Sei sulla pagina 1di 8

Personality and Individual Dierences 40 (2006) 16831690

www.elsevier.com/locate/paid

Sex dierences in face recognition and inuence of facial aect


Stefan G. Hofmann
a

a,*

, Michael Suvak a, Brett T. Litz

Department of Psychology, Boston University, 648 Beacon Street, 6th Floor, Boston, MA 02215-2002, United States b Boston Veterans Administration Medical Center, Boston University, School of Medicine, United States Received 28 February 2005; received in revised form 1 November 2005; accepted 5 December 2005 Available online 10 March 2006

Abstract To study sex dierences in the recognition of human faces with dierent facial expressions, 65 female and 64 male participants learned to associate names with various male and female neutral faces. During the recall phase, participants were then asked to name the same persons depicting dierent emotional expressions (neutral, happy, angry, and fearful). Females were faster than males at naming male faces, and males were faster than females at naming female faces. All participants were faster at naming neutral or happy female faces than neural or happy male faces. These results suggest that opposite-sex faces require less processing time than same-sex faces, which is consistent with an evolutionary account. 2006 Elsevier Ltd. All rights reserved.
Keywords: Sex dierence; Recognition; Facial aect; Emotional processing; Reaction time; Evolutionary psychology

1. Introduction The human face is an important communication tool for signaling dierent emotional states in humans. Certain emotions, such as fear, anger, and happiness, are expressed through distinctive facial expressions (Ekman, 1993). The recognition of these emotions is, to some extend, specic to

Corresponding author. Tel.: +1 617 353 9610; fax: +1 617 353 9609. E-mail address: shofmann@bu.edu (S.G. Hofmann).

0191-8869/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2005.12.014

1684

S.G. Hofmann et al. / Personality and Individual Dierences 40 (2006) 16831690

ones culture (Elfenbein & Ambady, 2003) and sex (Eagly, 1987; Fisher, 2000; Goos & Silverman, 2002; Grossman & Wood, 1993; Hess, Adams, & Kleck, 2004; Shields, 2002). For example, angry faces are typically more associated with men than with women, whereas happy faces are typically more associated with women than with men (Fisher, 2000). Little is known about sex dierences in processing faces and facial aect. The evolutionary argument for sex dierences in facial aect recognition is linked to the greater physical vulnerability of women, which aords an advantage to those who can make very rapid decisions about facial expressions, especially expressions that warn of danger or signal intimacy (Geary, 1998). Consistent with this notion are studies that suggest that females are better than males at distinguishing facial expressions (e.g., Thayer & Johnson, 2000; Guillem & Mograss, 2005). It could further be argued that individuals attend more strongly to the identity of potential mates, resulting in better recognition of opposite-sex faces. The goal of the present study is to examine sex dierences in facial recognition. For this purpose, we designed a new cued recall paradigm. During an acquisition phase, participants rst learned names associated with neutral faces. In a recall phase, participants were then presented with the pictures of the same individuals depicting neutral, happy, fearful, or angry facial expressions and were asked to name the person. Half of the participants viewed same-sex faces, while the other half viewed opposite-sex faces. Consistent with evolutionary theory, we expected that males would be faster at naming female faces than at naming male faces, whereas females would be faster at naming male faces than at naming female faces. Consistent with the sex stereotype, we further predicted that angry male faces would be more quickly named than angry female faces, whereas happy female faces would be more quickly named than happy male faces.

2. Method 2.1. Participants Participants of the study were 65 female and 64 male undergraduate students of psychology at Boston University. The average age of the sample was 19 (range: 1722, SD = 1.11). Males and females did not dier ethnic composition, v2(2, N = 128) = 1.72, p = .89. The majority was Caucasian (62%), or Asian (25.6%). The rights of human subjects were protected. Males (M = 19.34; SD = 1.15) were older than females (M = 18.25; SD = 1.53), t(128) = 4.68, p = 0.00. 2.2. Procedure We employed a cued-recall procedure to evaluate memory for faces. The faces were taken from Ekman and Friesen (1976) and displayed on a computer screen. The paradigm consisted of a learning phase in which participants saw the face of a person with a neutral facial expression and a rst name underneath. Participants were instructed to learn the association between the face and the name. The names were common and short male and female names (e.g., Bob, Jane). During the cued recall phase subjects were presented with the faces again and instructed to name the person as quickly as possible. This time, the face showed a neutral, happy, fearful, or angry

S.G. Hofmann et al. / Personality and Individual Dierences 40 (2006) 16831690

1685

expression. The participants response latency to name the face was measured with voice-activated computer software. The experimenter recorded participants responses to examine the response errors. The dependent variables were the correct responses and response latencies. Fig. 1 shows an example of a computer screen display during the learning phase and the cued recall phase. Half of the female participants were presented with pictures of female faces, and half of them were presented with pictures of male faces. Similarly, half of the male participants were presented with pictures of male faces and half were presented with female faces. Three face-name pairs were presented four times during the learning phase. The order of the face-word pairs was randomized within sets. All faces in the pictures were directly looking at the subject, and each face-name pair was presented for 3000 ms with a 2000 ms inter-trial interval between pairs. After the fourth presentation of the set of face-name pairs, the participants were tested to see if they had learned the pairs. The faces without the names were presented on the screen for 3000 ms, and the participants were instructed to say out loud the name of the face. If the participants provided the correct name for each face, the experimenter hit a key, and the program moved on to the next part of the task. If the participants made an error in naming one or more of the faces, the experimenter hit a dierent key, and another learning trial was presented. After the additional learning set, participants were tested again, and this process continued until the participant successfully learned the names of all of the faces. Following the learning phase, participants were instructed to remember the face-name pairs because they would be tested later. Five-minutes separated the learning phase from the recall phase. During this time, participants completed a demographics questionnaire. If they completed it in

Fig. 1. Example of the stimulus presentation during the learning phase (left) and recall phase (right).

1686

S.G. Hofmann et al. / Personality and Individual Dierences 40 (2006) 16831690

less than ve minutes, they were instructed to sit and wait for further instructions. At the end of this ve-minute period, the participants viewed the face-name pairs one last time before moving on to the recall phase. Each of the three faces displayed four dierent facial expressions during the recall phaseneutral, happy, fearful, and angry. Thus, the stimulus set for the recall phase consisted of 12 cues (three dierent faces with four dierent facial expressions). Each of the 12 cues was presented 10 times for a total of 120 trials. The 120 trials were divided into 10 blocks of 12 trials. Each block consisted of the 12 cues (emotional faces). Within each block, the order of the cues was randomly determined for each participant by the software program (DirectRT, 2002).

3. Results In general, participants made very few errors. More than 90% (91.2%) of the participants named 5% or fewer of the emotional faces incorrectly. Incorrect responses were excluded from the analyses. To remove outliers, responses latencies less than 333 ms and response latencies greater than 2 SDs above a participants mean response latency were excluded, and if the randomization procedure resulted in the same face being presented on two consecutive trials, the response latency for the second presentation of the face was excluded to eliminate priming eects. To determine whether the response latencies to certain emotional faces diered between males and females, we conducted a 2 (males vs. females; SEX) by 2 (male vs. female facial stimulus; FACE/SEX) by 4 (neutral, happy, fearful vs. angry facial EXPRESSION; within-subject factor) repeated measure ANCOVA with age of participants as the covariate. The results of the analysis

Table 1 Analysis of variance for response time latencies of naming emotional faces Source Between participants Participants Sex (S) Sex of Face (F/S) S F/S Error Within participants Facial expression (E) ES E F/S E F/S S Error Covariate Age E Age Error df 1 1 1 124 3 3 3 3 372 1 3 124 Mean Square 320091.973 51218.879 1274589.832 15500372.245 2411.170 1544.292 11310.484 1445.296 1549.967 21782.520 1431.635 1250003.002 F 2.56 0.41 10.20 p 0.11 0.52 0.00 Partial g2 0.02 0.00 0.08

1.56 1.00 7.30 0.93

0.20 0.40 0.00 0.43

0.01 0.01 0.06 0.01

0.17 0.92

0.68 0.43

0.00 0.01

Note. The sphericity assumption of the test was met (Mauchleys W: 0.92).

S.G. Hofmann et al. / Personality and Individual Dierences 40 (2006) 16831690

1687

are depicted in Table 1.1 The mean and standard errors of the reaction time for all conditions are depicted in Fig. 2. We observed signicant SEX by FACE/SEX and EXPRESSION by FACE/SEX interaction eects. The SEX by FACE/SEX interaction eect showed that male participants were faster than females at naming female faces, whereas females were faster than males at naming male faces (Fig. 3). The EXPRESSION by FACE/SEX interaction eect indicated that participants were faster at naming neutral or happy female faces than neutral or happy male faces. No such dierence was found for fearful and angry faces (Fig. 4). None of the other eects were statistically signicant.2

4. Discussion We expected that naming opposite-sex persons would be faster than naming same-sex persons, regardless of the facial expression. This hypothesis is consistent with an evolutionary perspective, which suggests that individuals attend more strongly to the identity of potential mates. In accordance with this prediction, we observed that males were faster than females at naming female faces, whereas females were faster than males at naming male faces. We further predicted that angry male faces would be generally more quickly named than angry female faces, whereas happy female faces would be more quickly named than happy male faces. Consistent with this sex stereotype hypothesis (Fisher, 2000; Geary, 1998), we found that neutral and happy female faces were more quickly identied than male neutral and happy faces. However, no dierences between male and female faces were found when they displayed fearful or angry emotions. This may suggest that strong negative emotions (fear and anger) dominate the processing of human faces and override the information about the sex of the face. In general, the ndings of this study are consistent with an evolutionary perspective and recent neuropsychological literature that report sex dierences in brain activation when viewing emotional faces (Adolphs, Damasio, Tranel, & Damasio, 1996; Cahill et al., 2001; Canli, Desmond, Zhao, & Gabrieli, 2002; Everhart, Shucard, Quatrin, & Shucard, 2001; Kilgore, Yurgelun, &

1 The clinical literature points to dierences in face processing between depressed and non-depressed individuals and people with and without social anxiety. Therefore, we also administered the Beck Depression Inventory (BDI; Beck, Ward, Mendelso, Mock, & Erbaugh, 1961) and the Social Phobia and Anxiety Inventory (SPAI; Turner, Beidel, Dancu, & Stanley, 1989). The inclusion of the covariates essentially replicated the results of the analyses reported here and showed signicant main eects of SEX (p = 0.03) and EXPRESSION (p = 0.04), and signicant interaction eects of EXPRESSION by FACE/SEX (p = 0.00) and SEX by FACE/SEX (p = 0.00). The BDI covariate eect was signicant, p = 0.001, but not the SPAI eect, p = 0.47. Follow-up analyses showed that higher depression scores were associated with longer response latencies. 2 In order to examine whether this eect was due to dierences in ethnicity among subjects, we also reanalyzed the results with only the 80 Caucasian subsample (40 males and 40 female participants). The pattern of results was replicated. We observed signicant SEX by FACE/SEX, F (1, 75) = 7.22, p = 0.007, g2 = 0.09, and EXPRESSION by FACE/SEX interaction eects, F(3, 73) = 5.42, p = 0.002, g2 = 0.18. None of the other eects were statistically signicant (all ps > 0.05).

1688

S.G. Hofmann et al. / Personality and Individual Dierences 40 (2006) 16831690


Viewing Male Faces
1050 Males Females

1000

Reaction Time (ms)

950

900

850

800

750

700

Ne

utr

al

Ha

pp

y Fe

ar

ful

An

gr

Viewing Female Faces


1050 Males Females

1000

Reaction Time (ms)

950

900

850

800

750

700

Ne

utr

al

Ha

pp

y Fe

ar

ful

An

gr

Fig. 2. Reaction time (mean and standard errors) of male and female participants for naming male faces (upper display) and female faces (lower display) with dierent emotional expressions.

Deborah, 2001). These studies support the notion that males and females dier in the underlying biological mechanism of facial aect processing. The study is limited by the use of a student sample. Assuming that the results can be replicated in non-student samples, we suggest that future studies examine the biological correlates of the sex dierences in information processing biases we reported here. For example, it would be interesting to replicate this paradigm in males and females while collecting neuroimaging or electrophysiological data. Interesting regions of interest would include the fusiform gyrus (e.g., Kanwisher, McDermott, & Chun, 1997) and the amygdala (e.g., Kilgore et al., 2001).

S.G. Hofmann et al. / Personality and Individual Dierences 40 (2006) 16831690

1689

1000

Males Females

Reaction Time (ms)

900

800

700
Male Faces Female Faces

Fig. 3. Reaction time (mean and standard errors) of male and female participants for naming male and female faces, collapsed across the dierent emotional expressions.

1000

Male Faces Female Faces

Reaction Time (ms)

900

800

700
u Ne tra l Ha pp y Fe ar ful An gr y

Fig. 4. Reaction time (mean and standard errors) of participants (collapsed for both sexes) for naming male and female faces with dierent emotional expressions.

References
Adolphs, R., Damasio, H., Tranel, D., & Damasio, A. R. (1996). Cortical systems for the recognition of emotion in facial expressions. Journal of Neuroscience, 16, 76787687. Beck, A. T., Ward, C. H., Mendelso, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561571.

1690

S.G. Hofmann et al. / Personality and Individual Dierences 40 (2006) 16831690

Cahill, L., Haier, R., White, N. S., Fallon, J., Kilpatrick, L., Lawrence, C., Potkin, S. G., & Alkire, M. T. (2001). Sexrelated dierence in amygdala activity during emotionally inuenced memory storage. Neurobiology of Learning and Memory, 75, 19. Canli, T., Desmond, J., Zhao, Z., & Gabrieli, J. (2002). Sex dierences in the neural basis of emotional memories. Proceedings of the National Academy of Science, 99, 1078910794. DirectRT (2002). Computer software. New York, NY: Empirisoft Corporation. Eagly, A. H. (1987). Sex dierences in social behavior: a social-role interpretation. Hillsdale, NJ: Erlbaum. Ekman, P. (1993). Facial expression and emotion. American Psychologist, 48, 384392. Ekman, P., & Friesen, W. (1976). Pictures of facial aect. Palo Alto, VA: Consulting Psychologist Press. Elfenbein, H. A., & Ambady, N. (2003). Universals and cultural dierences in recognizing emotions of a dierent cultural group. Current Directions in Psychological Science, 12, 159164. Everhart, D. E., Shucard, J. L., Quatrin, T., & Shucard, D. W. (2001). Sex-related dierences in event-related potentials, face recognition, and facial aect processing in prepubertal children. Neuropsychology, 15, 329341. Fisher, A. H. (Ed.). (2000). Gender and emotion. Social psychological perspectives. Cambridge, England: Cambridge University Press. Geary, D. C. (1998). Male and female: The evolution of human sex dierences. Washington, DC: American Psychological Association. Goos, L. M., & Silverman, I. (2002). Sex related factors in the perception of threatening facial expressions. Journal of Nonverbal Behavior, 26, 2741. Grossman, W., & Wood, W. (1993). Sex dierences in the intensity of emotional experience: a social role interpretation. Journal of Personality and Social Psychology, 65, 10101022. Guillem, F., & Mograss, M. (2005). Gender dierences in memory processing: Evidence from event-related potentials to faces. Brain and Cognition, 57, 8492. Hess, U., Adams, R. S., & Kleck, R. E. (2004). Facial appearance, gender, and emotion expression. Emotion, 4, 378388. Kanwisher, N., McDermott, J., & Chun, M. M. (1997). The fusiform face are: A module in human extrastrial cortex specialized for face perception. Journal of Neuroscience, 17, 43024311. Kilgore, W. D., Yurgelun, T., & Deborah, A. (2001). Sex dierences in amygdala activation during the perception of facial aect. Neuroreport, 12, 25432547. Shields, S. A. (2002). Speaking from the heart: Gender and the social meaning of emotion. Cambridge, England: Cambridge University Press. Thayer, J. F., & Johnson, B. H. (2000). Sex dierences in judgment of facial aect: A multivariate analysis of recognition errors. Scandinavian Journal of Psychology, 41, 243246. Turner, S. M., Beidel, D. C., Dancu, C. V., & Stanley, M. A. (1989). An empirically derived inventory to measure social fears and anxiety: The Social Phobia and Anxiety Inventory. Psychological Assessment, 1, 3540.

Potrebbero piacerti anche