Brain and Cognition 77 (2011) 412–418
Contents lists available at SciVerse ScienceDirect
Brain and Cognition journal homepage: www.elsevier.com/locate/b&c
Liberal bias mediates emotion recognition deficits in frontal traumatic brain injury Brandy L. Callahan a,b,⇑, Keita Ueda c, Daisuke Sakata c, André Plamondon a, Toshiya Murai c a
Université Laval, École de psychologie, Québec, Canada G1V 0A6 Centre de recherche Université Laval Robert-Giffard, Québec, Canada G1J 2G3 c Kyoto University, Graduate School of Medicine, Department of Neuropsychiatry, Sakyo Ku, Kyoto 6068507, Japan b
a r t i c l e
i n f o
Article history: Accepted 23 August 2011 Available online 25 September 2011 Keywords: Emotion Recognition Traumatic brain injury Frontal lobe Bias
a b s t r a c t It is well-known that patients having sustained frontal-lobe traumatic brain injury (TBI) are severely impaired on tests of emotion recognition. Indeed, these patients have significant difficulty recognizing facial expressions of emotion, and such deficits are often associated with decreased social functioning and poor quality of life. As of yet, no studies have examined the response patterns which underlie facial emotion recognition impairment in TBI and which may lend clarity to the interpretation of deficits. Therefore, the present study aimed to characterize response patterns in facial emotion recognition in 14 patients with frontal TBI compared to 22 matched control subjects, using a task which required participants to rate the intensity of each emotion (happiness, sadness, anger, disgust, surprise and fear) of a series of photographs of emotional and neutral faces. Results first confirmed the presence of facial emotion recognition impairment in TBI, and further revealed that patients displayed a liberal bias when rating facial expressions, leading them to associate intense ratings of incorrect emotional labels to sad, disgusted, surprised and fearful facial expressions. These findings are generally in line with prior studies which also report important facial affect recognition deficits in TBI patients, particularly for negative emotions. Ó 2011 Elsevier Inc. All rights reserved.
1. Introduction Long-term studies of traumatic brain injury (TBI) generally report notable changes in cognitive and social functioning in patients, and these are corroborated by reports of patients themselves as well as their caregivers. Difficulties in social cognition are often among the most marked of these post-TBI changes (Radice-Neumann, Zupan, Babbage, & Willer, 2007) and may include, for instance, a diminished personal experience of emotion (Hornak, Rolls, & Wade, 1996), a diminished ability to interpret nonverbal emotional cues (Henry, Phillips, Crawford, Ietswaart, & Summers, 2006) and difficulties understanding nonliteral, implied statements in language (McDonald & Flanagan, 2004). Such social difficulties may have profound implications for patients’ welfare and quality of life. Indeed, patients who exhibit greatest difficulty in emotion processing and social cognition are also those who report greatest difficulties in finding or returning to work and maintaining social relationships (Milders, Fuchs, & Crawford, 2003). Laboratory studies of emotion processing in TBI have generally focused on emotion recognition. Emotion recognition is commonly assessed using tests which require participants to attribute emo⇑ Corresponding author at: Université Laval, École de psychologie, Pavillon Félix-Antoine-Savard, Québec, Canada G1 V 0A6. Fax: +1 418 656 3646. E-mail address:
[email protected] (B.L. Callahan). 0278-2626/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.bandc.2011.08.017
tional labels to photographs of facial expressions or to judge the extent to which a given emotion is being expressed. Such stimuli are considered to be the most appropriate for assessing deficits in emotion processing because they reflect real-world representations of emotional expression, as compared to other emotional stimuli such as emotional word lists or pictures of emotional scenes. On tests of emotion recognition, TBI patients have consistently been shown to be impaired relative to healthy control subjects (Adolphs, Damasio, Tranel, Cooper, & Damasio, 2000; Adolphs, Tranel, & Damasio, 2003; Croker & McDonald, 2005; Henry, Phillips, Crawford, Theodorou, & Summers, 2006; Henry, Phillips, Crawford, Ietswaart, et al., 2006; Hopkins, Dywan, & Segalowitz, 2002; McDonald & Flanagan, 2004; McDonald & Saunders, 2005). In particular, negative emotions (anger, fear, sadness, disgust) appear to be less well-recognized by TBI patients than positive emotions (happiness, surprise) (Hopkins et al., 2002; Jackson & Moffat, 1987; Kucharska-Pietura, Phillips, Gernand, & David, 2003). Interestingly, these deficits appear to persist over time (Ietswaart, Milders, Crawford, Currie, & Scott, 2008) and are present even when the expression of emotion is clear, unambiguous and intense (Spell & Frank, 2000). Corroborating findings of emotion recognition deficits in TBI is the observation that these patients appear to have abnormal autonomic responses to emotional stimuli; that is, they show no change in facial musculature or skin conductance in response to emotionally valenced (positive and negative) stimuli, compared to healthy control
B.L. Callahan et al. / Brain and Cognition 77 (2011) 412–418
subjects (Soussignan, Ehrle, Henry, Schaal, & Bakchine, 2005). On self-reported measures, patients also note a significant blunting of their own emotional experiences since sustaining a TBI, especially in regard to negative emotions (Hornak et al., 1996). As such it appears that the experience of emotion, as well as the interpretation of others’ expressions of emotion, is significantly affected by TBI. The neural substrates of facial emotion processing include a network of cortical and subcortical structures, with the prefrontal cortex occupying a central part in this network. The orbitofrontal cortex (OFC), in particular, is hypothesized to play a key role in recognizing negative emotions (Bechara, Damasio, & Damasio, 2000). As such, it is not surprising that emotion recognition seems to be most affected in TBI patients who have sustained damage to frontal brain regions. Case report studies of patients with frontal-lobe lesions, such as those of Phineas Gage (Harlow, 1848) and EVR (Eslinger & Damasio, 1985), have reported outstanding behavioral changes which include inappropriate affect, disinhibition and lack of self-control, impaired insight, social withdrawal and apathy, and failure to perceive and respond to interpersonal cues including facial expression. In addition, these patients show such social dysfunction even if no other neurological or neuropsychological symptoms are present. Recent studies that have explored facial emotion recognition in TBI have exclusively used response accuracy as their dependent measure; that is, the proportion of correct responses of TBI patients is compared to that of control subjects as a measure of emotion recognition impairment. Surprisingly, no studies have yet investigated the nature of this impairment: Is a certain pattern observable in the types of mistakes being made by TBI patients? It is reasonable to assume that the answer to this question is yes. Studies with various patient populations have revealed a variety of response biases when participants are asked to make a judgement regarding emotional stimuli, particularly in cases where frontal brain structures are supposed to be affected. In several cases, this bias may lead participants to assign incorrect emotional labels to facial expressions. For instance, when asked to identify different facial emotions, depressed individuals show a clear tendency to misidentify neutral expressions as negative (Gollan, Pane, McCloskey, & Coccaro, 2008; Leppanen, Milders, Bell, Terriere, & Hietanen, 2004), reflecting a negative emotional bias. Similarly, patients with schizophrenia (Kohler et al., 2003) as well as men with mental retardation (Walz & Benson, 1996) have also been found to exhibit a negative bias, as they tend to misidentify neutral facial cues as negatively valenced. In other cases, the response bias may be influenced by the level of subjective criteria used by participants to make a judgement regarding facial expressions of emotion; this may lead to responses biases which are liberal (wherein the level of subjective evidence necessary to elicit a positive response is relatively low) or conservative (wherein a relatively high level of subjective evidence is required to elicit a positive response). For instance, when asked to label the level of fear displayed in fearful facial expressions, anxious patients tend to label mild or moderately fearful faces as ‘‘more fearful’’ than do healthy controls, reflecting a liberal bias in attributing fearful ratings (Frenkel, Lamy, Algom, & Bar-Haim, 2009). Similarly, participants presenting heightened sub-clinical paranoia have also been found to label mildly angry facial expressions as ‘‘more threatening’’ in anxietyevoking conditions than do participants with low paranoia (Westermann & Lincoln, 2010). Studies having specifically investigated such response biases in clinical samples are few, but their results suggest that different types of injury or dysfunction in the prefrontal cortex can affect the way in which emotional information is interpreted. Moreover, clarifying how emotional data is perceived by different patient populations can help in interpreting data collected from tests using emotional stimuli.
413
No research has yet looked at biases in emotional processing in frontal TBI, but results from prior research in other patient populations suggest that such a study would likely yield interesting and useful findings. As such, the aim of the present study was to assess facial emotion recognition in TBI patients having sustained frontallobe injuries and examine participants’ responses for patterns that may explain recognition deficits. 2. Materials and methods 2.1. Participants Fourteen patients (10 men and 4 women) having suffered frontal-lobe damage due to TBI were referred to the psychiatric department of Kyoto University and Kitano Hospital in Osaka. All patients suffered TBI by traffic accident and were discharged after having received conservative therapy. Damage sites were identified in the frontal lobe in all patients by MRI; the location of the lesion was identified by a number of medical doctors in our laboratory. All lesion sites were painted manually into the standard T2 image in MRIcro software (Nottingham, UK), and shown in Fig. 1. The overlap of lesion sites was centered in the bilateral ventromedial orbitofrontal cortex. No patient presented neurological deficit, except for impaired sense of smell caused by damage to the olfactory nerve. The comparison group consisted of 22 healthy subjects (11 men and 11 women) recruited from the local community, who were matched with the TBI group on age and education level. A comprehensive medical and psychiatric history was obtained from all participants. No participant reported a history of neurological disease, psychiatric or serious medical illness, head injury (apart from TBI in the patient group), stroke, antipsychotic medication use or substance abuse disorder, with the exception of one male TBI patient who had a history of somatisation disorder and had received medication. However, this participant had considerable social ability and held a regular occupation. All participants were right-handed. After a complete description of the study to participants, written informed consent was obtained. This study was granted approval by the Committee on Medical Ethics of Kyoto University. 2.2. Neuropsychological tasks 2.2.1. Verbal and performance IQ As part of the test battery administered to all subjects, verbal and performance IQ were assessed using the Wechsler Adult Intelligence Scale Revised (Wechsler, 1981). 2.2.2. Benton facial recognition test The Benton Facial Recognition Test (BFRT) (Benton, Hamsher, Varney, & Spreen, 1983) was administered to all participants in order to ensure their capacity to recognize and distinguish nonemotional human faces. In this 22-item test, participants were asked to match one of six different faces to a target face. 2.2.3. Frontal systems behavior scale The Frontal Systems Behavior Scale (FrSBe) (Grace & Malloy, 2001) is used to evaluate problems in daily life which are thought to be caused by damage to the prefrontal cortex. This scale consists of three subscales, Apathy, Disinhibition, and Executive Dysfunction, which are thought to evaluate dysfunction of the anterior cingulate, orbitofrontal cortex and dorsolateral prefrontal cortex, respectively. There are 46 questions regarding activities of daily living, in which subjects are asked to rate each item on a scale from 1 to 5 regarding daily life before and after their brain injury. Items within each of the subscales are summed and converted to a
414
B.L. Callahan et al. / Brain and Cognition 77 (2011) 412–418
Fig. 1. The lesion locations of the 14 patients were overlapped on a standard brain space. The number of subjects sharing a lesion site is indicated by a 14-graded color bar shown at right: Purple (leftmost of the bar) corresponds to the regions which are damaged in only one patient, while red (rightmost of the bar) corresponds to the regions which are damaged in all 14 patients.
t-score using normative data stratified by age and education. These t-scores were used in analyses presented hereafter, with higher scores reflecting stronger symptoms. Because TBI patients are known to have poor insight and judgement (Flashman & McAllister, 2002), we used the family (or staff) rating version of this scale with the assumption that these ratings would correspond more accurately to patients’ actual behavioral problems. This questionnaire was administered only to the TBI patient group. 2.2.4. Behavioral assessment of the dysexecutive syndrome The Behavioral Assessment of the Dysexecutive Syndrome (BADS) (Wilson, Alderman, Burgess, Emslie, & Evans, 1996) is a neuropsychological battery comprising six tests and two questionnaires to assess executive dysfunction. It is specifically designed to reflect real-world tasks requiring planning, organization, judgement and problem-solving. It was administered only to the TBI patient group. 2.2.5. Iowa gambling task The computerized version of the Iowa Gambling Task (IGT) (Bechara, Damasio, Damasio, & Anderson, 1994) was used, which was developed by our group and described previously (Fukui, Murai, Fukuyama, Hayashi, & Hanakawa, 2005). This task requires participants to draw cards from four virtual decks in order to win as much money as possible, and was used to evaluate planning and decision-making. Again, this task was administered to TBI patients only. 2.3. Experimental task The task used in the present study is identical to that described elsewhere (Adolphs, Tranel, Damasio, & Damasio, 1994). Briefly, 36 photographs depicting happy, sad, angry, fearful, surprised and disgusted facial expressions, as well as three photographs depicting neutral expressions, were selected from the Ekman series (Ekman & Friesen, 1976). Photographs were presented one at a time to all participants in random order on a computer screen, and this procedure was repeated six times in separate blocks. In each block, participants were asked to evaluate the extent to which the presented facial expressions reflected each of six emotions, using a scale from 0 (not at all) to 5 (very much). As such, each photograph was rated on all of six affect scales (happiness, sadness, anger, fear, surprise and disgust). It should be noted that although stimuli depicted Caucasian faces, this experimental task was considered appropriate to use in our Japanese sample in light of findings showing that emotion judgments do not seem to differ as a function of perceived nationality (Matsumoto, 2007).
subjects were calculated. This method allowed for clear comparison of each participant’s score with the norm in our sample. Next, in order to standardize participants’ scores to allow averaging over several faces, all Pearson correlations were Fisher Z-transformed. For all subjects, these Z scores were averaged across faces that expressed the same emotion (e.g. the average Z score for all happy faces), and these computed averages were used as the dependent measure in subsequent statistical analyses. It should be noted that higher Z values indicate stronger correlations with control subjects’ scores. All statistical analyses were performed using SPSS v.17.0 software for Windows. Statistical significance thresholds were set at p < 0.05.
3. Results As shown in Table 1, groups did not differ significantly in terms of sociodemographic variables or on verbal IQ. BFRT data indicated that all subjects were able to process facial features normally. Healthy participants had significantly higher performance IQ (PIQ) than TBI patients, which was not surprising considering the nature of the lesions sustained by the patients. In order to assess overall facial emotion recognition performance in both groups, Z scores for each of the seven facial expressions (happy, surprised, fearful, disgusted, angry, sad and neutral) were compared using a 2 7 repeated-measures ANOVA. This analysis revealed a significant overall effect of Group (F(1, 34) = 10.2, p = 0.003) and of Emotion (F(6204) = 139.7, p < 0.001). The Group Emotion interaction did not quite reach significance (F(6204) = 2.3, p = 0.074). Next, in order to uncover where differences arose, a one-way ANOVA was conducted, again using Z scores as the dependent variable. TBI patients were found to have significantly lower Z scores than controls, for surprised (F(1, 35) = 12.9, p = 0.001), fearful (F(1, 35) = 6.5, p = 0.016), disgusted (F(1, 35) = 7.3, p = 0.011) and sad (F(1, 35) = 10.5, p = 0.003) facial expressions. The group difference for facial expressions of anger was marginally significant (F(1, 35) = 4.0, p = 0.052), while differences for happy (F(1, 35) = 1.7, p = 0.197) and neutral faces (F(1, 35) = 1.6, p = 0.218) were not significant. Fig. 2 illustrates these data. To discover the response patterns that accounted for these group differences, a 2 (Group) 4 (Emotion) 6 (Rating) ANOVA was next conducted using participants’ raw rating scores for each
Table 1 Demographic and neuropsychological data for TBI and control subjects.
2.4. Data analysis Raw scores from the facial emotion recognition task were analyzed using Pearson correlation between each patient’s six ratings of a single face, and the average control group’s six ratings for the same face. In the case of healthy subjects, correlation coefficients between each healthy individual and the remaining 21 control
Sex (male/female) Age (years) Education (years) Verbal IQ Performance IQ
TBI n = 14
CON n = 22
p value
10/4 38.0 (12.9) 13.9 (2.7) 99.2 (17.7) 99.1 (16.8)
11/11 40.0 (7.7) 13.5 (2.6) 107.2 (15.4) 109.8 (14.1)
.207 .605 .638 .157 .047
Note: Figures are mean and standard deviation (SD) in parentheses.
415
B.L. Callahan et al. / Brain and Cognition 77 (2011) 412–418
(F(1, 34) = 0.0, p = 0.920) or disgusted (F(1, 34) = 3.5, p = 0.069) ratings given to fearful faces. These results are summarized in Fig. 3. To establish whether TBI patients exhibited a bias towards rating one emotional label more highly compared to others than did control participants, ratings for each label were computed across all seven facial expression. A 2 (Group) 6 (Rating) repeated measures ANOVA revealed that, overall, TBI patients gave higher fear (F(1, 34) = 5.2, p = 0.029) sadness (F(1, 34) = 4.9, p = 0.033) and anger ratings (F(1, 34) = 6.0, p = 0.020) to all faces than did controls, and marginally higher ratings for happiness (F(1, 34) = 4.1, p = 0.051), disgust (F(1, 34) = 3.7, p = 0.062) and surprise (F(1, 34) = 3.6, p = 0.067), suggesting a liberal attribution bias; indeed, the level of subjective evidence necessary to elicit a strong rating, regardless of the label, was apparently lower in TBI patients than in controls. To formally test the hypothesis that group differences in emotion recognition were explained by a liberal bias, a single composite Liberal score was computed by averaging participants’ raw scores for all seven facial expressions. This composite score was then used as a hypothesized mediator in bootstrapping analyses (Preacher & Hayes, 2008). This analysis was chosen because adding the hypothesized mediator as a covariate in a regression analysis does not allow to test the significance of the indirect effect. The change from a significant regression weight to a non-significant regression weight does not necessarily imply that the indirect effect is significant, and the absence of change in significance after the addition of a covariate does not necessarily imply that the indirect effect is insignificant (Holmbeck, 2002). Sobel’s test permits the investigation of the significance of the indirect path but has low power in small samples (MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002). Consequently, bootstrapping is suggested (Preacher & Hayes, 2008) with small sample sizes. As recommended by Hayes (2009), we will only report bootstrapping results. For all bootstrapping analyses, 5000 bootstrap samples were used and the Bias Corrected and Accelerated confidence intervals (BCa CI) were used to determine significance, thus allowing to formally test the indirect effect. If zero is not included in the interval, it is conceptually the same as rejecting the null hypothesis that the indirect effect is equal to zero. PIQ was entered as a covariate in all bootstrapping analyses to ensure that it did not account for group differences in emotion recognition. Bootstrapping analyses revealed that the indirect effect of Liberal bias was significant for anger, with a point estimate of .14 and 95% BCa CI of .39, .01; for disgust, with a point estimate of .14 and a 95% BCa CI of .40, .02; for happy with a point estimate of .49 and a 95% BCa CI of 1.13, .04; for sadness with a point estimate of .15 and a 95% BCa CI of .41, .02; and for surprise with a point estimate of .27 and a 95% BCa of .59, .04. The indirect
4 TBI CON
Z score
3
2
*
*
*
§
*
Angry
Sad
1
0 Happy
Surprised Fearful Disgusted
Neutral
Facial expression Fig. 2. Performance of TBI patients and control subjects for each stimulus type. p < 0.05; § < 0.10.
of the four facial expressions for which performance was found to be impaired (surprised, fearful, disgusted, sad). This analysis revealed significant main effects of Group (F(1, 34) = 6.0, p = 0.020), Emotion (F(3102) = 80,2, p < 0.001) and Rating (F(5170) = 34.3, p < 0.001), as well as an Emotion Rating interaction (F(15, 510) = 113.2, p < 0.001). No Emotion Group interaction was present, F(3102) = 0.7, p = 0.526, nor was a Ratings Group interaction (F(5170) = 0.5, p = 0.704), but a marginally significant Emotion Rating Group 3-way interaction (F(15, 510) = 2.1, p = 0.059) was revealed. Inspection of simple effects found that TBI patients gave significantly higher happiness (F(1, 34) = 4.4, p = 0.044), fear (F(1, 34) = 10.1, p = 0.003) and anger ratings (F(1, 34) = 6.1, p = 0.019) to sad faces than did healthy subjects. Ratings of sadness (F(1, 34) = 1.5, p = 0.237), surprise (F(1, 34) = 2.8, p = 0.102) and disgust (F(1, 34) = 3.3, p = 0.076) attributed to sad faces did not differ between groups. In the case of disgusted faces, TBI patients rated these as significantly more happy (F(1, 34) = 8.7, p = 0.006) and surprised (F(1, 34) = 4.4, p = 0.043) than did controls, but ratings of sadness (F(1, 34) = 3.6, p = 0.068), fear (F(1, 34) = 3.5, p = 0.071), anger (F(1, 34) = 1.7, p = 0.197) and disgust (F(1, 34) = 0.0, p = 0.908) attributed to disgusted faces did not differ between groups. TBI patients also gave significantly higher happiness (F(1, 34) = 4.9, p = 0.034), disgust (F(1, 34) = 7.8, p = 0.009) anger (F(1, 34) = 5.1, p = 0.031) and sadness ratings (F(1, 34) = 4.6, p = 0.040) to surprised faces than did healthy controls; only ratings of fear (F(1, 34) = 3.9, p = 0.057) and surprise (F(1, 34) = 0.0, p = 0.924) attributed to these faces were comparable between groups. Finally, TBI patients attributed significantly higher anger ratings to fearful faces (F(1, 34) = 5.5, p = 0.026) than did controls. There were no differences in sadness (F(1, 34) = 2.3, p = 0.135), happiness (F(1, 34) = 2.6, p = 0.118), surprised (F(1, 34) = 0.6, p = 0.432), fearful
5
TBI CON
Rating
4 3 *
* *
2
*
1
*
*
*
* *
*
0 Ha Su Fe Di An Sa
Ha Su Fe Di An Sa
Ha Su Fe Di An Sa
Ha Su Fe Di An Sa
Labels attributed to sad faces
Labels attributed to disgusted faces
Labels attributed to surprised faces
Labels attributed to fearful faces
Fig. 3. Intensity ratings (0–5) attributed to each facial expression for each emotion. Note: Ha = Happiness, Su = Surprise, Fe = Fear, Di = Disgust, An = Anger, Sa = Sadness. p < 0.05.
416
B.L. Callahan et al. / Brain and Cognition 77 (2011) 412–418
effect was not significant for fear, with a point estimate of .02 and a 95% BCa CI of .12, .05; nor for neutral, with a point estimate of .09 and a 95% BCa CI of .01, .32, with the latter approaching significance. In order to verify whether the overall liberal ratings for each label evidenced in the TBI group were associated with neuropsychological measures, correlation analyses were performed between the composite Liberal score and IGT, FrSBe and BADS scores. A single moderate, marginally significant negative correlation emerged between Liberal and IGT scores (r = 0.52, p = 0.055), suggesting that poorer IGT performance tended to be associated with more liberal scoring.
4. Discussion The purpose of the present study was to assess facial emotion recognition ability in individuals with frontal-lobe TBI using Adolphs’ (1994) task, and to discover whether a specific response pattern could explain recognition deficits. Our first question of interest was addressed by comparing patients’ and controls’ overall test performance using standardized Z scores. We found an overall emotion recognition deficit in the TBI patient group, which is in line with prior studies. Indeed, TBI patients have consistently been found to be impaired in recognizing facial expressions of emotion (Adolphs et al., 2000, 2003; Croker & McDonald, 2005; Henry, Phillips, Crawford, Ietswaart, et al., 2006; Henry, Phillips, Crawford, Theodorou, et al., 2006; Hopkins et al., 2002; McDonald & Flanagan, 2004; McDonald & Saunders, 2005). Closer inspection of our data revealed that group differences were attributable to particularly poor patient ratings of surprised, fearful, disgusted and sad facial expressions. This finding is also generally consistent with patients’ difficulty in recognizing negative emotions reported in the literature, and may be best understood by considering the damaged brain areas in TBI patients in our sample: lesion sites were focused in the OFC, an area known to play a key role in recognizing negative emotions (Bechara et al., 2000). Several previous behavioral studies have shown that TBI patients have notable difficulty processing negative facial expressions such as fear, disgust, sadness and anger (Hopkins et al., 2002; Jackson & Moffat, 1987; Kucharska-Pietura et al., 2003). While recognition of angry faces was not found to be significantly impaired in TBI patients in the present study, group differences were marginally significant (p = 0.052), possibly due to our small sample size. Patients’ impairment in identifying surprised faces is rather unexpected however, as prior studies have generally not reported this result. Impairment in identifying surprised faces is nonetheless consistent with the main finding of a liberal emotion attribution bias reported here. Interestingly, ratings for happy faces were very accurate in both groups. This is in line with findings reported extensively in the literature (see Babbage et al., 2011). Indeed, recognition of the facial expression of happiness is consistently found to be preserved after brain injury, even when recognition for all other expressions is impaired (e.g. Adolphs, Damasio, Tranel, & Damasio, 1996). Several explanations may be advanced to account for this result. Firstly, it is possible that happiness is more easily recognizable than other emotions simply by virtue of the fact that there are few positive emotions with which happiness can be confounded. As other authors have previously suggested (Adolphs et al., 1996), it may be argued that happiness is the only positive basic emotion and is thus easiest to recognize. Secondly, certain features of the expression of happiness are unique to this emotion: it is the only emotion which is nearly always expressed with the stereotypic smile. A number of negative emotions, on the other hand, may be expressed with a frown or furrowed brow, and are therefore probably more difficult to tell apart, especially if emotion recogni-
tion structures and pathways in the brain have been weakened by injury as was the case in our sample. Perhaps in addition to this, recent evidence sheds light on patients’ accuracy in recognizing happy faces: TBI patients have been found to show abnormal, decreased electrodermal activity compared to healthy volunteers when viewing facial expressions of negative, but not positive, emotions (Hopkins et al., 2002), suggesting that the ability to correctly recognize facial expressions of happiness may have a different physiological basis than the ability to recognize other emotions. The behavioral recognition deficits seen in TBI patients for negative emotions may therefore not extend to positive emotions. This study’s second question of interest was addressed by looking at individual ratings given to emotional faces for which recognition was impaired in TBI patients. Patients tended to give higher ratings of happiness to surprised and disgusted faces than did controls; higher ratings of disgust to surprised faces; higher ratings of anger to surprised, fearful and sad faces; higher ratings of sadness to surprised faces; and higher ratings of surprise to disgusted and sad faces than did healthy subjects. Furthermore, TBI participants were found to give higher emotion ratings to facial expressions overall. As such, they appear to display a liberal bias when selecting an appropriate emotion intensity score for emotional faces, selecting high ratings for all proposed labels regardless of whether the label is correct or not. In other words, this liberal bias signifies that TBI patients give high, appropriate ratings to correct labels, as well as high ratings to incorrect labels. Despite our small sample size, we were able to show, through bootstrapping analyses, that this pattern of responses was responsible for patients’ poor emotion recognition performance compared to healthy controls. As clinical research is often carried out with small sample sizes, we recommend that researchers in this field use bootstrapping to test indirect effects. When patients’ Liberal bias was examined against neuropsychological measures, no significant association was revealed. A marginal but non-significant negative association emerged between patients’ Liberal bias and scores on the IGT, which is generally considered to assess planning and decision-making. TBI patients’ IGT net scores were nearly zero in mean, indicating that patients did not learn from the negative feedback provided in the task. As suggested by Rolls and colleagues, TBI patients with lesions in OFC areas may have difficulty adjusting their choices because of the lack of the ability of reversal leaning (Fellows & Farah, 2003; Rolls & Grabenhorst, 2008). In our study, such deficits may have played a role in explaining patients’ difficulty regulating their emotion ratings, although these results should be interpreted cautiously because they did not reach statistical significance, perhaps due to our relatively small sample size. Nevertheless, it is reasonable to hypothesize that poor decisionmaking abilities in TBI may influence facial affect recognition, as several previous studies have found the ventromedial prefrontal region and surrounding areas to be associated with both emotion recognition and decision-making (for a review, see Zald & Andreotti, 2010). Moreover, one behavioral study has indicated a significant positive relationship between TBI patients’ ability to recognize facial expressions of emotion and measures of cognitive flexibility, a key component in successful decision-making and reversal learning (Milders, Ietswaart, Crawford, & Currie, 2008). Studies conducted with patients presenting other types of frontal-lobe impairment, such as schizophrenia, have also found an association between emotion recognition impairments and perseverative errors in decision-making tasks (Bryson, Bell, & Lysaker, 1997; Lee, Lee, Kweon, Lee, & Lee, 2009). Further studies are needed to clarify the strength and direction of the relationship between emotion recognition and executive functioning in TBI using specific measures of planning and executive dysfunction.
B.L. Callahan et al. / Brain and Cognition 77 (2011) 412–418
It remains unclear why patients consistently overestimate rather than underestimate intensity ratings. The explanation may lie in patients’ personal experience of emotion, which has been reported to be extremely blunted, particularly for negative emotions (Hornak et al., 1996). It has been suggested that facial affect recognition depends heavily on subjective emotional experiences and interpretation of one’s own emotional state (Adolphs et al., 2000; Hornak et al., 2003), and previous research has indeed shown poor facial affect recognition to be associated with self-reported reductions in patients’ own experiences of emotion (Croker & McDonald, 2005). It is possible that TBI patients may feel such weak emotional experiences that facial expressions of emotion displayed by others may appear overstated or exaggerated, though they may be unable to correctly identify the emotion. As such, when asked to rate the intensity of these emotions, TBI patients may be inclined to select abnormally high incorrect ratings. Prior studies having investigated the relationship between subjective emotional experience and facial affect recognition have indeed reported that changes in perceived emotion following TBI are associated with poorer performance on emotion recognition tasks (Croker & McDonald, 2005; Hornak et al., 2003), and one study further specified that it is a decrease in perceived emotion intensity that is associated with poorer emotion recognition (Croker & McDonald, 2005). Because these studies have not reported response patterns in facial affect recognition as we have done here, and because the present study did not incorporate a measure of subjective emotional experience, it is impossible to ascertain whether blunted affect is associated with inaccurate affect recognition due to liberal response patterns. One recent study having investigated the relationship between subjective emotional experience and perception of others’ emotions in healthy young adults provides evidence that partially supports the results reported here. Halberstadt and colleagues (2011) found that participants who reported being less expressive (i.e., seldom displayed emotion) perceived higher emotional intensity in facial stimuli than did more expressive individuals, which seems to support the possibility that the blunted affect of TBI patients in our sample may have led them to select across-the-board high ratings. On the other hand, these authors also found that participants who described themselves as less emotional (e.g., seldom felt intense emotion) perceived others’ emotional expressions as less intense than did more emotional participants. It therefore appears that interpretation of facial expressions is influenced to some extent by one’s own physical expression of emotion, at least in healthy adults, although this may not extend to the subjective experience emotion. Future studies, incorporating measures of subjective emotional experience as well as analyses of participants’ response patterns, are needed to investigate this issue in TBI patients. In sum, the present study replicated the finding of impaired facial emotion recognition in frontal TBI patients, and extended it to show that patients’ recognition deficit was due primarily to a liberal attribution bias which led them to overestimate the intensity of incorrect emotion labels. To our knowledge, this is the first study to characterize the nature of facial emotion recognition deficits in TBI patients. Future studies may explore different patterns of emotion attribution biases across TBI populations with different lesion sites, or evaluate the effect of cognitive rehabilitation on emotion recognition ability. Acknowledgment Brandy Callahan is supported by a doctoral award from the Alzheimer Society of Canada and an additional research award from the Centre de recherche sur le cerveau, le comportement et la neuropsychiatrie.
417
References Adolphs, R., Damasio, H., Tranel, D., Cooper, G., & Damasio, A. R. (2000). A role for somatosensory cortices in the visual recognition of emotion as revealed by three-dimensional lesion mapping. Journal of Neuroscience, 20(7), 2683–2690. Adolphs, R., Damasio, H., Tranel, D., & Damasio, A. R. (1996). Cortical systems for the recognition of emotion in facial expressions. Journal of Neuroscience, 16(23), 7678–7687. Adolphs, R., Tranel, D., & Damasio, A. R. (2003). Dissociable neural systems for recognizing emotions. Brain and Cognition, 52(1), 61–69. doi:10.1016/s02782626(03)00009-5. Adolphs, R., Tranel, D., Damasio, H., & Damasio, A. (1994). Impaired recognition of emotion in facial expressions following bilateral damage to the human amygdala. Nature, 372(6507), 669–672. Babbage, D. R., Yim, J., Zupan, B., Neumann, D., Tomita, M. R., & Willer, B. (2011). Meta-analysis of facial affect recognition difficulties after traumatic brain injury. Neuropsychology, 25(3), 277–285. doi:10.1037/a0021908. Bechara, A., Damasio, H., & Damasio, A. R. (2000). Emotion, decision making and the orbitofrontal cortex. Cerebral Cortex, 10(3), 295–307. Bechara, A., Damasio, A. R., Damasio, H., & Anderson, S. W. (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition, 50(1–3), 7–15. Benton, A. L., Hamsher, K. d. S., Varney, N. R., & Spreen, O. (1983). Contributions to neuropsychological assessment. A clinical manual. New York, NY: Oxford University Press. Bryson, G., Bell, M., & Lysaker, P. (1997). Affect recognition in schizophrenia: A function of global impairment or a specific cognitive deficit. Psychiatry Research, 71(2), 105–113. doi:10.1016/s0165-1781(97)00050-4. Croker, V., & McDonald, S. (2005). Recognition of emotion from facial expression following traumatic brain injury. Brain Injury, 19(10), 787–799. doi:10.1080/ 02699050500110033. Ekman, P., & Friesen, W. (1976). Pictures of facial affect. Palo Alto, CA: Consulting Psychologists Press. Fellows, L. K., & Farah, M. J. (2003). Ventromedial frontal cortex mediates affective shifting in humans: Evidence from a reversal learning paradigm. Brain, 126, 1830–1837. doi:10.1093/brain/awg180. Flashman, L. A., & McAllister, T. W. (2002). Lack of awareness and its impact in traumatic brain injury. NeuroRehabilitation, 17(4), 285–296. Frenkel, T. I., Lamy, D., Algom, D., & Bar-Haim, Y. (2009). Individual differences in perceptual sensitivity and response bias in anxiety: Evidence from emotional faces. Cognition & Emotion, 23(4), 688–700. doi:10.1080/02699930802076893. Fukui, H., Murai, T., Fukuyama, H., Hayashi, T., & Hanakawa, T. (2005). Functional activity related to risk anticipation during performance of the Iowa gambling task. Neuroimage, 24(1), 253–259. doi:10.1016/j.neuroimaging.2004.08.028. Gollan, J. K., Pane, H. T., McCloskey, M. S., & Coccaro, E. F. (2008). Identifying differences in biased affective information processing in major depression. Psychiatry Research, 159(1–2), 18–24. doi:10.1016/j.psychres.2007.06.011. Grace, J., & Malloy, P. F. (2001). Frontal Systems Behavior Scale (FrSBe): Professional manual. Lutz, FL: Psychological Assessment Resources. Halberstadt, A. G., Dennis, P. A., & Hess, U. (2011). The influence of family expressiveness, individuals’ own emotionality, and self-expressiveness on perceptions of others’ facial expressions. Journal of Nonverbal Behavior, 35(1), 35–50. doi:10.1007/s10919-010-0099-5. Harlow, J. M. (1848). Passage of an iron rod through the head. Boston Medical and Surgical Journal, 39, 389–393. Hayes, A. F. (2009). Beyond baron and Kenny: Statistical mediation analysis in the new millennium. Communication Monographs, 76(4), 408–420. Henry, J. D., Phillips, L. H., Crawford, J. R., Ietswaart, M., & Summers, F. (2006). Theory of mind following traumatic brain injury: The role of emotion recognition and executive dysfunction. Neuropsychologia, 44(10), 1623–1628. doi:10.1016/j.neuropsychologia.2006.03.020. Henry, J. D., Phillips, L. H., Crawford, J. R., Theodorou, G., & Summers, F. (2006). Cognitive and psychosocial correlates of alexithymia following traumatic brain injury. Neuropsychologia, 44(1), 62–72. doi:10.1016/j.neuropsychologia. 2005.04.011. Holmbeck, G. N. (2002). Post-hoc probing of significant moderational and mediational effects in studies of pediatric populations. Journal of Pediatric Psychology, 27(1), 87–96. doi:10.1093/jpepsy/27.1.87. Hopkins, M. J., Dywan, J., & Segalowitz, S. J. (2002). Altered electrodermal response to facial expression after closed head injury. Brain Injury, 16(3), 245–257. doi:10.1080/02699050110103346. Hornak, J., Bramham, J., Rolls, E. T., Morris, R. G., O’Doherty, J., Bullock, P. R., et al. (2003). Changes in emotion after circumscribed surgical lesions of the orbitofrontal and cingulate cortices. Brain, 126, 1691–1712. doi:10.1093/ brain/awg168. Hornak, J., Rolls, E. T., & Wade, D. (1996). Face and voice expression identification inpatients with emotional and behavioural changes following ventral frontal lobe damage. Neuropsychologia, 34(4), 247–261. Ietswaart, M., Milders, M., Crawford, J. R., Currie, D., & Scott, C. L. (2008). Longitudinal aspects of emotion recognition in patients with traumatic brain injury. Neuropsychologia, 46(1), 148–159. doi:10.1016/j.neuropsychologia. 2007.08.002. Jackson, H. F., & Moffat, N. J. (1987). Impaired emotional recognition following severe head-injury. Cortex, 23(2), 293–300.
418
B.L. Callahan et al. / Brain and Cognition 77 (2011) 412–418
Kohler, C. G., Turner, T. H., Bilker, W. B., Brensinger, C. M., Siegel, S. J., Kanes, S., et al. (2003). Facial emotion recognition in schizophrenia: Intensity effects and error pattern. American Journal of Psychiatry, 160(10), 1768–1774. Kucharska-Pietura, K., Phillips, M. L., Gernand, W., & David, A. S. (2003). Perception of emotions from faces and voices following unilateral brain damage. Neuropsychologia, 41(8), 1082–1090. Lee, S. J., Lee, H.-K., Kweon, Y.-S., Lee, C. T., & Lee, K.-U. (2009). The impact of executive function on emotion recognition and emotion experience in patients with schizophrenia. Psychiatry Investigation, 6(3), 156–162. doi:10.4306/ pi.2009.6.3.156. Leppanen, J. M., Milders, M., Bell, J. S., Terriere, E., & Hietanen, J. K. (2004). Depression biases the recognition of emotionally neutral faces. Psychiatry Research, 128(2), 123–133. doi:10.1016/j.psychres.2004.05.020. MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7(1), 83–104. doi:10.1037//1082-989x.7.1.83. Matsumoto, D. (2007). Emotion judgments do not differ as a function of perceived nationality. International Journal of Psychology, 42(3), 207–214. doi:10.1080/ 00207590601050926. McDonald, S., & Flanagan, S. (2004). Social perception deficits after traumatic brain injury: Interaction between emotion recognition, mentalizing ability, and social communication. Neuropsychology, 18(3), 572–579. doi:10.1037/08944105.18.3.572. McDonald, S., & Saunders, J. C. (2005). Differential impairment in recognition of emotion across different media in people with severe traumatic brain injury. Journal of the International Neuropsychological Society, 11(4), 392–399. doi:10.1017/s1355617705050447. Milders, M., Fuchs, S., & Crawford, J. R. (2003). Neuropsychological impairments and changes in emotional and social behaviour following severe traumatic brain injury. Journal of Clinical and Experimental Neuropsychology, 25(2), 157–172. Milders, M., Ietswaart, M., Crawford, J. R., & Currie, D. (2008). Social behavior following traumatic brain injury and its association with emotion recognition,
understanding of intentions, and cognitive flexibility. Journal of the International Neuropsychological Society, 14(2), 318–326. doi:10.1017/s1355617708080351. Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879–891. Radice-Neumann, D., Zupan, B., Babbage, D. R., & Willer, B. (2007). Overview of impaired facial affect recognition in persons with traumatic brain injury. Brain Injury, 21(8), 807–816. doi:10.1080/02699050701504281. Rolls, E. T., & Grabenhorst, F. (2008). The orbitofrontal cortex and beyond: From affect to decision-making. Progress in Neurobiology, 86(3), 216–244. doi:10.1016/j.pneurobio.2008.09.001. Soussignan, R., Ehrle, N., Henry, A., Schaal, B., & Bakchine, S. (2005). Dissociation of emotional processes in response to visual and olfactory stimuli following frontotemporal damage. Neurocase, 11(2), 114–128. doi:10.1080/ 13554790590922513. Spell, L. A., & Frank, E. (2000). Recognition of nonverbal communication of affect following traumatic brain injury. Journal of Nonverbal Behavior, 24(4), 285–300. Walz, N. C., & Benson, B. A. (1996). Labeling and discrimination of facial expressions by aggressive and nonaggressive men with mental retardation. American Journal on Mental Retardation, 101(3), 282–291. Wechsler, D. (1981). Manual for the wechsler adult intelligence scale—Revised. New York: Psychological Corporation. Westermann, S., & Lincoln, T. M. (2010). Using signal detection theory to test the impact of negative emotion on sub-clinical paranoia. Journal of Behavior Therapy and Experimental Psychiatry, 41(2), 96–101. doi:10.1016/j.jbtep.2009.10.007. Wilson, B. A., Alderman, N., Burgess, P. W., Emslie, H., & Evans, J. J. (1996). Behavioural assessment of the dysexecutive syndrome (BADS). St. Edmunds, UK: Thames Valley Test Company. Zald, D. H., & Andreotti, C. (2010). Neuropsychological assessment of the orbital and ventromedial prefrontal cortex. Neuropsychologia, 48(12), 3377–3391. doi:10.1016/j.neuropsychologia.2010.08.012.
