When Aggressive Individuals See the World More Accurately

Abstract

Previous research has suggested that aggressive individuals exhibit a bias to perceive nonangry expressions as angry. Another line of thinking, however, posits that aggression is a learned response to hostile environments and should be linked to social-cognitive skills suited to such environments. If so, aggressive individuals may exhibit greater perceptual sensitivity to subtle facial cues of anger. Three studies were conducted to test this proposal. In them, participants’ ability to discriminate between subtly different intensities of facial anger was tested. Aggressive participants generally displayed greater perceptual sensitivity to subtle cues of facial anger. This pattern could not be explained in terms of response bias and was specific to angry expressions. The results thus support the idea that aggression is associated with social-cognitive skills rather than bias and ineptitude.

Keywords

aggression facial expressions anger emotion perception signal detection

Imagine you are having a debate with a rather aggressive colleague. The discussion has been mostly civil thus far, but then your colleague makes a subtly rude comment. After noticing your facial expression, your colleague immediately accuses you of becoming angry. You feign ignorance and claim you were not angry at all. Privately, you suspect that your colleague is biased to perceive completely neutral facial expressions as angry. Yet, is it possible that your aggressive colleague was actually correct? Is it possible that he correctly perceived a subtle tinge of anger in your expression, even if you were not fully aware of it yourself?

This scenario contrasts two different views of aggression and facial expression perception. The first perspective suggests that aggressive individuals are biased to perceive anger in others’ facial expressions. By contrast, the second perspective suggests that aggressive individuals are actually highly accurate and capable of identifying even minute expressions of anger. Below, we review research consistent with both perspectives. To resolve inconsistencies in the literature, we suggest that signal detection paradigms are needed. Three studies are presented using such paradigms. Our results consistency favored the idea that aggressive individuals are perceptually sensitive to facial anger, rather than biased.

The Bias Perspective

Researchers have long argued that aggressive individuals exhibit biased perceptions of the social world. Beginning with Dodge (1980), studies have indicated that aggressive individuals tend to interpret the ambiguous actions of others in a hostile manner. The empirical support for this hostile attribution bias is robust. Although originally documented among subclinical aggressive children (Dodge, 1980; see Orobio de Castro, Veerman, Koops, Bosch, & Monshouwer, 2002, for a meta-analytic review), this effect has been demonstrated with juvenile offenders (e.g., Dodge, Price, Bachorowski, & Newman, 1990), subclinical aggressive adults (Dill, Anderson, Anderson, & Deuser, 1997; see Wilkowski & Robinson, 2008, 2010, for relevant reviews), and adult criminal offenders (e.g., James & Seagor, 2006; Walter, 2007).

According to Crick and Dodge’s (1994) social information processing theory, the hostile attribution bias may be due to earlier failures to properly encode social cues. Angry facial expressions are highly relevant in this regard, as they are one of the most well-established nonverbal signals of hostile intent. Both explicit (Knutson, 1996; Yik & Russell, 1999) and implicit (Adams, Ambady, Macrae, & Kleck, 2006; Wilkowski & Meier, 2010) sources of data indicate that people interpret angry expressions as signals of hostile intent.

Building on this perspective, a number of studies have investigated the possibility that aggressive individuals are biased to perceive nonangry facial expressions as angry. Most of these studies focused on misperceptions of expressions other than anger (Barth & Bastiani, 1997; Hall, 2006; Knyazaev, Bocharov, Slobodskaya, & Ryabichenko, 2008; Larkin, Martin, & McClain, 2002), although one study also included expressions displaying a mixture of anger and other emotions (Schultz, Izard, & Bear, 2004). The results have largely supported the claim that aggressive individuals display a biased tendency to perceive these expressions as angry.

Three of these studies also explored the ability to accurately perceive angry expressions as angry. These results have been less consistent, with one study suggesting that aggressive individuals are more accurate in this regard (Barth & Bastiani, 1997) and two other studies showing no significant relation along these lines (Hall, 2006; Larkin et al., 2002). However, these studies concentrated on participants’ perceptions of full, prototypical anger expressions. As reviewed in greater detail below, there is reason to believe that this may be a relatively insensitive measure.

The Sensitivity Perspective

In contrast to the above-presented perspective, several theorists have argued that aggression is not invariably associated with cognitive deficits (Cillessen & Mayeax, 2007; Hawley, 2007; Sutton, Smith, & Swettenham, 1999a). These theorists contend that at least some aggressive individuals exhibit unique cognitive skills that support their aggressive social style (Sutton, Smith, & Swettenham, 1999b; Vaughn, Vollenwider, Bost, Azria-Evans, & Snider, 2003).

Indeed, there are several compelling reasons to believe that aggressive individuals have more exposure to and thus more practice interacting with aggressive environments. Aggressive individuals are often the product of aggressive environments. They frequently emerge from homes characterized by harsh or abusive parenting styles (see Deuter-Deckard & Dodge, 1997; Patterson, DeBaryshe, & Ramsey, 1989, for relevant reviews). Similarly, they are more likely to be characterized by a history of exposure to violent media (see Anderson et al., 2003, for a review). Finally, aggressive individuals also create their own aggressive environment (Anderson, Buckley, & Carnegey, 2008). By engaging in unprovoked aggression, these individuals anger those around them and evoke retaliation against themselves.

Basic psychophysics research indicates that practice trying to perceive a stimulus leads to increased perceptual sensitivity to it (see Goldstone, 1998, for a review). For example, a recent investigation gave participants practice trying to discriminate subliminally presented stimuli that were initially perceived at chance levels of accuracy (Schwiedrzik, Singer, & Melloni, 2009). With adequate practice, these individuals’ perceptual sensitivity to such stimuli increased and surpassed chance accuracy. Importantly, this study found that bias was completely unaffected by practice.

The domain of aggressive stimuli is no different in this regard. Research indicates that individuals immersed in hostile environments become more perceptually sensitive to angry expressions (Pollak & Kistler, 2002; Pollak, Messner, Kistler, & Cohn, 2008; Pollak & Sinha, 2002). These studies specifically compared abused and nonabused children using a paradigm that is arguably more sophisticated than the studies reviewed above. Participants were shown a series of graded facial expressions (e.g., from 0% angry/100% neutral to 100% angry/0% neutral) and asked to identify each expression. Across studies, abused children were able to accurately perceive angry expressions at lower intensities than nonabused children.

Several details concerning these findings are highly relevant here. First, this difference in perceptual sensitivity was confined to moderate-intensity anger expressions. By the time full, prototypical anger is displayed, no differences were apparent between abused and nonabused children (Pollak et al., 2008; Pollak & Kistler, 2002; Pollak & Sinha, 2002). Second, these studies applied signal detection theory algorithms to separate perceptual sensitivity from bias. The results clearly indicate that abused children were more sensitive to facial signals of anger rather than more biased. The current investigation applied a similar paradigm to test the hypothesis that aggressive individuals are more perceptually sensitive to facial anger rather than more biased.

Separating Bias From Accuracy

Previous studies on aggression and facial expression perception have treated perceptions of angry and nonangry facial expressions as completely separate dependent measures. They have specifically focused on the “accurate” perception of angry expressions as angry and the “biased” perception of nonangry faces as angry. As psychophysics researchers long ago noted, such measures of bias and accuracy are not entirely separate (see Macmillan & Creelman, 2005, for an accessible introduction to this theory). The “accurate” perception of a presented stimulus can sometimes be due to a bias to indicate a stimulus is present, regardless of whether it is truly present or not. For example, if participants are rewarded every time they accurately identify a stimulus, they will begin to show an overall response bias toward indicating the stimulus is present (e.g., Levine, 1966).

More relevant to current concerns, the “biased” tendency to indicate that a stimulus is present under degraded conditions can represent increased perceptual sensitivity (e.g., Hecht, Shlaer, & Pirenne, 1941). This is especially true in the context of facial expressions. As the intensity of an expression can vary continuously, some individuals may be able to accurately identify an expression even at moderate or low intensities (Pollak, 2008).

In the current investigation, we employed methods developed by signal detection theorists to better separate accuracy from bias (Macmillan & Creelman, 2005). We modeled our studies after classic psychophysical investigations on perceptual thresholds (e.g., Hecht et al., 1941). In these studies, the intensity of a signal (e.g., the loudness of a sound) is varied, and the participants is asked to indicate when he or she can detect the presented stimulus. On each trial, there are four possible outcomes. First, the participant can successfully identify a stimulus that is present (i.e., a hit). Second, the participant can fail to identify a stimulus that was present (i.e., a miss). Third, the participant can correctly indicate that no stimulus was present (i.e., a correct rejection). Finally, the participant can falsely indicate that a stimulus was present when it was not (i.e., a false alarm).

Within such tasks, a bias toward indicating a stimulus is present is reflected by an increase in both hits and false alarms (Macmillan & Creelman, 2005). This pattern indicates that the participant has an overall tendency to indicate the stimulus is present regardless of whether it is truly present or not. The signal detection index of beta formally quantifies such biases. By contrast, perceptual sensitivity is reflected by an increase in the number of hits but not false alarms. Such a pattern indicates that the participant is correctly detecting an increase in stimulus intensity, rather than simply displaying an overall bias to indicate the signal is present. The signal detection index of d prime formally quantifies perceptual sensitivity.

The Current Studies

In our studies, we varied the intensity of facial signals of anger by presenting participants with a series of facial expressions displaying a mixture of two emotions. The critical faces contained different levels of anger (Study 1: 0% and 50% anger; Studies 2 and 3: 40%, 50%, and 60% anger). We predicted that aggressive individuals would be more sensitive to subtle increases in facial anger. We furthermore predicted that this relation could not be explained in terms of bias and that it would be unique to angry expressions.

Before proceeding, it is useful to elaborate on the precise personality measures and facial expression stimuli used in these studies. Regarding personality measures, we focused on the construct of physical aggression rather than the related constructs of verbal aggression, hostile interpersonal attitudes, or trait anger. The reason for this is simple: Past research has most clearly connected experience with aggressive environments to physical aggression itself. Harsh parenting styles (Deuter-Deckard & Dodge, 1997; Patterson et al., 1989) and violent media exposure (Anderson et al., 2003) both causally increase physical aggression. Similarly, acts of physical aggression have been found to evoke retaliatory responses from others (Anderson et al., 2008). Thus, there is a clear basis in the literature for predicting the physically aggressive individuals will have developed greater perceptual expertise with angry expressions.

By contrast, the evidence that exposure to aggressive environments is linked with the related personality constructs of verbal aggression, hostility, or trait anger is somewhat lacking. Nonetheless, we included measures of these related constructs in the current studies to help explore the generality of effects. We return to these related aspects of personality at the conclusion of the studies.

In regard to facial expression stimuli, we opted for stimuli that provided a great deal of experimental control to properly execute a signal detection paradigm (see Macmillon & Creelman, 2005). Given that real-world facial expressions are extraordinarily complex stimuli involving simultaneous movements of many facial muscles (e.g., Smith & Scott, 1997), we employed computer-generated expressions. These expressions were created using the software package, FaceGen Modeller 3.3, (Singular Inversions, 2004), which allows one to adjust expressions to precisely match an intended intensity (e.g., 50% anger). Stimuli created using this program have been used in multiple investigations (e.g., Oosterhof & Todorov, 2008; Schulte-Ruther, Markowitsch, & Fink, & Piefke, 2007; Todorov, Baron, & Oosterhof, 2008), and participants can recognize such expressions with comparable accuracy to posed facial expressions from standardized stimulus sets (e.g., Schulte-Ruther et al., 2007; see also our own pretest reported below). Thus, stimuli created using this program are both valid and provide adequate experimental control for current purposes.

Study 1

In our first study, participants were presented with a series of faces displaying an even mixture of two facial expressions. These included all possible combinations of anger, fear, sadness, happiness, and neutrality. For each expression, participants were asked to indicate which of the five relevant expressions was displayed. We predicted that physically aggressive individuals would be more capable of discriminating between faces displaying 50% anger and faces displaying no anger whatsoever (i.e., faces displaying a mixture of two other emotions). Furthermore, we predicted that this difference could not be explained in terms of a response bias, and that it would not generalize to perceptions of other expressions.

Method

Participants

Fifty-four undergraduate psychology students from North Dakota State University (29 female, 25 male; mean age = 20) participated in exchange for course credit.

Apparatus

Participants completed all measures on one of six Windows-based computers equipped with a five-button response box. E-Prime software (version 1.2; Schneider, Eschman, & Zuccolott, 2002) was used to administer the facial expression perception task, and MediaLab (Jarvis, 2008) software was used to administer the personality questionnaire.

Stimuli

Graded facial expressions were generated using FaceGen Modeller software (version 3.3; Singular Inversions, 2004). This software uses a large database of scanned faces to generate avatars that are realistic in appearance. The intensity of various emotional expressions can be varied within this program, ranging from 0% to 100%. We used this software to create 50%-50% blends of two expressions. All possible combinations of anger, fear, sadness, happiness, and neutrality were created. See Figure 1 for several example stimuli used in Study 1.

Figure 1.

Example stimuli from Study 1

Twenty exemplars of each combination were generated, resulting in a grand total of 200 facial expressions. Because the features that define facial masculinity are naturally confounded with the features defining facial anger (Becker, Kenrick, Neuberg, Blackwell, & Smith, 2007), faces were constrained to be gender neutral in appearance. Gender neutrality is objectively defined within this program as the midpoint of parameters that distinguish male from female faces within the database of scanned faces. To control for cross-racial disparities in facial recognition accuracy (Elfenbein & Ambady, 2003), faces were constrained to match the racial makeup of the vast majority of our subject pool (i.e., non-Hispanic Caucasian).

Both prior studies (e.g., Schulte-Ruther et al., 2007) and a pretest conducted within our own subject population (n = 7) indicated that unmixed expressions generated using this program can be recognized at levels of accuracy (Schulte-Ruther et al., 2007: 86.4%; our pretest: 81.0%) comparable to those found when using standardized photographs of posed facial expression (see Elfenbein & Ambady, 2003, for a meta-analytic review).

Measures

On each trial, an expression was randomly selected and presented. Participants were asked to indicate which expression they perceived (i.e., anger, fear, sadness, happiness, or neutrality) by pressing the 1, 2, 3, 4, or 5 button on the button box. To avoid making one expression systematically more prominent than others (e.g., by always assigning it to the 1 button), expressions were randomly assigned to different buttons on a trial-by-trial basis. For example, participants may press the 3 button for anger on one trial, but the 5 button for anger on the next trial. On each trial, response mappings were prominently displayed to both the right and left of the face.

Once participants completed the facial expression perception task, they next completed the Aggression Questionnaire (Buss & Perry, 1992). Of foremost interest was the Physical Aggression subscale of this instrument. This scale involves nine statements (e.g., “Once in awhile, I can’t control the urge to hit another person”) indicative of tendencies toward physical aggression. Participants indicate how accurately each statement describes them using a 1 (very inaccurate) to 5 (very accurate) response scale.

The Physical Aggression scale has been repeatedly validated for its ability to predict aggression in both laboratory and field settings (see Bettencourt, Talley, Benjamin, & Valentine, 2006; Bushman & Anderson, 1998, for relevant meta-analytic reviews). Moreover, it exhibits excellent convergent validity with other self-report measures of aggression and discriminate validity with related constructs (Buss & Perry, 1992; Martin, Watson, & Wan, 2000). Finally, this scale exhibited excellent internal validity in the current study (α = .87).

Procedures

Participants arrived at the laboratory and provided informed consent. They then completed the facial expression perception task followed by the Aggression Questionnaire. Tasks were always completed in this order because it is well established that increasing the salience of a personality trait will have effects on subsequent outcomes (Carver, 1979; Harmon-Jones, Lueck, Fearn, & Harmon-Jones, 2006). By contrast, we could locate no evidence suggesting that the completion of a facial expression perception task (or any similar task) can alter one’s responses to a personality questionnaire in a fashion that would confound our results. In fact, evidence indicates that one’s responses to the aggression questionnaire are highly stable over time (e.g., Buss & Perry, 1992).

Results

Anger perceptions

Our first analysis focused on the percentage of trials in which participants perceived anger in an expression. This analysis was conducted using the general linear model (GLM), which allows for the combination of discrete within-subject variables and continuous between-subject variables. Blend (50% anger vs. 0% anger) was entered as a discrete within-subject factor, and participants’ standardized physical aggression scores were entered as a continuous between-subject factor. Because physical aggression was higher among males (M = 2.70) than females (M = 1.83), t(52) = 4.21, p = .0001, gender was entered as a control variable. For all significant effects involving physical aggression, estimated means were calculated for participants low (mean − 1 SD) and high (mean + 1 SD) in physical aggression, as per the well-validated procedures introduced by Aiken and West (1991). Simple slope analyses were also conducted in all studies of this investigation to decompose significant interactions involving physical aggression (Aiken & West, 1991).

Supporting the validity of our manipulation, there was a main effect of blend, such that participants perceived anger more frequently for the 50% anger expressions (M = 47.4%) than for the 0% anger expressions (M = 2.8%), F(1, 51) = 327, p < .0001. Furthermore, there was a main effect of physical aggression, F(1, 51) = 4.99, p = .029, such that participants high in physical aggression indicated faces were angry on a higher percentage of trials (M = 11.0%) than did participants low in physical aggression (M = 9.6%).

However, both of these main effects were qualified by the predicted Blend × Physical Aggression interaction, F(1, 51) = 6.21, p = .016, partial η² = .11. Estimated means for this interaction are displayed in Figure 2. Although individuals low in physical aggression significantly differentiated between the 0% and 50% anger expressions, F(1, 51) = 94.06, p < .0001, individuals high in physical aggression more strongly differentiated between these two blends, F(1, 51) = 195.65, p < .0001.

Figure 2.

Anger perceptions as a function of blend and physical aggression, Study 1

Perceptions of other emotions

To ascertain whether the above effects were specific to perceptions of anger, parallel analyses were conducted for the perception of all other expressions present in this study (i.e., fear, sadness, and happiness). In these analyses, facial expressions were divided between those containing 50% and 0% of the relevant target emotion. Validating our stimuli, all analyses yielded a highly significant main effect of blend, all ps < .0001, such that the target expression was always perceived on a higher proportion of trials for the 50% blends than the 0% blends. More importantly, though, there were no significant effects involving physical aggression across all analyses, all ps > .05.

Signal detection analyses

Our first analysis suggested that participants high in physical aggression were more sensitive to subtle facial cues of anger, in that they more clearly discriminated between 0% and 50% anger expressions. To more definitively determine whether these effects are due to sensitivity or bias, the signal detection indices of d’ and Beta were next calculated (Macmillan & Creelman, 2005). The sensitivity index, d′, reflects a person’s sensitivity to the difference between the 0% and 50% conditions. The bias index, Beta, reflects a person’s bias to indicate a face is displaying a given emotion regardless of its actual expression. These two indices were calculated for all expressions examined in the current study using an Excel worksheet modeled after Table 1.1 in Macmillan and Creelman’s (2005, p. 21) user guide.

At the zero-order level, anger d′ exhibited a marginal positive relation with physical aggression, r = .23, p = .09, and Happy d′ exhibited a marginal negative relation with physical aggression, r = –.23, p = .09. Neither anger Beta (r = .01, p = .96) nor any remaining signal detection index approached a significant relation with physical aggression, all ps > .18. These findings suggest that physical aggression is related to sensitivity to angry facial cues, rather than bias.

Sensitivity scores tended to be positively correlated with each other, all rs > .32, all ps < .02, as might be expected from an emotional intelligence perspective (Mayer, Roberts, & Barsade, 2008). It is possible that a stronger relation between physical aggression and anger sensitivity would be observed when controlling for such overlapping abilities. This proved to be the case. In a multiple regression, all signal detection indices were simultaneously entered as predictors of physical aggression. Gender was added as a control variable. Anger d′ emerged as a significant and positive predictor of physical aggression, β = .52, p = .002. No other signal detection index significantly predicted physical aggression in this analysis, all ps > .10. Further analyses also indicated that gender did not significantly moderate the central finding of interest, p = .65.

Follow-up analysis

Finally, we sought to determine if the above-documented effects were specific to mixtures of anger with certain others expressions or if they were broadly applicable to all mixtures containing anger. In these analyses, we had no a priori hypotheses, but we nonetheless deemed it important to determine any boundary conditions to these effects. Our initial GLM analysis was repeated, this time with secondary expression (fear vs. sadness vs. happiness vs. neutrality) entered as an additional factor in the design.

This analysis yielded a significant Physical Aggression × Anger Blend × Secondary Expression interaction, F(3, 153) = 5.55, p = .001. To decompose this interaction, we conducted separate GLM analyses for each secondary expression. The Physical Aggression × Anger Blend interaction was significant for two secondary expressions, namely, happiness, F(1, 51) = 23.78, p < .0001, and neutrality, F(1, 51) = 4.84, p = .03. In both cases, participants high in physical aggression differentiated between the 0% and 50% anger blends (anger/happy blends: F = 138.22; anger/neutral blends: F = 145.39, all ps < .0001) more strongly than did participants low in physical aggression (anger/happy blends: F = 16.15; anger/neutral blends: F = 68.64, all ps < .0001). For the anger/fear and the anger/sadness blends, the Physical Aggression × Anger Blend interaction was not significant, all ps > .15. Thus, physical aggression was most clearly related to detecting subtle changes in anger when mixed with happiness or emotional neutrality.

Discussion

The results of Study 1 indicated that aggressive individuals are more perceptually sensitive to angry expressions, rather than more biased. Aggressive participants showed a stronger differentiation of 50% and 0% anger, and signal detection analyses confirmed that this was due to perceptual sensitivity rather than bias. Moreover, this results could not be explained in terms of more general-purpose perceptual abilities, as physical aggression was unrelated to the perception of other expressions.

Follow-up analysis also specified the boundary conditions for this effect. The effect emerged for mixtures of anger with happiness or neutrality but not with sadness or fear. It is possible that this is due to the context in which anger perception abilities are learned. Anger is one of the most frequently suppressed facial expressions (Gross, Richards, & John, 2006). Expressive suppression is frequently accomplished by masking anger with a fake non-Duchenne smile (Ekman, Friesen, & O’Sullivan, 1988) or by simply inhibiting the expression altogether (Gross & Levenson, 1993). Aggressive individuals thus may have more practice detecting facial anger against a background of feigned happiness or neutrality. Although the current investigation was not designed to directly test this idea, it does provide a cogent explanation that can be tested in future research.

Study 2

Given the novel nature of the Study 1 findings, we sought to conceptually replicate them in Study 2. In this study, our first goal was to focus on anger/happiness blends specifically. These blends are particularly interesting from an interpersonal perspective, as they contrast expressions of hostile intent (anger) with expressions typically associated with affiliative intent (happiness). Our second goal in conducting this study was to focus on more fine-grained differences in perceptual sensitivity. As such, participants were presented with faces displaying different degrees of ambiguity, namely, 40%, 50%, or 60% anger. The remaining percentage of each facial expression displayed happiness. We considered this a more difficult perceptual task that would more rigorously test participants’ perceptual abilities. We predicted that participants high in physical aggression would be more perceptually sensitive to this difference and that this relation could not be explained in terms of bias.

Method

Participants

Seventy-five undergraduate psychology students from North Dakota State University participated in exchange for course credit.¹

Stimuli

As in Study 1, FaceGen Modeller software (version 3.3) was used to generate mixed facial expressions. These expressions consisted of either 40% anger/60% happiness, 50% anger/50% happiness, or 60% anger/40% happiness. Eighty exemplars of each mixture were initially generated. As in Study 1, these faces were constrained to be non-Hispanic Caucasian and gender neutral in appearance.

Before conducting the main study, we sought to ensure that the manipulation was effective and to select a subset of stimuli for inclusion in the study proper. Six pretest participants were asked to identify the expression of these faces in a two-alternative forced-choice format. The effect of blend on anger perceptions was highly significant in this pretest, p < .0001, with all pairwise contrasts significant in the predicted direction at p < .0001. Pretest data were used to identify 40 exemplars of each blend for use in the main study that ideally matched the intended mixture. See Figure 3 for example stimuli used in Study 2.

Figure 3.

Example stimuli from Study 2

Apparatus, measures, and procedures

All apparatus, measures, and procedures were the same as those used in Study 1, with the following exceptions: Participants saw 40 faces of each blend (i.e., 40% anger/60% happy, 50% anger/50% happy, 60% anger/40% happy). Because there were clearly only two expressions involved in this study, participants were presented with only anger and happiness as response options. Participants pressed the 1 or 5 button of the button box to indicate their perceptions, with response mapping counter-balanced across participants (i.e., 1 = anger/5 = happy for some; 1 = happy/5 = anger for others). The physical aggression scale was once again internally reliable (α = .87).

Results

Anger perceptions

The percentage of trials in which participants perceived anger was first analyzed using the GLM. Blend (40% anger vs. 50% anger vs. 60% anger) was entered as a discrete within-subject factor, and participants’ standardized physical aggression scores were entered as a continuous between-subject variable. The analysis yielded a significant main effect of blend, F(2, 152) = 59.85, p < .0001. Confirming the validity of our manipulation, anger perceptions increased with the manipulated intensity of facial anger (40% anger blends: 33.1% of trials; 50% anger blends: 38.2% of trials; 60% anger blends: 45.5% of trials). There was also a marginal main effect of physical aggression, F(1, 76) = 3.43, p = .069, such that participants high in physical aggression perceived facial expressions to be angry on a higher proportion of trials (M = 41.5%) than did participants low in physical aggression (M = 36.3%).

Both of these main effects were qualified by the predicted Blend × Physical Aggression interaction, F(2, 152) = 3.51, p = .03, partial η² = .044. The estimated means for this interaction are displayed in Figure 4. As can be seen there, individuals high in physical aggression differentiated between the different blends of anger to greater degree, F(2, 152) = 46.46, p < .0001, than did individuals low in physical aggression, F(2, 152) = 17.39, p < .0001.

Figure 4.

Anger perceptions as a function of blend and physical aggression, Study 2

Signal detection analyses

The above analysis suggests that physical aggression is related to sensitivity rather than bias, in that aggressive individuals were more sensitive to slight changes in the percentage of anger displayed. To more definitively establish that physical aggression is related to sensitivity rather than bias, we next calculated the sensitivity index, d′, and the bias index, Beta. Signal detection algorithms are frequently applied in studies using graded stimuli (e.g., sounds varying in volume level; see Macmillan & Creelman, 2005). When this is done, d′ represents perceptual sensitivity to changes in stimulus intensity (i.e., from low-intensity to high-intensity stimuli). To calculate d′ in this manner, low-intensity stimuli are treated as signal-absent trials are in more standard signal detection analyses. We applied this approach to calculate total sensitivity and bias from the 40% anger blend to the 60% anger blends (Macmillan & Creelman, 2005).

At the zero-order level, physical aggression was positively correlated with the d′ sensitivity index, r = .26, p = .02, but not with the Beta bias index, r = .07, p = .48. When both d′ and Beta were simultaneously entered as predictors of physical aggression in a multiple regression, d′ emerged as a significant predictor, β = .27, p = .029, but the bias index, Beta, did not, β = –.04, p = .72.

Discussion

The results of Study 2 provided further evidence that aggressive individuals are more perceptually sensitive to facial anger rather than more biased. In this study, participants were presented with faces displaying subtly different mixtures of anger and happiness. Aggressive participants were more capable of detecting these subtle differences. Furthermore, there was no evidence that such individuals displayed a stimulus-independent bias toward perceiving expressions as angry.

Although Study 2 displayed several strengths, a critic could still argue that these effects are due to an increased ability to detect subtle facial cues of happiness, rather than anger. After all, facial happiness was inversely related to facial anger in this study’s stimuli. Although the results of Study 1 did not support this alternative interpretation, we nonetheless considered it prudent to conduct a third study to better support this claim.

Study 3

Our first goal in designing Study 3 was to explore the relation between physical aggression and the perception of angry/neutral blends. Thus, Study 3 focused on anger/neutral blends. Participants were presented with varying degrees of ambiguous anger, namely, 40%, 50%, or 60%, and the remaining percentage of each facial expression was left emotionally neutral. It was predicted that aggressive individuals would be more perceptually sensitive to these differences and that they would not display a stimulus-independent bias to label expressions as angry.

Beyond this, Study 3 was designed to provide separate sensitivity and bias indices for the perception of facial happiness. In addition to the anger/neutral block, participants also completed a second block in which they indicated their perceptions of happy/neutral blends. We expected that physical aggression would be unrelated to perceptual sensitivity of facial happiness, thus ruling out an alternative explanation of the results in Study 2.

Method

Participants

Eighty-four undergraduate psychology students (61 female, 23 male; mean age = 19.33) from North Dakota State University participated in exchange for course credit.

Stimuli

As in previous studies, FaceGen Modeller software was used to generate graded facial expressions. Faces were generated that displayed 40%, 50%, and 60% of either anger or happiness, with the remaining percentage left emotionally neutral. Eighty exemplars of each blend were initially generated, with the faces constrained to be non-Hispanic Caucasian and gender neutral in appearance.

Before beginning the study, we sought to ensure that the manipulation was successful and to select stimuli for inclusion in the study proper. Six pretest participants were asked to indicate which expression each face displayed in a two-alternative forced-choice task (i.e., anger vs. neutral for the anger/neutral blends; happy vs. neutral for the happy/neutral blends). There was a highly significant effect of blend in this pretest, p < .0001, with all contrasts significant at p < .04 in the predicted direction. These data were used to identify 40 expressions for use in the main study that most ideally matched the intended mixture. Example stimuli from Study 3 are displayed in Figure 5.

Figure 5.

Example stimuli from Study 3

Apparatus, measures, and procedures

All apparatus, measures, and procedures were identical to Study 1, with the following exceptions: Participants completed two blocks of trials, one involving 120 mixtures of anger and neutral expressions and the other involving 120 mixtures of happy and neutral expressions. Forty exemplars of each blend (i.e., 40% emotional, 50% emotional, 60% emotional) were included in each block. Only the relevant facial expressions were presented as response options within each block (i.e., anger/neutral or happy/neutral). Participants pressed the 1 or 5 button of the button box to indicate their perceptions. The order in which participants completed the blocks and the buttons to which expressions were assigned were separately counterbalanced across participants. As in prior studies, the Physical Aggression scale was internally reliable (α = .81).

Results

Emotion perceptions

We first analyzed the percentage of trials in which participants endorsed the emotional facial expression option (i.e., anger in the anger/neutral block and happiness in the happy/neutral block), using a GLM analysis. Block type (anger/neutral vs. happy/neutral) and blend (40% emotional vs. 50% emotional vs. 60% emotional) were entered as discrete within-subject factors, and participants’ standardized physical aggression scores were entered as a continuous between-subject factor. As in Study 1, males were higher in physical aggression (M = 2.71) than females (M = 1.77), t(82) = 6.12, p < .0001. Thus, gender was entered as a control variable.

This analysis revealed a significant main effect of blend, F(2, 162) = 207.84, p < .0001. Validating our manipulation, faces were endorsed as displaying an emotion (i.e., anger or happiness) on a higher proportion of trials as the intensity of the displayed emotion increased (40% emotion blends: 33.8% of trials; 50% emotion blends: 46.0% of trials; 60% emotion conditions: 59.7% of trials). There was also a main effect of block type, F(1, 81) = 50.91, p < .0001, such that emotion perceptions were more frequent in the happy/neutral block (M = 54.0%) than the angry/neutral block (M = 39.1%). These main effects were qualified by a Blend × Block Type interaction, F(2, 162) = 32.53, p < .0001. This interaction appeared to indicate that the block type main effect was more robust for the more intense blends (60% blends: M Diff = 23.5%; 50% blends: M Diff = 15.9%; 40% blends: M Diff = 5.1%).

More importantly, though, the predicted Blend × Block Type × Physical Aggression interaction was significant, F(2, 162) = 4.13, p = .018, partial η² = .048. The estimated means for this effect are displayed in Figure 6 (top panel: anger/neutral block; bottom panel: happy/neutral block). To decompose this interaction, we next analyzed each block in separate GLM analyses. Within the analysis of happy/neutral block, there were no significant effects involving physical aggression, all ps > .17. Within the analysis of the angry/neutral block, however, there was a significant Blend × Physical Aggression interaction, F(2, 162) = 4.42, p = .013. Although participants low in physical aggression successfully differentiated between the blends of anger, F(2, 162) = 25.20, p < .0001, participants high in physical aggression differentiated between these blends more strongly, F(2, 162) = 61.76, p < .0001. Thus, individuals high in physical aggression more strongly differentiated between different degrees of facial anger but not facial happiness.

Figure 6.

Perceptions of anger (top panel) and happiness (bottom panel) as a function of blend and physical aggression, Study 3

Signal detection analysis

The above analysis suggests that individuals high in physical aggression are more perceptually sensitive to facial anger. Moreover, it suggests that these effects cannot be explained in terms of response bias or perceptual sensitivity to happy expressions. To more definitively support these conclusions, we next calculated the d′ sensitivity and Beta bias indices for both angry and happy expressions separately considered. As in Study 2, these estimates reflected total perceptual sensitivity to changes in the intensity of an expression (i.e., from the 40% to the 60% emotion blends; Macmillan & Creelman, 2005).

At the zero-order level, physical aggression was positively correlated with anger d′, r = .29, p = .007, but not anger Beta, happy d’, or happy Beta, all ps > .40. To test the unique relation between physical aggression and anger d’, we next conducted a multiple regression. All signal detection indices were simultaneously entered as predictors of physical aggression, and gender was entered as a control variable. Anger d′ significantly predicted physical aggression, β = .28, p = .009, but no other signal detection index emerged as a significant predictor, all ps > .12. Additional analyses also indicated that neither gender, p = .62, nor counterbalancing factors, p = .94, significantly moderated the central finding of interest.

Discussion

Study 3 provided further evidence that aggressive individuals are more perceptually sensitive to facial anger. In this study, participants completed separate tasks requiring them to discriminate between different mixtures of facial anger and facial happiness. Aggressive individuals were more capable of discriminating different degrees of facial anger. Moreover, they did not demonstrate an overall bias toward the perception of facial anger or increased perceptual sensitivity to facial happiness. Thus, this study clearly links physical aggression with perceptual sensitivity to facial anger.

Related Aspects of Personality

Before proceeding to the General Discussion, it is important to consider whether the effects documented above are specific to physical aggression or generalize to related aspects of personality. The Aggression Questionnaire (Buss & Perry, 1992) used in the current investigation also contains subscales measuring individual differences in verbal aggression, hostile interpersonal attitudes, and angry emotions. Psychometric research has also confirmed that there are separable affective (i.e., anger), behavioral (i.e., aggression), and cognitive (i.e., hostility) components of trait aggression (e.g., Martin et al., 2000).

Based on our literature review, it was apparent that individuals high in physical aggression have had greater exposure to hostile environments (Anderson et al., 2003; Anderson et al., 2008; Patterson et al., 1989). Thus, there was a strong empirical foundation for predicting that such individuals have developed superior expertise in the perception of angry expressions. The related personality constructs of verbal aggression, hostility, and trait anger, by contrast, have been less heavily researched in this regard, and thus the empirical basis for a parallel prediction with these traits was more questionable. Nonetheless, we measured these aspects of personality across all studies (all αs > .77) to explore the potential generality of these effects.

Here, we report the results of analyses focusing on these related aspects of personality. In the interests of brevity, we concentrate on the most definitive analysis reported above for physical aggression, namely, the signal detection regression analyses. In these analyses, all signal detection indices available within a given study were simultaneously entered as predictors of the relevant personality construct in a multiple regression analyses. Gender was added as a covariate when available. These analyses were conducted separately for verbal aggression, hostility, and trait anger.

In Study 1, anger d′ significantly predicted verbal aggression (β = .40, p = .03), hostility (β = .46, p = .01), and trait anger (β = .59, p = .01). Neither anger Beta nor any other signal detection index consistently and significantly predicted these personality constructs. In Study 2, however, anger d′ was unrelated to verbal aggression (β = .01, p = .94), hostility (β = .12, p = .33), and trait anger (β = .05, p = .70). Anger Beta was similarly unrelated to these personality constructs. In Study 3, anger d′ was positively related to verbal aggression (β = .24, p = .051), hostility (β = .22, p = .08), and trait anger (β = .29, p = .02), although several of these relations were only marginally significant in magnitude (p < .10). Neither anger Beta nor any other signal detection index significantly predicted these personality constructs, all ps > .10.

Thus, perceptual sensitivity to facial anger was more consistently related to physical aggression relative to verbal aggression, hostility, and trait anger. There are multiple possible reasons for this finding. From an evocative standpoint (Anderson et al., 2008), physical aggressive behaviors would clearly evoke anger in others. Anger and hostility sometimes remain unexpressed (e.g., Gross & Levenson, 1993), and thus an individual’s tendencies toward anger and hostility would not evoke anger in others if they are not manifested in outward aggression. Verbal aggression may also not consistently evoke anger in others, as arguments are often viewed as helpful, but spirited, exchanges of information (e.g., Averill, 1983). Thus, physical aggression’s consistent relation with facial expression perception may be due to the increased evocation of anger in others, and thus to increased practice perceiving this expression.

Beyond this, it is also true that physical aggression alone necessitates face-to-face communication. Verbal aggression can take an indirect form (i.e., relationally aggressive acts, such as behind the back rumor spreading; Archer & Coyne, 2005), and anger and hostility can often be the result of prolonged rumination rather than immediate face-to-face reactions (see Wilkowski & Robinson, 2008). Thus, physical aggression may also be more directly related to facial expression perception simply because of its greater relevance to face-to-face interactions.

General Discussion

Summary of Hypotheses and Results

Historically, physical aggression has been seen as the result of cognitive biases and social ineptitude (e.g., Crick & Dodge, 1994, 1999). More recently, though, other theorists have argued that at least some aggressive individuals have developed unique cognitive skills that support their aggressive tactics (Cillessen & Mayeax, 2007; Hawley, 2007; Sutton et al., 1999a). Here, we hypothesized that through their increased exposure to aggressive environments (Anderson et al., 2003; Anderson et al., 2008; Patterson et al., 1989), physically aggressive individuals had developed a greater expertise in the perception of angry expressions. This implies a reinterpretation of previous findings suggesting that these individuals are biased to perceive nonangry facial expressions as angry (Barth & Bastiani, 1997; Hall, 2006; Knyazaev et al., 2008; Larkin et al., 2002; Schultz et al., 2004).

Three studies were conducted to test this hypothesis. These studies were based on classic signal detection studies of perceptual thresholds (e.g., Hecht et al., 1941). In them, participants were presented with faces displaying subtly different levels of anger (Study 1: 0% & 50%; Studies 2-3: 40%, 50%, & 60%), as well as other facial expressions. Sensitivity was measured as increases in perceptions of anger that covaried with the actual increase in displayed anger, and bias was measured as a stimulus-independent tendency to label facial expressions as angry (Macmillon & Creelman, 2005).

Across all studies, physically aggressive individuals were more sensitive to subtle differences in facial anger. This finding was demonstrated with two separate blends, namely, angry/happy blends (Studies 1 and 2) and angry/neutral blends (Studies 1 and 3). Moreover, this finding could not be explained in terms of bias or broader perceptual abilities, as physical aggression was consistently unrelated to signal detection indices of anger-related bias or perceptual sensitivity for other expressions.

Future research should continue to test these predictions in several fashions. The ecological validity of this phenomenon should also be tested using photographed facial expressions of real people. Beyond this, the applicability of this phenomenon to extreme levels of physical aggression (e.g., criminal offenders) should be tested. Presumably, this perceptual ability would be even more apparent among extremely aggressive individuals.

Are Aggressive Individuals Biased to Perceive Anger?

It is intuitive to suggest that aggressive individuals are prone to misperceive anger in others’ expressions (Hall, 2006). A number of studies have claimed to support biases of this type (e.g., Knyazaev et al., 2008; Larkin et al., 2002), but the procedures used were not ideal. In particular, such investigations have rarely compared people’s anger perceptions across different degrees of displayed anger. Without modeling such within-subject factors, we cannot be sure whether anger sensitivity, anger perception biases, or both are involved.

These methodological features are particularly problematic given that several theorists have argued that aggression may sometimes be associated with cognitive skills rather than cognitive incompetence (e.g., Cillessen & Mayeax, 2007; Hawley, 2007; Sutton et al., 1999a). In the present studies, we included the percentage of displayed anger as a within-subject factor in the design. Higher levels of physical aggression interacted with such within-subject manipulations in all studies, suggesting greater sensitivity to anger rather than greater bias. At least among adults, then, our results suggest that it is a mistake to suggest that aggressive individuals are biased to perceive anger. Instead, they are highly sensitive to perceiving subtle facial cues of anger.

On the basis of our results, we suggest a reinterpretation of the hostile attribution bias. This bias is such that aggressive children and adults are more likely to perceive hostile intent when such intent is ambiguous (Dill et al., 1997; Dodge et al., 1990; Epps & Kendall, 1995). Under such conditions, it may be problematic to infer that hostile attributions represent a bias. Instead, it is entirely possible that aggressive individuals are more sensitive to hostile cues in such situations and make their judgments accordingly. In other words, what appears to be a bias may instead be represent greater sensitivity to the hostile cues which are truly present. Support for this reinterpretation of the hostile attribution bias seems consistent with data showing that aggressive individuals are generally not more likely to attribute hostile intent in unambiguous situations (e.g., Dodge, 1980).

The Function of Perceptual Sensitivity

Learning studies have shown that practice perceiving a given stimulus increases a person’s perceptual sensitivity to that stimulus (Goldstone, 1998). From one perspective, our findings can be viewed in terms of this general principle. Aggressive individuals emerge from hostile environments (Anderson et al., 2003; Patterson et al., 1989) and act in ways that make other people angry (Anderson et al., 2008). As a result, they are likely to develop perceptual expertise in recognizing gradations in facial anger. Such ideas are consistent with findings that children exposed to abusive environments subsequently develop better anger perception skills (Pollak, 2008).

At a more basic level, though, one must ask why experience promotes perceptual learning at all. Presumably, human beings evolved the ability to learn from experience because it is functional (Pigliucci, 2001; Pulliam & Dunford, 1980). We learn to make finer perceptual distinctions because this prepares us to handle similar events in the future. From this perspective, then, learning to discriminate subtle facial cues of anger serves an adaptive function for aggressive individuals. This view stands in marked contrast to the classic social information processing perspective that aggression and the cognitive processing tendencies that support it are maladaptive in all respects (Crick & Dodge, 1994, 1999).

Although aggression is viewed as socially undesirable in modern society, it is nonetheless true that aggression is a viable strategy from an evolutionary perspective (Hawley, 2007). There are certainly trade-offs here, as aggression is likely to harm one’s relationships. Nonetheless, aggression is a mechanism that can be used to gain social status and access to scarce resources (Hawley, 2007; Maynard-Smith, 1974) and to deter victimization (Clutton-Brock & Parker, 1995).

In the present studies, we showed that aggressive individuals display greater perceptual sensitivity to facial anger. We suggest that this perceptual skill should support an aggressive interpersonal strategy. If so, greater anger sensitivity may predict some of the purported benefits of aggression, such as achieving status (Hawley, 2007) and deterring victimization (Clutton-Brock & Parker, 1995). By accurately perceiving hostile intent in one’s environment, aggressive individuals can better direct their aggressive actions in a fashion that achieves these evolutionarily adaptive ends.

Conclusion

Aggression has traditionally been seen as the result of cognitive biases and social ineptitude (Crick & Dodge, 1994, 1999). The current studies demonstrate that what may appear to be a bias is actually a finely tuned skill. Physically aggressive individuals were more perceptually sensitive to facial anger, rather than more biased. Such findings suggest that the cognitive characteristics of aggressive individuals may at least sometimes be indicative of a Machiavellian brand of social skill, rather than social incompetence.

Footnotes

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Adams

R. B.

Ambady

Macrae

C. N.

Kleck

R. E.

(2006). Emotional expressions forecast approach-avoidance behavior. Motivation and Emotion, 30, 179-188.

Aiken

L. S.

West

S. G.

(1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.

Anderson

C. A.

Berkowitz

Donnerstein

Huesmann

L. R.

Johnson

J. D.

Linz

. . . Wartella

(2003). The influence of media violence on youth. Psychological Science in the Public Interest, 4, 81-110.

Anderson

C. A.

Buckley

K. E.

Carnegey

N. L.

(2008). Creating your own hostile environment: A laboratory examination of trait aggressiveness and the violence escalation cycle. Personality and Social Psychology Bulletin, 34, 462-473.

Archer

Coyne

S. M.

(2005). An integrated review of indirect, relational, and social aggression. Personality and Social Psychology Review, 9, 212-230.

Averill

J.R.

(1983). Studies on anger and aggression: Implications for theories of emotion. American Psychologist, 38, 1145-1160.

Barth

J. M.

Bastiani

(1997). A longitudinal study of emotion recognition and preschool children’s social behavior, Merrill-Palmer Quarterly, 43, 107-128.

Becker

D. V.

Kenrick

D. T.

Neuberg

S. L.

Blackwell

K. C.

Smith

D. M.

(2007). The confounded nature of angry men and happy women. Journal of Personality and Social Psychology, 92, 179-190.

Bettencourt

B. A.

Talley

Benjamin

A. J.

Valentine

(2006). Personality and aggressive behavior under provoking and neutral conditions: A meta-analytic review. Psychological Bulletin, 132, 751-777.

10.

Bushman

B. J.

Anderson

C. A.

(1998). Methodology in the study of aggression: Integrating experimental and nonexperimental findings. In Geen

R. G.

Donnerstein

(Eds.), Human aggression: Theories, research, and implications for social policy (pp. 23-48). San Diego, CA: Academic Press.

11.

Buss

A. H.

Perry

(1992). The aggression questionnaire. Journal of Personality and Social Psychology, 63, 452-459.

12.

Carver

C. D.

(1979). A cybernetic model of self-attention processes. Journal of Personality and Social Psychology, 37, 1251-1281.

13.

Cillessen

C. H. N.

Mayeax

(2007). Variations in the association between aggression and social status: Theoretical and empirical perspectives. In Hawley

P. H.

Little

T. D.

Rodkin

P. C.

(Eds.), Aggression and adaptation: The bright side to bad behavior (pp. 135-156). Mahwah, NJ: Erlbaum.

14.

Clutton-Brock

T. J.

Parker

G. A.

(1995). Punishment in animal societies. Nature, 373, 209-216.

15.

Crick

N. R.

Dodge

K. A.

(1994). A review and reformulation of social information processing mechanisms in children’s social adjustment. Psychological Bulletin, 115, 74-101.

16.

Crick

N. R.

Dodge

K. A.

(1999). “Superiority” is in the eye of the beholder: A comment of Sutton, Smith and Swettenham. Social Development, 8, 128-131.

17.

Deuter-Deckard

Dodge

K. A.

(1997). Externalizing behavior problems and discipline revisited: Nonlinear effects and variation by culture, context, and gender. Psychological Inquiry, 8, 161-175.

18.

Dill

K. E.

Anderson

C. A.

Anderson

K. B.

Deuser

W. E.

(1997). Effects of aggressive personality on social expectations and social perceptions. Journal of Research in Personality, 31, 272-292.

19.

Dodge

K. A.

(1980). Social cognition and children’s aggressive behavior. Child Development, 51, 162-170.

20.

Dodge

K. A.

Price

J. M.

Bachorowski

Newman

J. P.

(1990). Hostile attributional biases in severely aggressive adolescents. Journal of Abnormal Psychology, 99, 385-392.

21.

Ekman

Friesen

W. V.

O’Sullivan

(1988). Smiles when lying. Journal of Personality and Social Psychology, 54, 414-420.

22.

Elfenbein

H. A.

Ambady

(2003). On the universality and cultural specificity of emotion recognition: A meta-analysis. Psychological Bulletin, 128, 203-235.

23.

Epps

Kendall

P. C.

(1995). Hostile attribution bias in adults. Cognitive Therapy and Research, 19, 159-178.

24.

Goldstone

R. L.

(1998). Perceptual learning. Annual Review of Psychology, 49, 585-612.

25.

Gross

J. J.

Levenson

R. W.

(1993) Emotional suppression: Physiology, self-report, and expressive behavior. Journal of Personality and Social Psychology, 64, 970-986.

26.

Gross

J. J.

Richards

J. M.

John

O. P.

(2006). Emotion regulation in everyday life. In Snyder

D. K.

Simpson

Hughes

J. N.

(Eds.), Emotion regulation in couples and families: Pathways to dysfunction and health (pp. 13-35). Washington, DC: American Psychological Association.

27.

Hall

C. W.

(2006). Self-reported aggression and the perception of anger in facial expression photos. Journal of Psychology, 140, 255-267.

28.

Harmon-Jones

Lueck

Fearn

Harmon-Jones

(2006). The effect of personal relevance and approach-related action expectation on relative left frontal cortical activity. Psychological Science, 17, 434-440.

29.

Hawley

P. H.

(2007). Social dominance in childhood and adolescence: Why social competence and aggression may go hand in hand. In Hawley

P. H.

Little

T. D.

Rodkin

P. C.

(Eds.), Aggression and adaptation: The bright side of bad behavior (pp. 1-30). Mahwah, NJ: Erlbaum.

30.

Hecht

Shlaer

Pirenne

M. H.

(1941). Energy at the threshold of vision. Science, 93, 585-587.

31.

James

Seager

J. A.

(2006). Impulsivity and schemas for a hostile world: Postdictors of violent behaviour. International Journal of Offender Therapy and Comparative Criminology, 50, 47-56.

32.

Jarvis

B. G.

(2008). MediaLab (Version 2008) [Computer Software]. New York, NY: Empirisoft Corporation.

33.

Knutson

(1996). Facial expressions of emotion influence interpersonal trait inferences. Journal of Nonverbal Behavior, 20, 165-182.

34.

Knyazaev

G. G.

Bocharov

A. V.

Slobodskaya

H. R.

Ryabichenko

T. I.

(2008). Personality-linked biases in perception of emotional facial expressions. Personality and Individual Differences, 44, 1093-1104.

35.

Larkin

K. T.

Martin

R. R.

McClain

S. E.

(2002). Cynical hostility and the accuracy of decoding facial expressions of emotions. Journal of Behavioral Medicine, 25, 285-292.

36.

Levine

J. M.

(1966). The effects of values and costs on the detection and identification of signals in auditory vigilance. Human Factors, 8, 525-537.

37.

Macmillan

N. A.

Creelman

C. D.

(2005). Detection theory: A user’s guide (2nd ed.). Mahwah, NJ: Erlbaum.

38.

Martin

Watson

Wan

C. K.

(2000). A three-factor model of trait anger: Dimensions of affect, behavior, and cognition. Journal of Personality, 68, 869-897.

39.

Mayer

J. D.

Roberts

R. R.

Barsade

S. G.

(2008). Human abilities: Emotional intelligence. Annual Review of Psychology, 59, 507-536.

40.

Maynard-Smith

(1974). The theory of games and the evolution of animal conflicts. Journal of Theoretical Biology, 47, 209-221.

41.

Oosterhof

N. N.

Todorov

(2008). The functional basis of face evaluation. Proceedings of the National Academy of Science, 105, 11087-11092.

42.

Orobio de Castro

Veerman

J. W.

Koops

Bosch

J. D.

Monshouwer

H. J.

(2002). Hostile attribution of intent and aggressive behavior: A meta-analysis. Child Development, 73, 916-934.

43.

Patterson

G. R.

DeBaryshe

B. D.

Ramsey

(1989). A developmental perspective on antisocial behavior. American Psychologist, 44, 329-335.

44.

Pigliucci

(2001). Phenotypic plasticity: Beyond nature and nurture. Baltimore, MD: John Hopkins University Press.

45.

Pollak

S. D.

(2008). Mechanisms linking early experience and the emergence of emotions: Illustrations from the study of maltreated children. Current Directions in Psychological Science, 17, 370-375.

46.

Pollak

S. D.

Kistler

(2002). Early experience is associated with the development of categorical representations for facial expressions of emotion. Proceedings of the National Academy of Science, 99, 9072-9076.

47.

Pollak

S. D.

Messner

Kistler

D. J.

Cohn

J. F.

(2008). Development of perceptual expertise in emotion recognition. Cognition, 110, 242-247.

48.

Pollak

S. D.

Sinha

(2002). Effects of early experience on children’s recognition of facial displays of emotion. Developmental Psychology, 38, 784-791.

49.

Pulliam

H. R.

Dunford

(1980). Programmed to learn: An essay on the evolution of culture. New York, NY: Columbia University Press.

50.

Schneider

Eschman

Zuccolotto

(2002). E-Prime (Version 1.2) [Computer Software]. Pittsburgh, PA: Psychology Software Tools.

51.

Schulte-Ruther

Markowitsch

H. J.

Fink

G. R.

Piefke

(2007). Mirror neuron and theory of mind mechanisms involved in face-to-face interactions: A functional magnetic resonance imaging approach to empathy. Journal of Cognitive Neuroscience, 19, 1354-1372.

52.

Schultz

Izard

C. E.

Bear

(2004). Children’s emotion processing: Relations to emotionality and aggression. Development and Psychopathology, 16, 371-387.

53.

Schwiedrzik

C. M.

Singer

Melloni

(2009). Sensitivity and perceptual awareness increase with practice in metacontrast masking. Journal of Vision, 9, 1-18.

54.

Singular Inversions. (2004). FaceGen Modeller (Version 3.3) [Computer Software]. Toronto, ON: Singular Inversions.

55.

Smith

C. A.

Scott

H. S.

(1997). A componential approach to the meaning of facial expressions. In Russell

J. A.

Fernandez-Dols

J. M.

(Eds.), The psychology of facial expression (pp. 229-254). Cambridge, UK: Cambridge University Press.

56.

Sutton

Smith

P. K.

Swettenham

(1999a). Bullying and “theory of mind”: A critique of the “social skills deficit” view of anti-social behavior. Social Development, 8, 117-127.

57.

Sutton

Smith

P. K.

Swettenham

(1999b). Social cognition and bullying: Social inadequacy or skilled manipulation? British Journal of Developmental Psychology, 17, 435-450.

58.

Todorov

Baron

S. G.

Oosterhof

N. N.

(2008). Evaluating face trustworthiness: A model based approach. Scan, 3, 119-127.

59.

Vaughn

B. E.

Vollenwider

Bost

K. K.

Azria-Evans

M. R.

Snider

J. B.

(2003). Negative interactions and social competence for preschool children in two samples: Reconsidering the interpretation of aggressive behavior for young children. Merrill-Palmer Quarterly, 49, 245-278.

60.

Walters

G. D.

(2007). Measuring proactive and reactive criminal thinking with the PICTS: Correlations with outcome expectancies and hostile attribution biases. Journal of Interpersonal Violence, 22, 371-385.

61.

Wilkowski

B. M.

Meier

B. P.

(2010). Bring it on: Angry facial expressions potentiate approach-motivated motor behavior. Journal of Personality and Social Psychology, 98, 201-210.

62.

Wilkowski

B. M.

Robinson

M. D.

(2008). The cognitive basis of trait anger and reactive aggression: An integrative analysis. Personality and Social Psychology Bulletin, 12, 3-21.

63.

Wilkowski

B. M.

Robinson

M. D.

(2010). The anatomy of anger: An integrative cognitive model of trait anger and reactive aggression. Journal of Personality, 78, 9-38.

64.

Yik

M. S. M.

Russell

J. A.

(1999). Interpretation of faces: A cross-cultural study of a prediction from Fridlund’s theory. Cognition and Emotion, 13, 93-104.