Content uploaded by Michael Slepian
Author content
All content in this area was uploaded by Michael Slepian on Jun 30, 2016
Content may be subject to copyright.
Psychological Science
24(11) 2335 –2338
© The Author(s) 2013
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0956797613487384
pss.sagepub.com
Short Report
In the card game of poker, players attempt to disguise
cues to the quality of their hand, either by concealment
(e.g., adopting the well-known, expressionless “poker
face”) or by deception. Recent work, however, demon-
strates that motor actions can sometimes betray inten-
tions. The same action can have different movement
dynamics depending on the underlying intention
(Becchio, Sartori, & Castiello, 2010), and these subtle dif-
ferences can be decoded by observers (Becchio, Manera,
Sartori, Cavallo, & Castiello, 2012; Sartori, Becchio, &
Castiello, 2011). Thus, professional poker players’ inten-
tions may be visible from their actions while moving
poker chips to place bets. Even though professional play-
ers may be able to regulate their facial expressions, their
motor actions could betray the quality of their poker
hand. In three studies, we tested this hypothesis by exam-
ining observers’ perceptions of poker-hand quality. We
also examined individual differences in sensitivity to non-
verbal behavior and potential diagnostic motor behaviors
as cues to hand quality.
Study 1
Twenty brief silent video clips (mean duration = 1.60 s,
SD = 0.68 s) of professional poker players placing a bet
were extracted from randomly sampled videos of the
2009 World Series of Poker (WSOP) tournament. Three
versions of each clip were produced: Unaltered clips
showed players’ bodies from the table up, face-only
clips showed players from the chest up, and arms-only
clips showed only players’ arms pushing chips into the
table. Each player’s objective likelihood of winning dur-
ing the bet was known (WSOP displays these statistics
on-screen; however, we kept this information from par-
ticipants by obscuring part of the screen). The number of
chips wagered was not confounded with the likelihood
of winning (i.e., chip values varied markedly—no partici-
pants were poker experts nor knew chip values; see the
Supplemental Material available online for information
about the game of poker, WSOP, and further method-
ological details).
Seventy-eight undergraduates were divided into three
groups based on the type of clip they were shown. Each
group viewed the 20 clips in a random order and judged
the quality of each poker hand (1 = very bad, 7 = very
good). Next, participants rated their overall confidence in
their judgments (1 = not at all confident, 7 = very confi-
dent) and their experience with poker (1 = none, 7 = a
lot). Finally, they completed a measure of nonverbal sen-
sitivity (Bänziger, Scherer, Hall, & Rosenthal, 2011).
Data were analyzed using multilevel linear models
with quality ratings of the hand depicted in each clip,
nested within participants, predicting objective likeli-
hoods of winning. Specifically, the model included par-
ticipants’ quality ratings at Level 1, a set of dummy codes
representing condition at Level 2 (the face-only condition
was the reference group because our primary hypothesis
concerned a comparison between judgments based on
facial expressions vs. arm movements or vs. upper-body
movements), and all interactions predicting objective
likelihoods of winning. This analysis revealed the pre-
dicted interaction between the arms-only (vs. face-only)
condition and quality ratings, b = 1.68, t(1554) = 2.88, p =
.004, such that the arms-only group’s ratings significantly
predicted likelihoods of winning, b = 0.94, t(1554) = 2.26,
p = .02, whereas the face-only group’s ratings marginally
inversely predicted likelihoods of winning, b = −0.74,
t(1554) = −1.81, p = .07. The interaction between the
487384PSSXXX10.1177/0956797613487384Slepian et al.Perception of Poker Hands
research-article2013
Corresponding Authors:
Michael L. Slepian, Tufts University, Department of Psychology, 490
Boston Ave., Medford, MA 02155
E-mail: michael.slepian@tufts.edu
Nalini Ambady, Stanford University, Department of Psychology, 450
Serra Mall, Stanford, CA 94305
E-mail: nambady@stanford.edu
Quality of Professional Players’ Poker
Hands Is Perceived Accurately From
Arm Motions
Michael L. Slepian1, Steven G. Young2, Abraham M. Rutchick3,
and Nalini Ambady4
1Tufts University; 2Fairleigh Dickinson University; 3California State University, Northridge;
and 4Stanford University
Received 11/1/12; Revision accepted 4/2/13
at TUFTS UNIV on November 9, 2013pss.sagepub.comDownloaded from
2336 Slepian et al.
upper-body (vs. face-only) condition and quality ratings
was not significant, b = 0.95, t(1554) = 1.65, p = .10.
Reconducting these analyses with the individual-differ-
ence measures entered as predictors revealed no two- or
three-way interactions, ps > .07.1
We also examined participants’ accuracy scores, which
were computed by correlating participants’ poker-hand
ratings with players’ objective likelihoods of winning. If
these scores were significantly different from zero, per-
formance was different from chance (Table 1). Correlations
between these accuracy scores and participants’ nonver-
bal sensitivity, poker experience, and overall confidence
in their judgments were separately explored (Table 1).
These analyses also showed that judgments in the face-
only group were marginally worse than chance, which
suggests that players exhibited deceptive facial cues.
When isolating arm movements, however, analyses
showed that untrained participants judged the quality of
poker hands better than chance, which suggests that per-
ceptions of arm movements exert an independent influ-
ence on judgments of poker-hand quality. Judgments
made when viewing the players’ upper body (arm
motions plus the face) were at chance. Additionally,
when watching arm motions only, participants’ nonver-
bal sensitivity and poker experience were positively cor-
related with their accuracy.
Study 2
In Study 2, we replicated the arms-only accuracy finding
from Study 1 with a new set of silent video clips to ensure
the generalizabilty of the effect. Twenty-two new, ran-
domly sampled, chest-down close-ups of players placing
bets during the 2009 WSOP were extracted from video
clips as in Study 1 (mean duration = 1.54 s, SD = 0.74 s).
Again, the number of chips wagered was not confounded
with the likelihood of winning (see the Supplemental
Material). Thirty undergraduates judged poker-hand
quality from these new clips. As in the previous study,
data were analyzed with a multilevel model. Results rep-
licated those of Study 1. When participants viewed arm
motions, their judgments again predicted the objective
quality of professional poker players’ hands, b = 1.46,
t(558) = 2.70, p = .004. Participants’ performance was
greater than chance when they judged poker-hand qual-
ity from viewing players’ arm motions (Table 1).
Study 3
Players who have strong poker hands should be more
confident than players who have weak hands, and per-
haps this confidence is expressed in motor actions. To
the extent that participants’ poker-hand quality ratings
were influenced by player confidence, having partici-
pants judge player confidence could yield similar results.
Previous work demonstrates that anxiety disrupts smooth-
ness of body movement (Beuter & Duda, 1985), which
suggests that confidence (i.e., lack of anxiety) might be
revealed via smoother actions. Therefore, in Study 3, we
had participants in one condition judge player confi-
dence, and in a second condition, they judged how
smoothly the chips were pushed into the center of the
table. If greater confidence in players relates to smoother
motor action, smoothness judgments might also predict
likelihoods of winning.
Forty undergraduates viewed the same randomly
ordered videos from Study 2, judging player confidence
(“How confident does this person seem?”) or action
smoothness (“How smooth is this person’s movement?”;
1 = not at all, 7 = very). They subsequently completed
the measure of nonverbal sensitivity used in Study 1. We
ran a multilevel model, including participants’ quality
Table 1. Mean Accuracy in All Conditions and Correlations Between Accuracy and Individual-Difference Measures
Correlations
Study and condition Mean accuracy Nonverbal sensitivity Poker experience Confidence in judgments
Study 1
Upper body .02 [–.06, .09] .14 .14 .19
Face only –.07 [–.15, .01] .17 –.32 –.26
Arms only .07 [.01, .14] .40* .39* .26
Study 2 .15 [.11, .19] — — —
Study 3
Player confidence .15 [.07, .24] .46* — —
Smoothness of movement .29 [.22, .36] .14 — —
Note: Accuracy scores are the correlation of participants’ ratings of the quality of poker hands with players’ objective likelihoods of winning.
Values in brackets are 95% confidence intervals (created using Fisher’s transformed zs and then converted back to r values). If the 95%
confidence interval includes zero, accuracy is at chance.
*p < .05.
at TUFTS UNIV on November 9, 2013pss.sagepub.comDownloaded from
Perception of Poker Hands 2337
ratings at Level 1, a dummy code representing judgment
condition (with the player-confidence condition as the
reference group) at Level 2, and the interaction predict-
ing objective likelihoods of winning. Analyses revealed
a main effect of participants’ quality ratings, b = 3.33,
t(855) = 4.17, p < .001, but no significant interaction of
ratings with judgment condition, b = 0.54, t(855) = 0.58,
p = .56. Reconducting this analysis with the addition of
participants’ nonverbal-sensitivity scores and all interac-
tions did not reveal any significant main effects of non-
verbal sensitivity or interactions with nonverbal sensitivity
and other variables, ps < .64. Thus, both player confi-
dence and smoothness judgments significantly predicted
likelihoods of winning, which suggests that movement
smoothness might be a valid cue for assessing poker-
hand quality. It is unknown, however, how participants
interpreted “smoothness” or whether the players’ move-
ments that participants rated as smooth were truly
smoother than other players’ movements. Other physical
factors, such as speed, likely played a role (see Patel,
Fleming, & Kilner, 2012).
As in Study 1, we also explored correlations between
participants’ nonverbal sensitivity and accuracy scores.
Participants’ nonverbal sensitivity significantly correlated
with their accuracy as indexed by ratings of players’ con-
fidence, but not with their accuracy as indexed by ratings
of players’ smoothness of movement (Table 1), which
suggests the possibility that individual differences in non-
verbal sensitivity can be overcome when participants are
explicitly directed to attend to potentially diagnostic
motor cues.2
Discussion
In three studies with two unique video sets, observers
naive to the quality of professional players’ poker hands
could judge, better than chance, poker-hand quality from
merely observing players’ arm actions while placing bets.
The accuracy of participants’ judgments when viewing
players’ upper bodies was no different from chance, and
when observing players’ faces, participants’ accuracy was
nearly worse than chance, which suggests that players’
facial cues were deceptive. Arm motions might provide a
more diagnostic cue to poker-hand quality than other
nonverbal behaviors. Additionally, correlations between
nonverbal sensitivity and accuracy from viewing arm
motions suggest a positive relationship between the two
(see Table 1), and movement smoothness might be a
valid cue for assessing poker-hand quality, although
more research is needed to document the moderators of
the present effects.
These findings are notable because the players in the
stimulus clips were highly expert professionals compet-
ing in the high-stakes WSOP tournament. Additionally,
judges were untrained observers (cf. Ekman & O’Sullivan,
1991) watching clips on average less than 2 s long (see
Ambady & Rosenthal, 1992). Nevertheless, professional
poker players’ motor actions were revealing, enabling
perceivers to decode poker-hand quality from minimal
visual information. Even in very restrictive settings, motor
actions can yield important diagnostic information.
Author Contributions
M. L. Slepian, S. G. Young, A. M. Rutchick, and N. Ambady
conceived and designed the studies. M. L. Slepian, S. G. Young,
and A. M. Rutchick conducted the studies and analyzed the
data. All authors wrote the manuscript.
Acknowledgments
We thank Saheela Mehrotra for assistance in conducting the
experiments and Aneeta Rattan, Takuya Sawaoka, and Jessica
Salerno for helpful comments on an earlier version of this
manuscript.
Declaration of Conflicting Interests
The authors declared that they had no conflicts of interest with
respect to their authorship or the publication of this article.
Funding
This work was supported in part by National Science Foundation
Grant BCS-0435547 to N. Ambady and by a National Science
Foundation Graduate Research Fellowship to M. L. Slepian.
Supplemental Material
Additional supporting information may be found at http://pss
.sagepub.com/content/by/supplemental-data
Notes
1. The Quality Rating × Upper-Body Condition (vs. Face-Only
Condition) × Participant Confidence interaction was significant,
b = −0.71, t(1476) = −2.19, p = .03, but subsequent two-way
interactions were nonsignificant, ps > .07, which makes it dif-
ficult to interpret the three-way interaction.
2. Additionally, smoothness judgments yielded larger accu-
racy than confidence judgments. This is an example of when
judgments in a “micro” domain (physical properties of action)
may be a more diagnostic cue than judgments in a “molar”
domain (the meaning behind an action), whereas the reverse
is typically the case (see Weisbuch, Slepian, Clarke, Ambady,
& Veenstra-Van der Weele, 2010). Such conclusions about
greater accuracy, or higher correlations, in one condition than
in the other must be made with caution, however, because
neither nonverbal sensitivity nor judgment condition signifi-
cantly interacted with quality ratings in predicting objective
likelihoods to win.
References
Ambady, N., & Rosenthal, R. (1992). Thin slices of expressive
behavior as predictors of interpersonal consequences: A
meta-analysis. Psychological Bulletin, 111, 256–274.
at TUFTS UNIV on November 9, 2013pss.sagepub.comDownloaded from
2338 Slepian et al.
Bänziger, T., Scherer, K. R., Hall, J. A., & Rosenthal, R. (2011).
Introducing the MiniPONS: A short multichannel version
of the Profile of Nonverbal Sensitivity (PONS). Journal of
Nonverbal Behavior, 35, 189–204.
Becchio, C., Manera, V., Sartori, L., Cavallo, A., & Castiello,
U. (2012). Grasping intentions: From thought experiments
to empirical evidence. Frontiers in Human Neuroscience,
6, 117. Retrieved from http://www.frontiersin.org/Human_
Neuroscience/10.3389/fnhum.2012.00117/full
Becchio, C., Sartori, L., & Castiello, U. (2010). Toward you: The
social side of actions. Current Directions in Psychological
Science, 19, 183–188.
Beuter, A., & Duda, J. L. (1985). Sport psychology analysis of
the arousal/motor performance relationship in children
using movement kinematics. Journal of Sport & Exercise
Psychology, 7, 229–243.
Ekman, P., & O’Sullivan, M. (1991). Who can catch a liar?
American Psychologist, 46, 913–920.
Patel, D., Fleming, S. M., & Kilner, J. M. (2012). Inferring subjec-
tive states through the observation of actions. Proceedings
of the Royal Society B: Biological Sciences, 279, 4853–4860.
Sartori, L., Becchio, C., & Castiello, U. (2011). Cues to inten-
tion: The role of movement information. Cognition, 119,
242–252.
Weisbuch, M., Slepian, M. L., Clarke, A., Ambady, N., &
Veenstra-Van der Weele, J. (2010). Behavioral stability
across time and situations: Nonverbal versus verbal consis-
tency. Journal of Nonverbal Behavior, 34, 43–56.
at TUFTS UNIV on November 9, 2013pss.sagepub.comDownloaded from
