Psychological Science

2 downloads 0 Views 821KB Size Report
Mar 26, 2013 - Marcus R. Munafò, School of Experimental Psychology, University of. Bristol .... Positive and Negative Affect Schedule (PANAS; Watson, Clark, ...
Psychological Science http://pss.sagepub.com/

Increasing Recognition of Happiness in Ambiguous Facial Expressions Reduces Anger and Aggressive Behavior Ian S. Penton-Voak, Jamie Thomas, Suzanne H. Gage, Mary McMurran, Sarah McDonald and Marcus R. Munafò Psychological Science published online 26 March 2013 DOI: 10.1177/0956797612459657 The online version of this article can be found at: http://pss.sagepub.com/content/early/2013/03/25/0956797612459657

Published by: http://www.sagepublications.com

On behalf of:

Association for Psychological Science

Additional services and information for Psychological Science can be found at: Email Alerts: http://pss.sagepub.com/cgi/alerts Subscriptions: http://pss.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav

>> OnlineFirst Version of Record - Mar 26, 2013 What is This?

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013

459657 research-article2013

PSSXXX10.1177/0956797612459657Penton-Voak et al.Emotion Recognition and Aggressive Behavior

Psychological Science OnlineFirst, published on March 26, 2013 as doi:10.1177/0956797612459657

Research Article

Increasing Recognition of Happiness in Ambiguous Facial Expressions Reduces Anger and Aggressive Behavior

Psychological Science XX(X) 1­–10 © The Author(s) 2013 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0956797612459657 pss.sagepub.com

Ian S. Penton-Voak1, Jamie Thomas2, Suzanne H. Gage1, Mary McMurran3, Sarah McDonald1, and Marcus R. Munafò1 1

School of Experimental Psychology, University of Bristol; 2School of Health Sciences, University of Wales Institute, Cardiff; and 3Institute of Mental Health, University of Nottingham

Abstract The ability to identify emotion in other people is critical to social functioning. In a series of experiments, we explored the relationship between recognition of emotion in ambiguous facial expressions and aggressive thoughts and behavior, both in healthy adults and in adolescent youth at high risk of criminal offending and delinquency. We show that it is possible to experimentally modify biases in emotion recognition to encourage the perception of happiness over anger in ambiguous expressions. This change in perception results in a decrease in self-reported anger and aggression in healthy adults and high-risk youth, respectively, and also in independently rated aggressive behavior in high-risk youth. We obtained similar effects on mood using two different techniques to modify biases in emotion perception (feedback-based training and visual adaptation). These studies provide strong evidence that emotion processing plays a causal role in anger and the maintenance of aggressive behavior. Keywords emotions, aggressive behavior, facial expressions Received 3/6/12; Revision accepted 8/1/12

Facial expressions of emotion are a particularly salient cue for interpreting social behavior and subsequently deciding on appropriate courses of action. Recent work suggests that deficits in the ability to detect emotion in other people may contribute to the development and maintenance of anger and aggressive behavior. For example, a tendency to interpret ambiguous emotional expressions as negative rather than positive (i.e., a negativity bias) may result in inappropriate, potentially aggressive, behavior (Dodge, 1993). Aggressive individuals interpret a variety of ambiguous social cues as hostile (the hostile-attribution bias; Nasby, Hayden, & Depaulo, 1980), and it has been argued that such biases may be causal in the development of aggressive behavior (Crick & Dodge, 1994). Intermittent explosive disorder, conduct disorder, and antisocial behavior in children, adolescents, and adults are all associated with deficits in perception of emotional expression (Best, Williams, & Coccaro, 2002; Bowen & Dixon, 2010; Fairchild, Van Goozen, Calder,

Stollery, & Goodyer, 2009; Sato, Uono, Matsuura, & Toichi, 2009). Such studies commonly find that neutral and ambiguous facial expressions are interpreted more negatively by aggressive individuals than by others, a pattern consistent with the operation of a hostile-attribution bias. Such biases may be self-reinforcing: Misattributing hostile intent in others may in fact elicit hostile behavior from previously unaggressive individuals, a self-fulfilling prophecy that may lead to a vicious cycle of aggressive attributions, expectations, and responses. Thus, biases in emotional processing may form an important component in the maintenance of aggressive behavior.

Corresponding Author: Marcus R. Munafò, School of Experimental Psychology, University of Bristol, 12a Priory Rd., Bristol BS8 1TU, United Kingdom E-mail: [email protected]

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013

Penton-Voak et al.

2 Some researchers have attempted to correct deficits in emotion processing, particularly in young people with behavioral problems. For example, Dadds et al. (2006) showed that children with psychopathic traits had an impairment in fear recognition, but that this deficit could be corrected simply by directing the young people to pay attention to the eyes of facial stimuli depicting emotion. More recently, Dadds, Cauchi, Wimalaweera, Hawes, and Brennan (2012) investigated whether an emotionrecognition training task could be an effective intervention in a sample of young people with diverse behavioral difficulties. Although there was no beneficial effect of this training on the group as a whole, participants with callous-unemotional traits showed significant improvements in conduct in response to the training. It is also possible that biases in emotional processing are causal in the development of psychopathology. Recent work on cognitive neuropsychological models of antidepressant drugs’ action indicates that antidepressants do not enhance mood directly, but rather have their effect via changes in emotional processing (Harmer, Goodwin, & Cowen, 2009). The modification of biases in emotional processing through behavioral techniques has effects on mood consistent with this model (PentonVoak, Bate, Lewis, & Munafò, 2012). In a series of experiments, we sought to investigate whether analogous processes may occur in the domains of anger and aggression. We experimentally manipulated biases in facial emotion recognition to elucidate the role that the identification of emotion may play in anger and aggressive behavior.

population, through advertisements on and around the University Precinct at the University of Bristol (Table 1). Participants were randomly assigned to receive an emotion-recognition modification procedure or a control procedure. Materials. Using established techniques (Tiddeman, Burt, & Perrett, 2001), we generated prototypical happy and angry composite images from 20 individual male faces showing a happy facial expression and the same 20 individuals showing an angry expression. The original images came from the Karolinska Directed Emotional Faces (Lundqvist, Flykt, & Öhman, 1998). These prototypical images were used as endpoints to generate a linear morph sequence that consisted of images that changed incrementally from unambiguously happy to unambiguously angry, with emotionally ambiguous images in the middle. This sequence had 15 equally spaced images that we used as experimental stimuli (see Fig. 1 for the endpoints and examples of intermediate stimuli). The State Anger subscale of the State-Trait Anger Expression Inventory-2 (STAXI-2; Spielberger, 1999) was used to record self-reported state anger. The Positive and Negative Affect Schedule (PANAS) was used to assess participants’ positive and negative affect (Watson, Clark, & Tellegen, 1988). Procedure. Testing took place within a single session. Each session consisted of three phases: baseline, training, and test. The baseline and test phases were identical and consisted of 45 trials, in which each of the stimuli from the morph sequence was presented three times. Images were presented, in random order, for 150 ms, preceded by a fixation cross (1,500–2,500 ms, randomly jittered). Stimulus presentation was followed by a mask of visual noise (150 ms) and then a prompt asking the participant

Experiment 1 Method Participants. We recruited 40 healthy volunteers (14 male, 26 female; ages 18–30 years) from the general

Table 1.  Descriptive Statistics for Experiments 1 and 3 Experiment 1 Measure Age (years) STAXI-2 PANAS positive affect PANAS negative affect

Control training (n = 20) 22.65 17.50 21.55 12.95

(0.76) (0.92) (1.64) (0.94)

Experiment 3

Modification training (n = 20) 22.80 15.25 21.70 10.75

(0.84) (0.12) (1.57) (0.22)

No adaptation (n = 27) 23.30 16.63 23.30 12.85

(0.60) (0.43) (1.64) (0.45)

Adaptation (n = 26) 26.08 15.85 25.96 12.38

(1.35) (0.41) (1.65) (0.56)

Note: The table presents means, with standard errors in parentheses. State anger and positive and negative affect were measured using the State-Trait Anger Expression Inventory-2 (STAXI-2; Spielberger, 1999) and the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988); the values presented here are from the posttraining assessment.

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013

Emotion Recognition and Aggressive Behavior 3

Fig. 1.  Illustration of the stimuli and design of Experiment 1. Stimuli consisted of a morphed continuum of 15 images ranging from an unambiguously happy face to an unambiguously angry face. At baseline, participants rated these images as happy or angry, in a two-alternative forced-choice procedure, and each participant’s balance point between these categorical judgments was estimated (top row). Participants in the modification condition received biased feedback intended to shift this balance point. This feedback was based on each participant’s baseline balance point, with the “correct” classification shifted two morph steps toward happy; that is, the two images nearest the balance point that the participant would have classified as angry at baseline were classified as happy in the feedback. Participants in the control condition received feedback that was directly based on their baseline categorization. A final test phase, identical to the baseline assessment, was employed to confirm whether the modification training was successful in shifting the balance point and to assess the magnitude of the shift (bottom panel).

to respond (a two-alternative forced-choice judgment of whether the face was happy or angry). Within participants, classification responses to such morph continua tend to shift monotonically from one response to the other across the continuum (Young et al., 1997). Therefore, we were able to obtain a simple estimate of the balance point at which each participant was equally likely to perceive happiness or anger in the presented faces by calculating the number of “happy” responses as a proportion of the total number of trials. Each trial in the training phase was similar to a trial in the baseline and test phases with respect to the intertrial interval and stimulus presentation, but feedback was provided following the participant’s response. In the control condition, feedback was directly based on the participant’s baseline balance point. Responses were classified as “correct” when the participant identified images below the original balance-point image as happy, and faces above that image as angry, and otherwise were classified

as “incorrect.” Feedback was a message saying, “Correct/ Incorrect! That face was happy/angry.” In the modification condition, feedback was also based on the participant’s baseline balance point, but the “correct” classification was shifted two morph steps toward the angry end of the continuum, so that the two images nearest the balance point that the participant would have classified as angry at baseline were considered happy for purposes of feedback. Each face from the 15-face continuum was presented twice (with presentation order randomized) within each block. Six training blocks were delivered, for a total of 180 training trials. Following the test phase, each participant’s balance point was again calculated, using the same method as in the baseline phase, to establish whether the procedure modified the participant’s perception of ambiguous emotional expressions. Participants completed the questionnaire measures of state anger and positive and negative affect both before and after the training procedure.

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013

Penton-Voak et al.

4 Statistical analysis. Linear regression was used to assess the relationship between training condition and state anger following training, with adjustment for baseline state anger and sex. STAXI-2 scores were positively skewed and therefore log-transformed for statistical analyses.

Results Participants in the modification condition showed a shift in balance point relative to participants in the control condition, b = 1.33, SE = 0.37, t (36) = 3.62, p = .001 (see Fig. 2). In addition, participants in the modification condition reported lower levels of state anger after the training procedure relative to those in the control

condition, b = −0.05, SE = 0.02, t (36) = 2.52, p = .016. Training accounted for 3% of the variance in state anger rated immediately following training, ΔF (1, 36) = 6.37, p = .016. Our data demonstrate that the experimental modification of emotional-expression perception altered self-reported aggressive mood. We did not observe any effects of the modification training on measures of positive or negative affect ( ps > .1; see Table 1 for descriptive statistics).

Discussion These results suggest that emotion-recognition biases regarding faces on the anger-happiness continuum play a causal role in subjective anger. Thus, modification

Experiment 1: Control Condition Proportion of “Angry” Responses

Posttraining

.8 .6 .4 .2 .0

1.0

Proportion of “Angry” Responses

1.0

Posttraining

.6 .4 .2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Frame of the Happy-Angry Continuum

Frame of the Happy-Angry Continuum

Experiment 2: Control Condition

Experiment 2: Modification Condition

Baseline

1.0

Posttraining .8 .6 .4 .2 .0

Baseline

.8

.0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Proportion of “Angry” Responses

Proportion of “Angry” Responses

1.0

Experiment 1: Modification Condition

Baseline

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Frame of the Happy-Angry Continuum

Baseline Posttraining

.8 .6 .4 .2 .0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Frame of the Happy-Angry Continuum

Fig. 2.  Mean proportion of “angry” responses to each frame of the continuum before and after training in Experiment 1 (healthy volunteers; top row) and Experiment 2 (high-risk youth; bottom row). Graphs in the left column present results for participants in the control condition, and graphs in the right column present results for participants in the modification condition. Posttraining data for the high-risk youth in Experiment 2 come from the test phase of the last session.

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013

Emotion Recognition and Aggressive Behavior 5 training of the kind used in this experiment may result in beneficial changes in behavior. Conduct disorder in adolescents and young adults, which encompasses antisocial behavior, delinquency, and criminality, has major societal and economic consequences, and is a risk factor for continued criminal offending in adulthood (Scott, Knapp, Henderson, & Maughan, 2001). Our results, therefore, motivated us to investigate whether our training procedure would have an effect on independent assessments of aggressive behavior among individuals with high levels of aggression, and to examine the longevity of any effects over a longer follow-up period.

Experiment 2 Method Participants.  For our second experiment, we recruited 46 adolescents (33 male, 13 female; ages 11–16 years) who had been referred to a youth program, either by the courts or by schools, as being at high risk of committing a crime; we conducted a further screening so that only those with a high frequency of aggressive behavior were included in the sample. Approximately 70% of participants had an official record of criminal offending. Participants were randomly assigned to an emotion-recognition modification procedure (n = 23; mean age = 13.22 years, SE = 0.28) or a control procedure (n = 23; mean age = 13.43 years, SE = 0.35). Materials. The facial stimuli were the same as in Ex­periment 1. Participants completed a self-rated behavior diary, based on the measure developed by McMurran (2007) and modified for use in an adolescent population. This diary covered six categories of aggressive behavior, ranging from verbal aggression to direct physical aggression. The same six categories were included in a staffrated behavior diary that was completed by staff members of the youth program who were blind to participants’ training condition. The staff member rating each participant was required to make a judgment about how often that participant displayed each of the behaviors over the preceding week, on a scale from 0 (not present at all) to 100 ( present all the time). Scores were averaged over the six categories. Procedure. The modification training was similar to that in Experiment 1, except that participants completed four sessions, once a day over a period of either 4 or 5 days (270 trials per session: 45 baseline, 180 training, 45 test). The control training was a treatment condition in which “correct” classifications were moved one step toward the angry end of the continuum for feedback purposes. We incorporated a small modification into the

control training for ethical reasons, given the results of Experiment 1; this procedure ensured that all participants received some degree of “treatment.” The baseline and test phases were the same as in Experiment 1 except that they were repeated over multiple days. Participants completed the daily self-report diary at the end of each day in the baseline week (just prior to training) and 2 follow-up weeks. For each participant, we calculated a score for each week; scores were summed over the 5 assessment days (weekdays) in each week. For the staff-rated diary, program staff recorded the incidence of aggressive behavior in the week prior to training and in the 2 weeks following completion of training. Statistical analysis. Linear regression was used to assess the relationship between training condition and both self-reported and staff-rated aggressive behavior following training, with adjustment for baseline aggressive behavior, age, sex, ethnicity, and history of criminal offending.

Results As in Experiment 1, participants in the modification condition showed a shift in their balance point relative to participants in the control condition (at the end of the last session), b = 4.22, SE = 1.23, t (36) = 3.33, p = .003 (Fig. 2).1 Ratings of aggressive behavior by program staff were available for all 46 participants, irrespective of whether they had completed training or not (i.e., intentto-treat analysis). Intent-to-treat analysis is based on initial treatment intent, and avoids potential confounding due to nonrandom dropout. These data indicated that youth in the modification condition showed lower levels of staff-rated aggressive behavior 2 weeks after the training procedure relative to those in the control condition, b = −2.40, SE = 0.67, t (39) = 3.77, p = .001 (Fig. 3a). Training accounted for 16% of the variance in staff-rated aggressive behavior 2 weeks after the completion of training, ΔF (1, 39) = 14.21, p = .001. When this analysis was restricted to the 34 participants who completed the training procedure, the reduction in staff-rated aggressive behavior in the modification condition relative to the control condition was considerably greater, b = −6.32, SE = 1.11, t (27) = 5.69, p < .001. Data on self-rated aggressive behavior were available for the 34 participants who completed the training procedure and self-report diary. These data also indicated lower aggressive behavior 2 weeks after training in the modification condition compared with the control condition, b = −4.74, SE = 1.26, t (27) = 3.61, p = .001 (Fig. 3b). Data from 1 week after training revealed similar results for both measures of aggressive behavior (Fig. 3).

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013

Penton-Voak et al.

6

b

Control 17

Modification

15 13 11 9 7

Control 10

Aggressive Behavior (Self-Rated)

Aggressive Behavior (Staff-Rated)

a

Pretraining

+7 Days

+14 Days

Modification

9 8 7 6 5

Pretraining

Posttraining

+7 Days

+14 Days

Posttraining

Fig. 3.  Results from Experiment 2: (a) mean staff-rated aggressive behavior for all 46 participants and (b) mean self-rated aggressive behavior for the 34 participants who completed training as a function of time of assessment (pretraining, 1 week after training, and 2 weeks after training). Each graph shows results separately for participants in the modification condition and those in the control condition. Error bars represent ±1 SE. The data shown here were adjusted for age, sex, ethnicity, and previous criminal offending.

Discussion These results support and extend those of Experiment 1, and strengthen the evidence for a causal role of the perception of emotion in anger and aggressive behavior. These effects persist for at least 2 weeks, showing a nonsignificant trend toward an increase in strength over time. One interpretation of these data is that the initial training procedure establishes a virtuous cycle, whereby the alteration in perception of facial expression favoring the perception of happiness over anger may lead to changes in social behavior, which are then reciprocated and reinforced in social interaction. One interesting feature of our data (see Fig. 2) is that after training, participants in the modification condition appeared to generate the typical sigmoidal function that is often seen in categorical judgments of emotion in typical healthy volunteer populations (i.e., undergraduate students). Participants in the control condition, however, did not show this change in behavior, despite receiving feedback based on their original threshold, which would be expected to improve categorization even while leaving the threshold unchanged. We speculate that because the feedback received by participants in the modification condition was inconsistent with their prior expectations, they engaged a learning mechanism to reevaluate ambiguous emotional expressions on the basis of feedback received. In contrast, participants in the control condition received feedback that was statistically consistent with their prior beliefs, and were apparently not motivated to

change their responses in any way, despite the possibility of reducing negative feedback. However, whether the training procedure addresses this more fundamental aspect of emotion recognition (i.e., induces more pronounced categorical perception of emotional expression), or addresses attentional and motivational deficits (or both), remains to be seen. The method we used in Experiments 1 and 2 relies on explicit feedback to change categorical judgments of emotion. If emotion perception plays a causal role in anger and aggressive behavior, manipulating the perception of emotion should lead to comparable effects on behavior irrespective of the method used. We therefore explored whether similar effects could be obtained in the absence of the explicit feedback provided in Experiments 1 and 2, by using a visual adaptation methodology to manipulate categorical judgments of emotional faces. Adaptation occurs as a result of the prolonged viewing of a stimulus and alters the perception of subsequent related stimuli. For example repeated presentations of a line tilted to the left can make a subsequently presented vertical line appear to tilt to the right (Gibson & Radner, 1934). Similar perceptual aftereffects are found for many categories of visual stimuli, including those in motion, as observed in the waterfall illusion (Barlow & Hill, 1963), and appear to reflect common processes across the visual system. However, adaptation can lead not only to low-level visual effects, but also to high-level aftereffects in face perception, changing the perception of traits such as sex, race, identity, and expression. So, for example,

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013

Emotion Recognition and Aggressive Behavior 7 exposure to a series of female faces can make a subsequently presented androgynous face look male, and exposure to angry faces can make subsequent faces look happy (Webster, Kaping, Mizokami, & Duhamel, 2004; Webster & MacLeod, 2011). Although such adaptation techniques are typically used to investigate the psychological representation of facial attributes, we exploited the perceptual aftereffects of adaptation to investigate the relationship between emotion perception and mood.

Experiment 3 Method Participants. We recruited 53 healthy volunteers (21 male, 32 female; ages 20–51 years) from the general population, through advertisements on and around the University Precinct at the University of Bristol (Table 1). Participants were randomly assigned to receive an emotion-recognition modification procedure (adaptation) or a control procedure (no adaptation). Materials.  Facial stimuli and questionnaire instruments (STAXI-2 and PANAS) were the same as in Experiment 1. Procedure. Testing took place within a single session. As in Experiments 1 and 2, the session consisted of three phases. The baseline and test phases to assess balance point were identical to those in Experiments 1 and 2, but the explicit-feedback training phase was replaced by a 20-min visual adaptation phase. The questionnaires were administered both before and after the adaptation phase. The adaptation phase consisted of a 2-back memory task, in which 10 individual faces from the Karolinska face set were presented sequentially (and repeatedly) to participants, who were instructed to respond by indicating whether the identity of the face presented on a given trial matched that of the face presented two trials previously. This task ensured that participants paid attention to the images displayed, in order to enhance adaptation effects (Rhodes et al., 2011). In the adaptation condition, each of the faces displayed an angry expression. In the no-adaptation condition, 5 of the faces were angry, and 5 were happy. The identities of the faces were the same in the two conditions. Each face was displayed for 2,000 ms, with a 20-ms intertrial interval. The images were spatially jittered by an amount that varied randomly from the center of the screen by up to ±51 pixels in the x-coordinate and ±38 pixels in the y-coordinate (approximately 10% in each direction of the 1024- × 768-pixel display; image size = 562 × 765 pixels, viewing distance ≈ 1 m). During presentation of a face, participants indicated whether it was a target (i.e., had the same identity as the face presented two trials previously) or a nontarget (i.e., had a different identity than

the face presented two trials previously). Failure to respond within the 2,000-ms presentation resulted in no response being recorded for that trial. Each adaptation block consisted of 48 trials, in which eight targets were presented. Thirteen blocks were presented, for a total of 624 trials. The session length, about 21 min for each participant, was approximately equal to length of the six training blocks presented in the feedback training of Experiment 1. Statistical analysis. Linear regression was used to assess the relationship between condition and state anger following training, with adjustment for baseline state anger and sex. STAXI-2 scores were positively skewed and therefore log-transformed for statistical analyses.

Results As demonstrated in many other experiments (e.g., Webster et al., 2004), participants in the adaptation condition (but not those in the no-adaptation condition) showed a shift in balance point at test (i.e., were more likely to see ambiguous faces as happy following exposure to angry faces), b = 2.30, SE = 0.47, t (49) = 4.90, p < .001 (see Fig. 4). Furthermore, and crucially, participants in the adaptation condition reported lower levels of state anger after the adaptation procedure relative to those in the control condition, b = −0.06, SE = 0.03, t (49) = 2.01, p = .050. Adaptation training accounted for 6% of the variance in state anger rated immediately following training, ΔF(1, 49) = 4.05, p = .050. As in our first experiment, we did not observe clear effects of the training procedure on measures of positive or negative affect ( ps > .09). (See Table 1 for descriptive statistics.)

Discussion These results confirm that modifying the recognition of emotion in ambiguous emotional expressions using two different mechanisms (explicit feedback and visual adaptation) gives rise to comparable subjective effects on anger. It is notable that the effects observed in Experiments 1 and 3 were highly similar in magnitude, which is consistent with a common effect being achieved regardless of the method by which emotion recognition is modified.

General Discussion The results of our experiments strongly suggest that biases in the perception of emotional facial expressions play a causal role in subjective anger and aggressive behavior. These effects were observed up to 2 weeks after completion of training in Experiment 2, and there was a tendency for the decrease in aggressive behavior to strengthen over

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013

Penton-Voak et al.

8

Adaptation Condition

No-Adaptation Condition 1.0

Proportion of “Angry” Responses

Proportion of “Angry” Responses

1.0

Baseline Postadaptation

.8 .6 .4 .2 .0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Frame of the Happy-Angry Continuum

Baseline Postadaptation

.8 .6 .4 .2 .0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Frame of the Happy-Angry Continuum

Fig. 4.  Mean proportion of “angry” responses to each frame of the continuum before and after training in Experiment 3, in which we employed a visual adaptation paradigm to induce a change in emotion perception. Results are presented separately for participants in the adaptation condition (right panel) and those in the no-adaptation condition (left panel).

time. Experiment 3 indicates that employing a completely different mechanism to manipulate emotion perception results in a comparable change in self-reported mood. Although previous attempts to use emotion-recognition training as an intervention have had limited success (Dadds et al., 2012), they have concentrated on the perception of relatively unambiguous emotional expressions. By contrast, we modified biases in the interpretation of ambiguous emotional expressions, and this may prove to be an important distinction. Our results raise the possibility that a symmetric manipulation may be possible: That is, training participants to classify ambiguous faces as angry (rather than happy) may increase both state anger and, potentially, aggressive behavior. Although demonstrating this effect would be scientifically useful, given the apparent longevity of the effects in Experiment 2 and in our related work on low mood (Penton-Voak et al., 2012), it may be inappropriate. Our results suggest a mechanism that is similar in important respects to mechanisms described in recent cognitive models of how antidepressants have their effects. Harmer et al. (2009) suggested that changes in emotion-processing biases occur rapidly (within hours) of commencing antidepressant usage. They proposed that these changes allow cognitive reconsolidation, resulting in improved mood; that is, mood changes are caused by changes in emotion perception, rather than vice versa. Our results with both state anger and, more directly, aggressive behavior suggest that a similar process may operate in the maintenance of aggressive behavior. That is, the effects of the intervention we tested may be

mediated by the establishment of virtuous cycles that positively reinforce nonaggressive behavior in everyday social interaction. Our results are consistent with Crick and Dodge’s (1994) social information-processing model of aggressive behavior in young people, which describes six steps that lead to enacting competent social behaviors that reflect social adjustment. These six stages are (a) encoding of external and internal cues, (b) interpretation and mental representation of those cues, (c) clarification or selection of a goal, (d) response access or construction, (e) response decision, and (f) behavioral enactment. Manipulating the perception of ambiguous emotional cues of potential threat, as we did in our experiments, would seem to affect the first or second stage of this model. The evidence that a hostile-attribution bias is linked to social maladjustment is robust (Crick & Dodge, 1994), and our findings add further experimental weight to earlier suggestions that hostile-attribution biases precede social maladjustment in young people, and may have a causal role in aggressive behavior (Dodge, Bates, & Pettit, 1990; Rabiner & Coie, 1989). The neuropsychological mechanisms underlying the effects we have reported will be the subject of further studies. Given that the mood changes we report can be elicited via two experimental techniques, one of which is completely nonverbal, it seems unlikely that our findings are the result of top-down, higher cognitive processes alone. Adaptation to facial expressions causes a change in activation of areas of the superior temporal sulcus (Furl, van Rijsbergen, Treves, Friston, & Dolan, 2007),

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013

Emotion Recognition and Aggressive Behavior 9 which does not persist for long periods, although the dynamics of face adaptation are not well understood and may occur across various time scales (Webster & MacLeod, 2011). Given the likely short-term effects of such visual experience, adaptation techniques may not induce the long-term effects on aggressive behavior that were evident in Experiment 2; nevertheless, these findings reinforce the conclusion that experimentally induced biases in emotion processing play a causal role in mood. Clinically, simple emotion training techniques, which can be administered with speed and ease, and in principle remotely, might be a cost-effective means of reducing aggressive behavior in target populations. This possibility, and the exact nature of the mechanisms that underlie the behavioral changes we observed, will be the focus of further work. Acknowledgments We are grateful to the participants who took part in our experiments, and in particular to the young people who gave up their time to take part in the second experiment. We also are grateful to the staff at the Youth Offending Team in South Wales for their support and assistance, and to Mike Goldman, head of the Youth Offending Team, for allowing us to conduct this research within his organization. I. S. P.-V. and M. R. M. contributed equally to this work.

Declaration of Conflicting Interests I. S. P.-V. and M. R. M. are directors of Jericoe Ltd., which develops software for assessing and modifying emotion perception. The other authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

Note 1. Despite the absence of a baseline difference in balance point between the samples in Experiments 1 and 2, our healthy volunteer sample discriminated between stimuli better than our high-risk youth sample, likely because of differences in attention and motivation between the groups.

References Barlow, H. B., & Hill, R. M. (1963). Evidence for a physiological explanation of the waterfall phenomenon and figural aftereffects. Nature, 200, 1345–1347. Best, M., Williams, J. M., & Coccaro, E. F. (2002). Evidence for a dysfunctional prefrontal circuit in patients with an impulsive aggressive disorder. Proceedings of the National Academy of Sciences, USA, 99, 8448–8453. Bowen, E., & Dixon, L. (2010). Concurrent and prospective associations between facial affect recognition accuracy and childhood antisocial behavior. Aggressive Behavior, 36, 305–314. 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.

Dadds, M. R., Cauchi, A. J., Wimalaweera, S., Hawes, D. J., & Brennan, J. (2012). Outcomes, moderators, and mediators of empathic-emotion recognition training for complex conduct problems in childhood. Psychiatry Research, 199, 201– 207. Dadds, M. R., Perry, Y., Hawes, D. J., Merz, S., Riddell, A. C., Haines, D. J., . . . Abeygunawardane, A. I. (2006). Attention to the eyes and fear-recognition deficits in child psychopathy. British Journal of Psychiatry, 189, 280–281. Dodge, K. A. (1993). Social-cognitive mechanisms in the development of conduct disorder and depression. Annual Review of Psychology, 44, 559–584. Dodge, K. A., Bates, J. E., & Pettit, G. S. (1990). Mechanisms in the cycle of violence. Science, 250, 1678–1683. Fairchild, G., Van Goozen, S. H. M., Calder, A. J., Stollery, S. J., & Goodyer, I. M. (2009). Deficits in facial expression recognition in male adolescents with early-onset or adolescenceonset conduct disorder. Journal of Child Psychology and Psychiatry, 50, 627–636. Furl, N., van Rijsbergen, N. J., Treves, A., Friston, K. J., & Dolan, R. J. (2007). Experience-dependent coding of facial expression in superior temporal sulcus. Proceedings of the National Academy of Sciences, USA, 104, 13485–13489. Gibson, J. J., & Radner, M. (1934). Adaptation, after-effect and contrast in the perception of tilted lines. Journal of Experimental Psychology, 20, 453–467. Harmer, C. J., Goodwin, G. M., & Cowen, P. J. (2009). Why do antidepressants take so long to work? A cognitive neuropsychological model of antidepressant drug action. British Journal of Psychiatry, 195, 102–108. Lundqvist, D., Flykt, A., & Öhman, A. (1998). The Karolinska Directed Emotional Faces - KDEF. Stockholm, Sweden: Karolinska Institutet. McMurran, M. (2007). The relationships between alcoholaggression proneness, general alcohol expectancies, hazardous drinking, and alcohol-related violence in adult male prisoners. Psychology, Crime & Law, 13, 275–284. Nasby, W., Hayden, B., & Depaulo, B. M. (1980). Attributional bias among aggressive boys to interpret unambiguous social-stimuli as displays of hostility. Journal of Abnormal Psychology, 89, 459–468. Penton-Voak, I. S., Bate, H., Lewis, G., & Munafò, M. R. (2012). Effects of emotion perception training on mood in undergraduate students: Randomised controlled trial. British Journal of Psychiatry, 201, 71–72. Rabiner, D., & Coie, J. (1989). Effects of expectancy inductions on rejected children’s acceptance by unfamiliar peers. Developmental Psychology, 25, 450–457. Rhodes, G., Jeffery, L., Evangelista, E., Ewing, L., Peters, M., & Taylor, L. (2011). Enhanced attention amplifies face adaptation. Vision Research, 51, 1811–1819. Sato, W., Uono, S., Matsuura, N., & Toichi, M. (2009). Misrecognition of facial expressions in delinquents. Child and Adolescent Psychiatry and Mental Health, 3, 27. Retrieved from http://www.capmh.com/content/3/1/27 Scott, S., Knapp, M., Henderson, J., & Maughan, B. (2001). Financial cost of social exclusion: Follow up study of antisocial children into adulthood. British Medical Journal, 323, 191.

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013

Penton-Voak et al.

10 Spielberger, C. D. (1999). The State-Trait Anger Expression Inventory-2 (STAXI-2): Professional manual. Odessa, FL: Psychological Assessment Resources. Tiddeman, B. P., Burt, M., & Perrett, D. I. (2001). Prototyping and transforming facial textures for perception research. IEEE Computer Graphics and Applications, 21, 42–50. Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070.

Webster, M. A., Kaping, D., Mizokami, Y., & Duhamel, P. (2004). Adaptation to natural facial categories. Nature, 428, 557–561. Webster, M. A., & MacLeod, D. I. (2011). Visual adaptation and face perception. Philosophical Transactions of the Royal Society B: Biological Sciences, 366, 1702–1725. Young, A. W., Rowland, D., Calder, A. J., Etcoff, N. L., Seth, A., & Perrett, D. I. (1997). Facial expression megamix: Tests of dimensional and category accounts of emotion recognition. Cognition, 63, 271–313.

Downloaded from pss.sagepub.com at University of Nottingham on April 7, 2013