E28, E29, E31, E32, and E33) and
is explained by one (E3). On the other hand, E2, "S is lying," explains
only one fact (E23) and is jointly explained by E14 and E15 (indicated
in the figure by arrows that join one another). In addition, the concept
"S is lying" contradicts "S is telling the truth" and "S is shy" (E3),
both strongly activated concepts. Finally, "S is telling the truth" receives
further activation from an analogy (indicated in the figure with winding
lines).6 Specifically, the participant draws an analogy between her own
behavior when she lies and the behavior of the target whom she perceives
as deceptive. Because the participant sounds genuine (E20) when she tells
the truth (El9), she perceives S's genuine behavior (E21) as a sign of
his honesty (El).

[IMAGE CHART] Captioned as: Figure 1:

[IMAGE CHART] Captioned as: Figure 2

The Construction of Mental Models: Content and Structure

The framework advanced in this study assumes that there are causal connections
between the multiple concepts that play a role in the process of deception
detection. In other words, in any given mental model, concepts are activated
because they explain the occurrence of other concepts. Whereas the previous
section illustrated the ways in which perceptions of veracity may be affected
by multiple concepts at the same time, this section examines specific substructures
and content within participants' mental models. Such an analysis is useful
for several reasons. First, it illustrates whether there are patterns in
substructure use. That is, when detecting deception, are some substructures
more common than others? Second, such an examination demonstrates how specific
concepts (e.g., "she is telling the truth") are causally related to others.
This is important to understand the nature of deception detection from
a coherence framework because the location of a concept in the overall
network affects the degree to which that concept is activated. Finally,
such an analysis is useful because it helps us understand how people use
specific types of information to reach conclusions about others' veracity.
Indeed, in this study, a total of 1,514 concepts were generated (an average
of 25.23 per participant), but not all were used in the same way

For these reasons, the following sections explicate more specifically the
ways in which the content (i.e., general cues and inferences) of participants'
mental models is structurally related to the concepts of truthfulness and
deceptiveness. Seven different substructures are identified.

Substructure 1: signs of deception. One common substructure found in participants'
mental models was one in which concepts of truthfulness or deceptiveness
were used to directly or indirectly explain a behavior or inference. Several
examples are shown in Figure 3. Such substructures are labeled "signs of
deception" because, as seen in each example, the concept "S is lying" explains
some behavior or inference.

[IMAGE CHART] Captioned as: Figure 3:

Recall that the theory of explanatory coherence assumes that concepts that
have more explanatory breadth receive more activation than those that do
not. Thus, concepts explaining a large number of "signs" should win out
over those that do not. This is significant when one considers that "signs"
were the most common substructure, used 671 times by participants. Not
only did all of the participants use it to judge veracity; 53% of all the
concepts generated by participants were found in this type of substructure.


It is also informative to examine the general nature of concepts that were
found in this substructure, for such an examination indicates what type
of information people perceive as indicative of veracity. Of the concepts
that were contained in this type of substructure, 34% (277) were nonverbal
in nature (e.g., "S's legs were shaking"); 24% (196) described verbal behaviors
(e.g., "R said she went to the movies"); and 41% (329) were inferences
that were made about the targets or the targets' stories (e.g., "S is uncomfortable").


Substructure 2: constituents of deception. In addition to using signs,
39 out of the 60 participants made inferences about veracity by identifying
pairs of concepts that, in their minds, constituted deceptive/truthful
behavior. Examples of such "constituents of deception" are also illustrated
in Figure 3. In each example, two concepts are paired to explain why the
target is lying or telling the truth. In the theory of explanatory coherence,
constituents receive activation, but not as much activation as a concept
that explains data by itself.

Interestingly, the nature of the concepts in this substructure differed
from those in the previously discussed substructure in several ways. First,
although nonverbal behaviors were frequently viewed as signs of veracity,
no concepts about targets' nonverbal behaviors were seen as "constituents
of deception/honesty." Second, when participants used targets' verbal behaviors
as signs, concepts were either specific (e.g., "R said she took Jack to
the movies") or general (e.g., "R's answers lacked detail") in nature.
However, when viewed as constituents of deception, concepts about verbal
behaviors were of a specific nature in every case but one (i.e., 98 out
of 99 cases). Finally, when participants used inferences (e.g., "S is nervous")
as signs, all inferences were about the targets or the targets' stories
(e.g., "S is nervous" or "R's story is not believable"). On the other hand,
when inferences were used as constituents, they also included assumptions
about the nature of reality (e.g., "no Asians are named Barb"). As will
be discussed later, such assumptions illustrate the importance of knowledge
structures in the process of deception detection.

An examination of the ways in which these concepts are paired indicates
that it is what targets say specifically that is used to make inferences
about their veracity. To be sure, of the 80 pairs of concepts that were
seen as constituting deception, 76 involved at least one concept concerning
something specific that the targets said. In 22 of those cases, the specific
verbal statement was paired with another contradictory verbal statement.
In 23 of those cases, the specific statement was paired with a contradictory
assumption about the target (e.g., "R said she dated Jack, a punker" but
"R is not the type of person to date a punk rocker"). And in 31 of those
cases, the specific statement was paired with a belief about the nature
of reality (e.g., "R said Jack, a punker, was a good dancer" but "Punk
rockers can't dance"). In short, then, although people believe that deceptive
acts influence both what is said and how it is said, they look for deception
itself by considering what is said and whether it contradicts itself or
something that is believed.

Substructure 3: top-down processing. As discussed earlier, Buller and Walther
(1989) argued that top-down deception detection occurs when a higher order
schema is activated first and, in turn, lower level schemata are activated
to search for more data. Although it could be argued that all cognitive
processing in this study was top-down (because participants were told that
this was a deception study), the clearest evidence for top-down processing
can be seen in this third substructure, where a single concept, often demonstrating
the operation of some stereotype or bias, was used to explain why the target
was either lying or telling the truth (e.g., "R is telling the truth" was
explained by "R is a female"). Such concepts affect deception detection
because they presumably influence the way in which subsequent data are
processed. Moreover, such concepts are significant because, according to
Thagard's (1989) principles, truthful and deceptive concepts receive activation
when they are explained by higher order concepts (see above).

With 29 participants using it, this substructure was also common. Regarding
content, 39 of these substructures included inferences about the targets
(e.g., "R is shy") to explain veracity, 8 used something the targets said
(e.g., "R said she was hurt in the relationship") to explain veracity,
and 3 used some other form of data ("This is a study on deception") to
explain veracity.

Substructure 4: analogies. Thagard (1989) argued that concepts receive
activation when they are supported by an analogy, the fourth substructure
(also illustrated in Figure 3). Even so, with only 15 participants using
them (16 analogies were generated in all), analogies were the least common
substructure in this study It is interesting to note that participants
used themselves as a reference point (e.g., "When I lie, I act like the
target") in 14 of the analogies and used people in general (e.g., "When
people lie, they act like the target is acting") in only 2 of the analogies.


Substructure 5: cohypotheses. Substructures 5 and 2 are similar because,
in them, two concepts are used together to explain a third (see Figure
3). The difference is that in Substructure 5, one of the explaining hypotheses
is the truthful or deceptive concept (i.e., "S is lying" or "R is telling
the truth"). The implication is that in this substructure, the deceptive/truthful
concepts receive less activation than if they explain another concept by
themselves (see Thagard's principles discussed earlier).

With 23 participants using it a total of 33 times, this substructure was
fairly common. In every case, the truthfulness/deceptiveness concepts were
paired with inferences either about the nature of reality (e.g., "Lying
causes you to slip up," which occurred in 9 cases) or about the target
or the target's story (e.g., "R is projecting," which occurred in 24 cases).


Substructure 6: explaining away. According to the theory of explanatory
coherence, concepts become deactivated when they contradict highly activated
concepts. Thus, concepts with many contradictions will become deactivated
and will not be used to judge veracity. Such contradictions are perhaps
the most intriguing substructure, for they indicate the ways in which individuals
"explain away" discrepant information. For instance, one participant believed
that the target was telling the truth but was bothered that the target
also seemed nervous-a behavior that, for the participant, was indicative
of deception. To overcome the discrepancy, the participant generated an
alternative (contradictory) explanation for the nervous behavior, "the
target is being videotaped." Thus, the participant was still able to see
the target as truthful while, at the same time, account for the discrepant
behavior. Forty participants generated 74 such alternative concepts.

Substructure 7: isolates. The last type of substructure included concepts
that were not structurally related to the truthful/deceptive concepts in
participants' mental models. For example, a participant might have said
"R was stupid" while not seeing R's stupidity as related to R's veracity.
Such concepts, because not related to deception, did not affect the activation
of truthful and deceptive concepts in participants' mental models. Twenty-six
of the participants generated 66 of such isolated concepts. This and information
regarding the other substructures is summarized in Table 1.

Change Over Time

The process of person perception is a dynamic one. As individuals are exposed
to more information, their perceptions change to assimilate, to accommodate,
or to explain away that information. The findings from this study indicate
that the process of detecting deception is no exception: The mental models
of participants in this study grew and changed over time. After making
an initial attribution (i.e., the stimulus target is lying or the stimulus
target is telling the truth), participants had opportunities to change
their attributions five times. Results indicated that 31 participants changed
their mind at least once during the interview. Fifteen of those changed
their mind twice, and 3 changed their mind three times. Changes from "lie"
to "truth" or vice versa occurred at about the same frequency.

[IMAGE TABLE] Captioned as: TABLE 1

To illustrate such change, Figure 4 diagrams the associative network of
a participant when she was first asked to make an attribution and when
she was last asked to make an attribution. The dark rectangles indicate
which concepts were generated at Time 1, and the lighter rectangles indicate
those that were generated at Time 6. Activation levels for both times are
provided as well.

As can be seen, the participant initially believed that the target was
being deceptive but eventually changed her mind. What happened to cause
this change? At Time 1, concept E1, "S is telling the truth," was less
activated than E2, "S is lying," for several reasons. First, it explained
less data (i.e., E2 explained E3 and E4, whereas El explained nothing);
second, unlike E2, it was not explained by higher order concepts (E6 and
E7; and finally, it contradicted E2, a highly activated concept. By Time
6, however, several changes occurred to reverse these activation levels.
First, E1 now explains more concepts (E8 through E14) than E2, which explains
no more than it did before. Second, E2 contradicts the now heavily activated
E1. And finally, to account for E3 ("S is fidgety") and E4 ("S can't remember
things that happened"), which are consistent with the belief that "S is
lying" and inconsistent with the now highly activated "S is telling the
truth," two alternative explanations, E15 and E16, are added to the network.
Now, the once deceptive behavior is explained away, resulting in less activation
for E2. In short, the participant comes to believe that the target is no
longer behaving deceptively.

The Idiosyncratic Nature of Deception Detection The previous discussion
of substructures illustrates that at one level, it

may be possible to make generalizations about the process of judging veracity.
Indeed, when concepts are organized into substructures, it is clear that
at least some of the substructures (e.g., signs of deception) were used
by nearly all of the participants. On the other hand, an analysis that
makes its focus either narrower or broader indicates that, to a large extent,
deception detection is idiosyncratic.

[IMAGE CHART] Captioned as: Figure 4:

First, when one narrows the analysis to single concepts within a mental
model, it becomes clear that the information different participants used
to detect deception was highly variable and idiosyncratic. This was most
apparent in the different knowledge structures that participants activated.
For example, one participant noted that "Asian women want father figures"
and assumed that the stimulus target was lying because "he said his Asian
girlfriend wanted a sensitive guy" Another believed that "one hour was
not a long time to wait for someone" and assumed that the stimulus target
was lying because "she was upset about waiting an hour for her boyfriend."


Idiosyncrasies were also apparent in other substructures. Examples include
participants who believed targets were lying or truthful because targets
were "shy" or "nice-looking" (top-down processing) and participants who
disagreed about whether smiling was a sign of deception or truthfulness.s


Second, when one broadens the analysis to an examination of entire mental
models, it becomes clear that the participants in this study showed considerable
variation in the coherence (i.e., how well concepts "fit together" to form
a whole) and structure of the models they used to judge veracity. For example,
Figures 5 and 6 illustrate the mental models of two participants, one more
coherent than the other. As can be seen, the network in Figure 5 is less
fragmented (i.e., the concepts are more interconnected) than the network
in Figure 6. The network in Figure 5 coheres around the concept "S is lying"
(E2), which is not only supported by two strong clusters of reasons (E26
through E29 and E6, E10, and Ell), but also supports a large number of
lower order concepts. On the other hand, the network in Figure 6 coheres
around the concept El, "R is telling the truth," but El is not as central
to its network as E2 was in the previous example. Instead, in Figure 6,
there are several fragmented inferences, some that contradict (i.e., El
and E2) and others (i.e., E3 and E13) that support data but are relatively
isolated from the rest of the network.

A comparison of Figures 2 and 4 illustrates the considerable degree to
which the organization and use of cognitive substructures vary from participant
to participant. Such differences are even more dramatic when one considers
that both mental models were constructed while watching the same video
(i.e., the male stimulus target was lying), and that both participants
ultimately reached the same conclusion about the target's veracity (i.e.,
the target was telling the truth). As can be seen, compared with the model
in Figure 4, the one in Figure 2 includes more signs, more isolates, and
an analogy. Although both models include constituents of deception and
contradictions, the content of these substructures varies considerably.


[IMAGE CHART] Captioned as: Figure 5:

[IMAGE CHART] Captioned as: Figure 6:

Results of Tested Hypotheses

Preliminary Tests

Before testing the hypotheses of this study, statistics were run to determine
whether the dependent variables (attributions and confidence) were affected
by the nature of the stimulus tapes (i.e., male or female stimulus target
lying or telling the truth) and the condition within which the tape was
viewed (i.e., with or without stopping the tape to assess participants'
attributions).9

After Bartlett's test of sphericity and tests for multicollinearity among
the dependent measures indicated it was warranted, a multivariate analysis
of variance (MANOVA) was computed to determine whether viewing condition,
target sex, or target veracity affected participants' continuous attributions
of targets' veracity (i.e., the 9-point Likert-type scale where 1 = definitely
deceptive and 9 = definitely truthful) or the level of confidence with
which such attributions were made (1 = extremely confident, 9 = not at
all confident). With the use of Wilks's criterion, the combined dependent
variables were significantly affected by the viewing condition, lamda =
..89, F(2, 111) = 6.90, p < .001, eta ^sup2^ = .11; and the sex of the target,
lamda = .94, F(2, 111) = 3.51, p < .05, eta ^sup2^ = .06; but not by the
targets' actual veracity, lamda = .99, F(2, 111) = .33, p > .05; or by
interactions between target sex, target veracity, and viewing condition,
lamda = .99, F(2, 111) = .24, p > .05; target sex and viewing condition,
lamda = .97, F (2,111) = 1.49, p > .05; target sex and target veracity,
lamda = .97, F (2,111) = 1.49, p > .05; and target veracity and viewing
condition, lamda = .99, F (2,111) = .39, p > .05.

Follow-up univariate tests were conducted to examine more closely the effects
of the viewing condition and target sex on the dependent variables. Results
indicated that although participants in the control group and the interviewed
group did not differ significantly in their attributions of the targets,
F(1, 112) = 2.52, p > .05, the control group was significantly less confident
in making its attributions, F(1, 112) = 13.27, p < .01, eta ^sup2^ = .11).
Moreover, although participants were not more confident when judging the
male target than when judging the female target, F(1, 112) = .74, p > .05,
the female target was perceived as significantly more deceptive than was
the male target, F(1, 112)] = 6.99, p < .01, eta = .06. In short, these
results indicate that caution should be taken when making generalizations
about confidence across viewing conditions and about attributions across
stimulus targets.

Hypothesis 1: Predicting Attributions of Deceptiveness

As discussed previously, the most activated concept in an individual's
mental model should predict the attribution that he or she makes. To test
this hypothesis, a phi coefficient was computed to examine the relationship
between the attribution a participant made (i.e., "the target is lying"
or "the target is telling the truth") and the most activated concept in
that participant's simulated mental model (i.e., "the target is lying"
or "the target is telling the truth"). Results indicated a significant
relationship (phi = .86, chi ^sup2^ = 40.27, df = 1, p < .001), thereby
supporting the first hypothesis of this study

Hypothesis 2: Predicting Confidence

For each concept in a participant's mental model, the ECHO simulation produced
activation levels. An independent variable was created by computing the
difference between the activation levels of the deceptive and truthful
concepts in each mental model (i.e., the activation level of "the target
is telling the truth" was subtracted from the activation level of "the
target is lying"). It was hypothesized that the size of this difference
score should predict the confidence with which an attribution is made.
That is, smaller differences should lead to less confidence and larger
differences should predict more confidence. To test this hypothesis, regression
analysis was used. Results, however, did not support the hypothesis, R
= .01, F(1, 58) = .85, p > .05.

To investigate possible reasons for this lack of support, further analyses
were conducted. A significant negative correlation was found to exist between
the activation level of the participants' "truthful" and "deceptive" concepts
(r = -.88, p < .001). This resulted in difference scores with little variability.
Moreover, as indicated by the high negative correlation, any variation
in these concepts' difference scores could be primarily due to chance.
Thus, variability due to error, which presumably would not be systematically
related to the confidence variable, may have been the actual reason for
the lack of support for the hypothesis.

Because of this statistical problem, it was decided that Hypothesis 2 should
be examined by using the activation score of a single concept instead of
the difference between the activation levels of two concepts to predict
attributional confidence. Specifically, it was suspected that the activation
level of the most positively activated concept (i.e., either "the target
is telling the truth" or "the target is lying") in a participant's mental
model would predict that person's degree of attributional confidence (i.e.,
higher activation should lead to higher confidence, whereas lower activation
should lead to lower confidence).lo To test this, regression analysis was
used. Results indicated a significant relationship, R = .39, F(1, 58) =
10.58, p < .002, R ^sup2^ = .15, thereby supporting this study's second
hypothesis.

DISCUSSION

This study focused on understanding the detailed and dynamic mental models
that people develop for the purpose of judging other individuals' veracity.
To that extent, this study has reached beyond previous research to provide
new insights into the process of deception detection. By building on the
work of Thagard (1989) and Miller and Read (1991), and by implementing
a novel design and analysis that mapped and measured participants' mental
models, this study not only revealed that a "connectionist," cognitive
science framework provides a unique way of understanding how people integrate
large amounts of verbal and nonverbal information, inference, and knowledge
for the purpose of detecting deception; it also showed that coherence analysis
is useful for predicting outcomes of the detection process. Indeed, both
of this study's hypotheses were confirmed.

First, tests of Hypothesis 1 indicated that the activation of concepts
in a person's simulated mental model predicted the type of attribution
that a person made in "real life." In other words, people decided that
someone was lying or telling the truth depending on which of these two
concepts was the most activated in their network. Support for this hypothesis
is significant because it suggests that when considered together, Thagard's
(1989) seven theoretical principles of explanatory coherence provide a
tenable framework for describing the process by which attributions of veracity
are made. Specifically, because activations were determined through an
implementation of these theoretical principles, and because activations
corresponded to the attributions made by participants, it is apparent that
the theory of explanatory coherence is useful for both describing and predicting
the process and outcomes of deception detection. In short, this study suggests
that deception detection can be conceptualized as a process by which multiple
concepts competing within the same mental model become more or less activated
according to a small set of theoretical principles of coherence. Concepts
receiving the most activation are used to attribute honesty or dishonesty.


Second, tests of Hypothesis 2 indicated that the activation of concepts
in a network (i.e., the most activated concept in a network) predicted
the confidence an individual had when making an attribution. This, of course,
makes sense because concepts that receive higher levels of activation are
connected to, or supported by, a greater amount of input or data, which
presumably would lead an individual to be more confident about his or her
attribution. As with Hypothesis 1, support for this hypothesis illustrates
that coherence theory and analysis are useful for predicting the outcomes
of deception.

In addition to supporting these hypotheses, this study promotes theoretical
understanding of cognition's role in the process of deception detection
in other ways. First, this study shows how information regarding deception
is cognitively represented and, by doing so, suggests that deception detection
is far more intricate than previously conceptualized. For instance, interpersonal
deception theory (Buller & Burgoon, 1996; Burgoon, 1994) asserts that deceptive
acts contain many parts, including the deceptive message itself, ancillary
behaviors intended to bolster credibility and image, and unintended behaviors
that leak deceptive intent. The framework presented in this study illustrates
the ways in which all of these information sources and much more can be
integrated to judge deception. Indeed, in addition to verbal and nonverbal
information, this study showed the importance of knowledge and inference
in forming impressions of deceit. A primary contribution, therefore, is
an understanding of the ways in which cognitive networks operate holistically,
processing multiple and possibly contradictory pieces of information simultaneously.11
One advantage of such an approach is that a concept or explanation (e.g.,
he is lying), which may have been perceived as coherent when evaluated
in isolation, may be seen as incoherent when evaluated comparatively. In
addition, such an approach allows us to see how multiple explanations of
the same behavior may be active in a mental model at the same time and
how some of these explanations win out over others (see Read, 1992). Finally,
an implication of this complexity is that detecting deception can be cognitively
demanding, requiring detectors to integrate large amounts of information
from multiple sources. This notion is consistent with interpersonal deception
theory (Buller & Burgoon, 1996), which asserts that both the enactment
and detection of deception are cognitive demanding tasks and that deception
detectors must deal with ongoing, rapidly changing information that may
be incongruent or ambiguous.

This study indicated that deception detection is intricate not only by
showing the ways in which people integrate multiple sources of information,
but also by showing that information relevant to deception detection is
structured in a variety of ways that have not been addressed previously.
Specifically, although this study, like prior research, found that individuals
attempt to detect deception by searching for verbal and nonverbal signs
indicative of deceit, it identified several additional substructures by
which deception detection operates. Examples of such substructures were
referred to as "constituents of deception," "analogies," and "explaining
away"

The first of these ("constituents of deception") indicates that detectors
do not simply look for signs of deception; they look for deception itself.
That is, detectors organize what they perceive as inconsistencies (e.g.,
"R said Dan was caring" but "R said Dan lied to her") differently from
the way in which they organize information that they believe signals deceit.
This distinction is important, because differences in the ways in which
concepts are related to one another in a network can affect the degree
to which such concepts become more or less salient for a communicator.


It is also worth mentioning that some of the inconsistencies that were
believed to constitute deceit were based, in part, on the participants'
personal biases or stereotypes. For example, one participant thought that
"R was lying" because "R claimed that Jack, a punker, drove recklessly,
when, in fact, punkers don't drive reckless." That such personalized knowledge
is used when attempting to explain behavior is consistent with previous
research in the area of social cognition (e.g., see Read, 1987; Schank
& Abelson, 1977; Wilensky, 1983), thereby supporting the notion that a
social cognitive framework is a useful avenue for understanding deception
detection. Moreover, the identification of "constituents" as a substructure
for detecting deception indicates that the process of judging veracity
cannot be understood fully without also understanding the role of personalized
knowledge and the manner in which such knowledge is structured.

A second substructure previously unidentified in deception research is
the "analogy." This study indicated that detectors readily use themselves
and, to some extent, others as references for judging the veracity of a
third party. That is, detectors compare the way in which they behave when
lying to the behavior of the person they are trying to detect. If the behaviors
match, the detector is more likely to view the "detectee" as deceptive.
It should be noted that the role of analogies in the process of attribution
and causal reasoning has been noted elsewhere (e.g., Kelley, 1972; Read
1983, 1984; Read & Cesa, 1991). Moreover, the use of analogies indicates,
once again, the theoretical importance of acknowledging the integral role
that knowledge structures play in the perception of deception. People use
stored knowledge about themselves and others to detect deception.

A third previously neglected substructure is called "explaining away" and
involves an alternative explanation for an inference or behavior. As discussed
earlier, the theoretical implication of such alternative explanations is
that they show how individuals are able to decode inconsistent behaviors
(e.g., simultaneously observe behaviors that indicate deceptiveness and
behaviors that indicate truthfulness) and still reach a conclusion about
a person's veracity. This substructure also underlines the notion that
deception detection is just one part of a larger attributional process.
Specifically, in many participants' mental models, an attribution of deceptiveness
was just one of many possible interpretations for observed behaviors. For
example, in one person's mental model, "the target is being interviewed"
and "the target is lying" competed to explain why "the target is fidgeting."
Thus, this study indicates that the process of deception detection cannot
be understood completely without an examination of the multiple interpretations
competing within a person's mental model. Indeed, information that, at
first glance, may seem unrelated to judgments of veracity (e.g., "S was
tired"; "A lot of time has passed since what S is saying occurred"; "S
is shocked he had an Asian girlfriend") may function as alternative interpretations
that are used to "explain away" competing attributions of deceptiveness.
The framework presented in this study provides a useful approach for understanding
how such alternative interpretations become more coherent than others and,
in turn, are used to make sense of social interactions.

It is worth pointing out that when examining the content of these and the
other substructures, this study deviates from the pattern of most previous
deception research by showing that verbal information had a strong influence
on veracity judgments. This is significant considering that ever since
Mehrabian's (1972) claim about the preeminence of nonverbals, and Ekman
and Friesen's (1969) assumption that deception is revealed more through
verbal than nonverbal cues, deception scholars have presumed that detection
judgments derive almost exclusively from nonverbals and have designed their
studies accordingly. Because of this, little work has examined deceptive
message designs (for notable exceptions, see Bavelas, Black, Chovil, &
Mullett,1990; Burgoon, Buller, Guerrero, Afifi, & Feldman, 1996; McCornack,
1992). The results of this study point to the importance of research that
considers the effects of both verbal and nonverbal information in the process
of enacting and perceiving deception.

In addition to indicating a number of substructures that influence the
coherence of deception attributions, this study contributes to an understanding
of the dynamic nature of deception detection. Past research (e.g., Buller
& Burgoon,1996; Burgoon, 1996; Burgoon, Bullet, Ebesu, White, & Rockwell,
1996; Buller, Strzyzewski, & Comstock, 1989, 1991; Stiff & Miller, 1986)
has conceptualized deception as a process, but the conceptualization of
deceptiveness judgments as a process has been ignored. This study, however,
addressed how and why people's attributions change over time. As they are
exposed to new information, people add to and alter their mental models
for judging veracity. Such changes can sway their perceptions so that a
person who was once seen as deceptive can subsequently be viewed as truthful
(or vice versa). This finding, and the finding that over half of the participants
changed their attributions at least once while watching the stimulus tapes,
contradicts past research that indicates that initial impressions of a
deceiver tend to anchor and bias later attributions (see Zuckerman et al.,
1987; Zuckerman, Koestner, Colella, & Alton, 1984; Zuckerman et al., 1981).
It is consistent, however, with literature in social cognition (e.g., Read,
1987), which argues that observed behaviors, inferences, and knowledge
can be used to transform collections of attributions.

In addition to illustrating the dynamic nature of deception detection,
this study supports Miller and Read's (1991) argument that Thagard's simulation
provides an intriguing methodology to explore individual differences in
the coherence of mental models. Indeed, whereas there were similarities
in the substructures used by participants in this study, deception detection
was highly idiosyncratic in other respects. First, the content of mental
models varied considerably from person to person. In some cases, such content
was surprisingly rich. For example, one participant believed that "R,"
the female stimulus target, said "Dan, an ex-boyfriend, was a liar" because
"R was lying" and because "R was projecting." Another participant believed
that "R was lying" because "R said Jack (a punker) was a good dancer" and
"Punk rockers are not good dancers." It is worth noting that, as in the
latter example, the idiosyncratic content in mental models was especially
noticeable in the different knowledge structures that were activated by
participants to judge veracity. To be sure, this study suggests that the
role of knowledge in the process of deception detection is more extensive
and complex than previously imagined. For instance, prior work has argued
that knowledge gained from behavioral familiarity should affect perceptions
of deception. This study showed that, in addition, personalized knowledge
regarding the nature of reality (e.g., "Punkers don't drive reckless")
is used in multiple ways to make attributions regarding another person's
deceptiveness. Such detailed connections among specific inferences, knowledge,
and data cannot be examined or understood as readily with previously used
methods. Thus, a contribution of the framework in this study is that it
enables a richer understanding of the specific sources of information used
in the process of deception detection.

Second, this study indicated that the coherence of mental models for judging
veracity varied considerably from person to person. Thus, even when presented
with the same set of facts (i.e., the same stimulus video), people constructed
vastly different mental models. Moreover, such models differed even when
participants made the same attribution. The power of Thagard's model is
that it helps us understand why people differ in how they view deceptive/truthful
communicators. The model suggests that to understand why people differ
in their perceptions of detection, researchers must consider the specific
ways in which particular individuals combine data, inference, and knowledge
to construct meaning. Moreover, the study shows that attributions of deceptiveness
may not depend on the observation of any particular behavior but rather
on the configurational array within which such observations are embedded.


Of course, the finding that individuals use vastly different types of information
and generate idiosyncratic judgment models has implications for current
research in deception; this study suggests that we may need to reconsider
popular assumptions regarding cross-individual consistency in certain deception-related
behaviors. Because the content and coherence of mental models appear to
be so idiosyncratic, it may be impossible to find commonalities across
individuals when examining deception detection at these levels of analysis.
However, at a different level, because this study found that particular
substructures (see above) were used by nearly all of the participants,
it may be possible to make generalizations concerning the use of such substructures.
Although such an analysis is beyond the scope of this study, future research
might examine whether there are conditions under which the various substructures
are used.

By way of example, it is possible that people are required to expend more
cognitive effort when using some substructures than when using others.
Nonverbal signs, for instance, may activate attributions of deception almost
automatically, whereas constituents of deception, which are often made
up of contradictory verbal statements or verbal statements that contradict
knowledge held by the detector, may require closer scrutiny to the potentially
deceptive message. With this in mind, individual differences in communicator
characteristics such as the need for cognition (see Petty & Cacioppo, 1986),
which is associated with carefully scrutinizing incoming information, and
conversational attentiveness (Burns, Seiter, & Miller, 1990), which is
associated with increased attention to verbal communication, may also be
associated with the use of the constituents' substructure. On the other
hand, individuals who are low in the need for cognition and who are conversationally
inattentive may rely more on signs to detect deception.

Similarly, a communicator's motivations and/or ability to attend to a message
may also affect the types of substructures he or she uses to detect deception.
For example, when communicators are motivated to detect deception (e.g.,
they are led to be suspicious or believe they will be rewarded for successful
detection), they may scrutinize messages more and, as a result, use more
constituents than communicators who are not motivated. Moreover, when communicators
are unable to attend to, or scrutinize, a message (e.g., they are distracted,
they are involved in an interaction rather than simply observing it), they
may rely on signs of deception more than other substructures.

Finally, the degree to which a detector is familiar with a communication
source may also affect the substructures used by the detectee. Specifically,
because people possess more knowledge about intimates than they do about
strangers, and because knowledge structures are more typical in some substructures
(e.g., analogies, constituents, cohypotheses) than others (e.g., signs),
it is possible that some substructures will be used more when judging the
veracity of an intimate than when judging the veracity of a stranger. Future
research is needed to explore these possible conditions of substructure
use.

Besides that, this study hints at a number of other potential avenues for
future research on the topic of deception. First, future research should
examine the relationship between concepts discussed in this study and accuracy
at detecting deception. With this in mind, however, it should be noted
that only particular concepts may be related to accuracy. For example,
although it was shown that the strength of subjects' overall mental models
predicted the inferences they made about veracity and the confidence with
which they made such inferences, there is no reason to suspect that the
strength of these mental models will be related to the accuracy with which
individuals are able to detect deception. Specifically, when constructing
a mental model of another person, an individual may rely on several types
of information (i.e., nonverbal behaviors, stereotypes, and so forth).
However, all of these types of information are not necessarily accurate
indicators of deception (see Riggio & Friedman, 1983). Thus, to the extent
that individuals use erroneous clues to deception, they can be inaccurate
in their attribution even though their mental models for deceptiveness
may be highly activated.

On the other hand, it is possible that, when their content is also considered,
the use of particular substructures may be related to accurate deception
detection. For instance, as noted above, many of the "constituents of deception"
identified in this study were composed of contradictory statements that
the stimulus target made. Because such statements are by definition deception,
when they are identified, they should promote accurate detection more than
other types of structures. On the other hand, when "constituents of deception"
are not composed of such contradictory statements, their use should not
facilitate accurate detection.

Second, further work is needed to explore the confidence with which individuals
make judgments of veracity. Thagard (1989), for example, noted that each
of the concepts in a network can be assigned certain weights before the
ECHO program is run. Future work could therefore ask participants to assign
their own values to the inferences they make on the basis of how confident
they were about making the inferences. It is possible that such a procedure
could enhance our ability to predict confidence.

Third, future work should explore the effects of the order in which information
is received and change on the construction of mental models. Specifically,
does the order in which people receive information affect how likely they
are to select one interpretation over another (Read & Miller, 1989)? Also,
do individuals who change their attributions over the course of time make
more accurate attributions? And if so, what types of changes lead to more
accurate perceptions? As people's mental models develop further, are we
better able to predict their attributions over time? How are models affected
when new information is added? Clearly, Thagard's (1989) simulation provides
a methodology that would be useful for answering such questions. For instance,
the simulation makes it possible to examine hypothetically the ways in
which information that is added to a system affects coherence. As Miller
and Read (1991) noted, "If part of the network was changed (i.e., for example,
if one belief was altered), the model may be able to predict the likely
higher order structures around which the resulting model would cohere"
(p. 87).

In other words, by using coherence analysis, it might be possible to predict
what concepts in a model would need to be altered or added to change a
person's attribution regarding veracity.

On the other side of the coin, and as alluded to earlier, the theory of
explanatory coherence suggests a number of possibilities for future research
on the enactment of deception. From this perspective, the primary task
facing a deceiver who wants to be perceived as truthful is that he or she
must try to construct the most coherent lie possible. How do individuals
attempt to construct and tell coherent lies? The theory of explanatory
coherence would argue that a deceiver must create and integrate facts that
activate attributions of truthfulness, provide alternative and simple explanations
for data that are explained by deceptiveness, and avoid contradictions.
Clearly, the task of constructing a lie that appears truthful is a difficult
one, and some individuals may be more skilled at it. For instance, do Machiavellians
or socially skilled individuals tell more coherent lies, and, if so, are
their stories believed more by others because of it? Are factors such as
the spontaneity of the lie (see Cody, Marston, & Foster, 1984) related
to the coherence or believability of the lie? Lastly, do people tell more
coherent lies to those they know than to those who are strangers?

Finally, contemporary work (Buller & Burgoon, 1996) has argued that a complete
understanding of deception requires an examination of the ways in which
deceivers and detectors mutually influence one another in dynamic, ongoing
interactions. For example, interpersonal deception theory (Buller & Burgoon,
1996; Burgoon, 1994; Burgoon & Buller, 1994) asserts that a liar's behavior
may cause a receiver to react suspiciously. Such behavior, in turn, may
affect the liar's subsequent behavior. Although this study did not permit
senders and receivers to interact, the framework presented can be used
in future work to extend what is known about the interactional nature of
deception detection. For example, the framework presented in this study
would enable a researcher to examine (a) how the construction of a mental
model leads a detector to be suspicious, (b) how the mental model affects
the detector's subsequent behaviors (e.g., behaviors indicative of suspiciousness),
(c) how the detector's behavior is used as new input in the mental model
of the deceiver, and (d) how the amended model of the deceiver affects
subsequent deceptive behavior.

Before concluding, some limitations regarding the generalizability of this
study must be mentioned. First, the lies told by the stimulus targets were
not sanctioned as lies might be in the real world. Furthermore, although
the lies were long and narrative in nature, the fact that they were told
in front of a video camera and that stimulus targets were permitted to
plan their lies might have affected how the lies were told. Third, because
the deceptions created by stimulus subjects in this study could be characterized
as fabrications, the results may not generalize to attributions made when
other forms of deception are used (for a discussion of alternative forms
of deception, see Buller & Burgoon, 1996). Fourth, because the majority
of judges in this study were female, the results may only generalize to
situations with similar populations. Fifth, although the experimenter made
every effort to respond neutrally to participants' answers, there may have
been experimenter expectancy effects, because the interviewer was not blind
to the conditions in the study.

It should be noted that although this study suggests that the analyses
described above depict neural processes, self-reports, inferences, and
the ECHO simulation may not be veridical indicators of neural activation.
As Thagard (1989) noted,

Despite ECHO's parallelism and use of vague neural metaphor of connections,
I have not listed neural plausibility as one of its advantages, because
current knowledge does not allow any sensible mapping from nodes of ECHO
representing concepts to anything in the brain. (p. 458) Even so, it has
been argued that ECHO's parallelism has advantages independent of the brain
analogy (Thagard, 1986) and that, considering the neonatal state of the
study of cognition, researchers should pursue the study of the mind using
multiple methodologies.12

Finally, the stimulus tape-viewing conditions were artificial, particularly
in cases where the tape was stopped to acquire attributional information.
Specifically, even though participants did not seem to have difficulty
producing the data, and even though participants were instructed to indicate
to the interviewer the thoughts that they had had while watching the videotaped
targets, stopping the tapes was intrusive and therefore may have resulted
in more elaborated mental models than are present in a typical interaction.
Clearly, collecting on-line data during the course of an interaction presents
methodological problems that should be addressed in future research. In
the meantime, it should be noted that this study may be more generalizable
to situations where individuals are given time to carefully consider the
veracity of the communication they observe (e.g., during breaks in courtroom
deliberations).

The conceptualization of deception detection advanced in this study differs
considerably from the ways in which it has been viewed previously. This
study portrays it as a dynamic and changing process by which individuals
attempt to integrate, structure, and make sense of an array of inferences,
knowledge, and verbal and nonverbal behaviors. As such, it provides a richer
understanding of how and why people decide whether someone is lying or
behaving honestly. It is hoped that the current study represents only a
beginning for examinations of this type, because the ways in which people
attempt to make sense of others around them is integral not only for an
understanding of the process of deception detection but also for an understanding
of the process of communication itself.

[IMAGE TABLE] Captioned as: APPENDIX A
 Captioned as: APPENDIX B
 Captioned as: APPENDIX C

[IMAGE TABLE] Captioned as: APPENDIX D

NOTES

1. It is interesting to note that previous research (e.g., Riggio & Friedman,
1983) has demonstrated that these scripts are not very accurate (i.e.,
behaviors used to detect deception are not generally the same behaviors
that are enacted when telling lies).

2. Interview time between each video segment and for each participant varied
considerably, with some participants giving several responses and others
few responses. Data regarding specific amounts of time were not collected.
My impressions are that interview segments ranged somewhere between 2 and
10 minutes and averaged about 5 to 6 minutes.

3. For further discussion of ECHO and its implementation of the principles
of explanatory coherence, see Thagard (1989) and Miller and Read (1991).


4. The activation (aj) of a concept (]) is updated using the following
equation: a ^subj^(t + 1) = a ^subj^(t)(1 - d) + enet ^subj^[max - a^subj^(t)]
+ inet^subj^[a ^subj^(t) - min].

Here d is a decay parameter that decrements each concept at every cycle,
enet, is the net excitatory input and ineti is the net inhibitory input.
enet,is equal to Sigmaw^subij^, a^subi^(t) if w^subij^ > 0, and inet^subj^
is equal to Sigmaw^sub4^a^subi^(t) if w^subij^ < 0, where w^subij^ is the
weight between concepts i and j. max is the maximum activation value possible
(1) and min is the minimum activation value possible (-1). The degree to
which a concept becomes activated depends on the number of concepts connected
to it, the strength and valence of the links, and the activation level
of connected concepts (see Miller & Read, 1991; Thagard, 1989).

5. It is important to note that scores indicating which concepts were most
activated in a mental model were derived independently from judgments of
veracity. 6. Analogies lead to increased activation for those concepts
involved in the analogy. 7. Because participants' observations and inferences
were collected in matrix form (see Method section), a network analysis
of the substructures and content within participants' mental models was
possible.

8. Despite the idiosyncratic nature of the specific content of concepts,
it should be noted that, consistent with previous research, there were
consistencies regarding the general nature of signs participants used.
For instance, all of the participants used inferences and verbal and nonverbal
cues as signs of deceptiveness. Of those used, 34% of cues were nonverbal
in nature (e.g., "S's legs were shaking"), 25% were verbal in nature (e.g.,
"R's answers were vague"), and 41% were inferences made about the targets
or the targets' stories (e.g., "R is very uncomfortable").

9. Although an examination of accuracy was beyond the scope of this study,
because they may be of interest to deception scholars, results of accuracy
tests are presented here. To determine participants' accuracy, their dichotomous
attributions of targets' veracity (i.e., "the target is lying" or "the
target is telling the truth") were checked against the targets' actual
behavior. Overall, 66 participants were accurate in their judgments, whereas
54 were inaccurate. A series of phi coefficients were computed to assess
the effects of viewing condition, target sex, and target veracity on partiapants'
accuracy Results indicated that the accuracy of participants who were interviewed
(48% were accurate) was not significantly different from the accuracy of
participants who viewed the stimulus tapes nonstop (42% were accurate)
(phi = .067, chi^sup2^ = .539, df = 1, p > .05). Moreover, targets were
not judged more accurately when lying (33% of deceptive targets were judged
accurately) than when telling the truth (50% of truthful targets were judged
accurately) (phi = .17, chi^sup2^ = 3.43, df = 1, p > .05), and the female
target was not judged more accurately than the male target (female was
judged accurately in 40% of the cases and male was judged accurately in
45% of the cases) (phi = .05, chi^sup2^ = .307, df = 1, p > .05).

10. Because Miller and Read (1991) argued that negatively activated concepts
in a network fade away, whereas positively activated concepts are used
to make attributions, it made theoretical sense to use participants' most
positively activated concept scores to test Hypothesis 2. Indeed, because
positively activated concepts presumably guide the attribution process,
the level at which such concepts are activated should predict attributional
confidence.

11. As an application and illustration of parallel distributed processing,
this study should also be of interest to scholars in the field of cognitive
science. The issue here centers around whether traditional models of cognition
that require vast numbers of microsteps if implemented sequentially can
provide a plausible account of the rapid nature of human thought (for further
discussion, see McClelland, Rumelhart, & Hinton, 1986). This study illustrates
that parallel distributed processing models may provide a more reasonable
account of information processing by showing how a number of different
pieces of information may be kept in mind at once and simultaneously influence
one another. 12. For additional limitations of the ECHO simulation, refer
to Thagard (1989).

REFERENCES

Bavelas, J. B., Black, A., Chovil, N., & Mullett, J. (1990). Equivocal
communication. Newbury Park, CA: Sage.

Bond, C. F Jr., Kahler, K. N., & Paolicelli, L. M. (1985). The miscommunication
of deception:

An adaptive approach. Journal of Experimental Social Psychology, 21, 331-345.
Brandt, D. R., Miller, G. R, & Hocking, J. E. (1980). The truth deception
attribution: Effects of familiarity on the ability of observers to detect
deception. Human Communication Research, 6, 99-108.

Buller, D. B., & Burgoon, J. K. (1996). Interpersonal deception theory.
Communication Theory, 6(3), 203-242.

Buller, D. B, Strzyzewski, K D., & Comstock, J. (1989, November). The efficacy
of probing as a deception detection strategy: The effect of suspicion and
knowledge of the source. Paper presented at the annual meeting of the Speech
Communication Association, San Franciso. Buller, D. B., Strzyzewski, K.
D., & Comstock, J. (1991). Interpersonal deception I: Deceivers'

reactions to receivers' suspicions and probing. Communication Monographs,
58,1-24. Buller, D. B., & Walther, J. B. (1989, February). Deception in
established relationships: Application of schema theory. Paper presented
at the annual meeting of the Western Speech Communication Association,
Spokane, WA.

Burgoon, J. K. (1994). Prologue to special issue: Interpersonal deception.
Journal of Language and Social Psychology, 13(4), 357-365.

Burgoon, J. K., & Buller, D. B. (1994). Interpersonal deception: Ill. Effects
of deceit on perceived communication and nonverbal behavior dynamics. Journal
of Nonverbal Behavior, 18,155-184.

Burgoon, J. K., Buller, D. B., Ebesu, A. S., & Rockwell, P (1994). Interpersonal
deception: V

Accuracy in deception detection. Communication Monographs, 61(4), 303-325.
Burgoon, J. K., Buller, D. B., Ebesu, A. S., White, C. H., & Rockwell,
P (1996). Testing interpersonal deception theory: Effects of suspicion
on communication behaviors and perceptions. Communication Theory, 6(3),
243-267.

Burgoon, J. K., Buller, D. B., Guerrero, L. K., Afifi, W. A., & Feldman,
C. M. (1996). Interpersonal deception: XII. Information management dimensions
underlying deceptive and truthful messages. Communication Monographs, 63(1),
50-69.

Burns, D. M., Seiter, J. S., & Miller, L. C. (1990, February). Who remembers
best what others say? The assessment of conversational attentiveness. Paper
presented at the annual meeting of the Western Speech Communication Association,
Phoenix, AZ.

Cody, M. J., Marston, P J., & Foster, M. (1984, May). Paralinguistic and
verbal leakage of deception as a function of attempted control and timing
of questions. Paper presented at the annual meeting of the International
Communication Association, San Francisco. Comadena, M. E. (1982). Accuracy
in detecting deception: Intimate and friendship relationships. In M. Burgoon
(Ed.), Communication yearbook 6 (pp. 446-472). Beverly Hills, CA: Sage.


Cupach, W R, & Metts, S. (1986). Accounts of relational dissolution: A
comparison of marital and non-marital relationships. Communication Monographs,
53, 311-334. Ekman, P, & Friesen, W V. (1969). Nonverbal leakage and clues
to deception. Psychiatry, 32, 88-105.

Fiedler, K., & Walka, I. (1993). Training lie detectors to use nonverbal
cues instead of global heuristics. Human Communication Research, 20,199-223.


Gordon, R A., Baxter, J. C., Rozelle, R M., & Druckman, D. (1987). Expectations
of honest,

evasive, and deceptive nonverbal behavior. Journal of Social Psychology,
127, 231-233. Hale, J. S., & Stiff, J. B. (1990). Nonverbal primacy in
veracity judgments. Communication Reports, 3, 75-83.

Harvey, J. H., Agostinelli, G., & Weber, A. L. (1989). Account making and
the formation of expectations about close relationships. In C. Hendrick
(Ed.), Close relationships (pp. 3942). Newbury Park, CA: Sage.

Harvey, J. H., Wells, G. L., & Alvarez, M. D. (1987). Attribution in the
context of conflict and separation in close relationships. In J. H. Harvey
& R E Kidd (Eds.), New directions in attribution research (pp. 235-260).
Hillsdale, NJ: Lawrence Erlbaum. Kelley, H. H. (1972). Attribution in social
interaction. In E. E. Jones, D. E. Kanouse, H. H. Kelley, R E. Nisbett,
S. Valins, & B. Weiner (Eds.), Attribution: Perceiving the causes of behavior
(pp. 151-174). Morristown, NJ: General Learning Press. Kintsch, W (1988).
The role of knowledge in discourse comprehension: A construction-inte

gration model. Psychological Review, 95, 163-182.

Kraut, R E. (1978). Verbal and nonverbal cues in the perception of lying.
Journal of Personality and Social Psychology, 36, 380-391.

Kraut, R E., & Poe, D. (1980). Behavioral roots of person perception: The
deception judgments of customs inspectors and laymen. Journal of Personality
and Social Psychology, 39, 784-798.

Maier, N.RE & Janzen, J. C. (1967). Reliability of reasons used in making
judgments of honesty and dishonesty. Perceptual and Motor Skills, 25,141-151.
McClelland, J. L., Rumelhart, D. E., & Hinton, G. E. (1986). The appeal
of parallel distributed processing. In D. E. Rumelhart, J. L. McClelland,
dc the PDP Research Group, Parallel distributed processing: Explorations
in the microstructure of cognition (Vol. 1, pp. 3-44). Cambridge: MIT Press.


McCornack, S. A. (1992). Information manipulation theory. Communication
Monographs, 59, 1-16.

McCornack, S. A., & Parks, M. R (1985). Deception detection and relationship
development: The other side of trust. In M. McLaughlin (Ed.), Communication
yearbook 9. Beverly Hills, CA: Sage.

Mehrabian, A. (1972). Nonverbal communication. Chicago: Aldine-Atherton.
Miller, L. C., & Read, S. J. (1991). On the coherence of mental models
of persons and relationships: A knowledge structure approach. In GJ.O.
Fletcher & E Fincham (Eds.), Cognition in close relationships (pp. 69-99).
Hillsdale, NJ: Lawrence Erlbaum. O'Sullivan, M., Ekman, P., & Friesen,
W V. (1988). The effect of comparisons on detecting deceit. Journal of
Nonverbal Behavior, 12, 203-216.

Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central
and peripheral routes to attitude change. New York: Springer-Verlag.

Read, S. J. (1983). Once is enough: Causal reasoning from a single instance.
Journal of

Personality and Social Psychology, 45, 323-334.

Read, S. J. (1984). Analogical reasoning in social judgment: The importance
of causal theories. Journal of Personality and Social Psychology, 46, 4-25.
Read, S. J. (1987). Constructing causal scenarios: A knowledge structure
approach to causal

reasoning. Journal of Personality and Social Psychology, 52, 288-302. Read,
S. J. (1992). Constructing accounts: The role of explanatory coherence.
In M. L. McLaughlin, M. J. Cody, & S. J. Read (Eds.), Explaining one's
self to others: Reason-giving in a social context (pp. 3-19). Hillsdale,
NJ: Lawrence Erlbaum. Read, S. J., & Cesa, I. L. (1991). This reminds me
of a time when . . .: Expectation failures in reminding and explanation.
Journal of Experimental and Social Psychology, 27,11-25. Read, S. J., &
Miller, L. C. (1989). Explanatory coherence in understanding persons, interactions,
and relationships. Behavioral and Brain Sciences, 12, 485-486.

Riggio, R E., & Friedman, H. S. (1983). Individual differences and cues
to deception. Journal of Personality and Social Psychology, 45, 899-915.


Riggio, R E., Tucker, J., & Widaman, K. E (1987). Verbal and nonverbal
cues as mediators of

deception ability. Journal of Nonverbal Behavior, Il, 126-143. Rumelhart,
D. E. (1984). Schemata and the cognitive system. In R S. Wyer Jr., & T.
K Srull (Eds.), Handbook of social cognition (Vol. 1, pp. 161-185). Hillsdale,
NJ: Lawrence Erlbaum. Rumelhart, D. E., & McClelland, J. L. (1986). Parallel
distributed processing: Explorations in the microstructure of cognition.
Vol. 1: Foundations. Cambridge: MIT Press. Schank, R. C., & Abelson, R
P (1977). Scripts, plans, goals and understanding. Hillsdale, NJ: Lawrence
Erlbaum.

Seiter,J. S., &Wiseman, R. L. (1995). Ethnicity and deception detection.
Journal of the Northwest Communication Association, 23, 24-38.

Stiff, J. B., Hale, J. L., Garlick, R. & Rogan, R G. (1990). Effect of
cue incongruence and social normative influences on individual judgments
of honesty and deceit. Southern Communication Journal, 55,192-205.

Stiff, J. B., & Miller, G. R (1984, May). Deceptive behaviors and behaviors
which are interpreted as deceptive: An interactive approach to the study
of deception. Paper presented at the annual meeting of the International
Communication Association, San Francisco. Stiff, J. B., & Miller, G. R
(1986). "Come to think of it . . .": Interrogative probes, deceptive communication,
and deception detection. Human Communication Research, 12, 339-357.

Streeter, L. A., Krauss, R M., Geller, V, Olson, C., & Apple, W. (1977).
Pitch change during attempted deception. Journal of Personality and Social
Psychology, 35, 345-350. Thagard, P (1986). Parallel computation and the
mind-body problem. Cognitive Science, 10,

301-318.

Thagard, P (1989). Explanatory coherence. Behavioral and Brain Sciences,
12, 435-467. Toris, C., & DePaulo, B. M. (1985). Effects of actual deception
and suspiciousness of deception

on interpersonal perception. Journal of Personality and Social Psychology,
47,1063-1073. Wilensky, R (1983). Planning and understanding: A computational
approach to human reasoning. Reading, MA: Addison-Wesley.

Zuckerman, M., Fischer, S. A., Osmun, R. W., Winkler, B. A., & Wolfson,
L. R (1987).

Anchoring in lie detection revisited. Journal of Nonverbal Behavior, 11,
4-12. Zuckerman, M., Koestner, R., Colella, M. J., & Alton, A. O. (1984).
Anchoring in the detection of deception and leakage. Journal of Personality
and Social Psychology, 47, 301-311. Zuckerman, M., Koestner, R, & Driver,
R. E. (1981). Beliefs about cues associated with deception. Journal of
Nonverbal Behavior, 6, 105-114.

John S. Setter (PhD, University of Southern California, 1993) is an assistant
professor in the Department of Languages, Philosophy and Speech Communication
at Utah State University. A previous version of this article was presented
at the National Communication Association's convention in New Orleans,
1994. This article was developed from part of the author's dissertation
directed by Lynn C. Miller at the University of Southern California. The
author would like to thank the following people for their invaluable contributions
toward the completion of this project: Lynn C. Miller; Michael Cody; Steven
Read; Scott Smith; Verlaine McDonald; Susan Avanzino; Kerry Osborne; Tom
Hollihan; Gwen Brown; Jon Bruschke; Harold Kinzer; Cindy Gallois; Steven
McComack; one anonymous reviewer; Nancy Birch; and Rhonda, Delores, and
Debora Seiter. Please address correspondence concerning this article to
John S. Seiter, Department of Languages, Philosophy and Speech Communication,
Utah State University, Logan, UT 84322-0720; phone: (435) 797-0138; e-mail:
jsseiter@cc.usu.edu.

Reproduced with permission of the copyright owner.
Further reproduction or distribution is prohibited without permission.


=============================== End of Document ================================