تعامل احساسات شناختی در اسکیزوفرنی: اثرات ضمنی و صریح حالت صورت

Abstract Working memory (WM) and emotion classification are amongst the cognitive domains where specific deficits have been reported for patients with schizophrenia. In healthy individuals, the capacity of visual working memory is enhanced when the material to be retained is emotionally salient, particularly for angry faces. We investigated whether patients with schizophrenia also have an enhanced WM capacity for angry faces. We compared 34 inpatients with schizophrenia and 34 age-, handedness- and gender-matched control participants in three separate tasks. In the WM task, participants saw two faces with angry, happy or neutral emotional expressions for 2 s and had to decide whether a probe face presented after a 1 s delay was identical to one of them. In the emotion classification task, they had to assign these faces to the appropriate categorical emotion. They also rated faces for valence and arousal. Although patients performed generally worse on the working memory task, they showed the same benefit for angry faces as control participants. However, patients were specifically impaired for angry faces on the emotion classification task. These results indicate preserved implicit emotion processing in schizophrenia patients, which contrasts with their impairment in explicit emotion classification. With regard to clinical practice, our findings underline the importance of assessing responsiveness to emotions in patients with schizophrenia, with a view possibly to utilize preserved implicit emotion processing in cognitive remediation programs.

1. Introduction The ability to retain socio-emotional information in working memory (WM) is fundamental to normal social cognition. Visual WM is a limited capacity memory store that has been shown to be able to retain approximately two faces at any one time (Jackson & Raymond, 2008). Effective storage of facial information in visual WM is essential for fluid and coherent face-to-face communication. The majority of social interactions take place within an emotional context, and previous research has found that facial expression modulates the degree to which face identities are successfully remembered: WM capacity is greater for angry compared to happy or neutral faces (Jackson et al., 2008 and Jackson et al., 2009). This ‘angry face benefit’ in WM might serve to ensure a timely and appropriate response when one's personal safety is in threat, for example preparation of approach or avoid reactions. Working memory deficits in schizophrenia have been widely reported (Silver, Feldman, Bilker, & Gur, 2003). They affect both the verbal and the visuospatial modality (Reichenberg & Harvey, 2007) and often precede clinical symptoms (Hambrecht, Lammertink, Klosterkoetter, Matuschek, & Pukrop, 2002). Working memory deficits are amongst the core candidates for neurocognitive trait markers of the disorder, supported by their presence in first-degree relatives (Park, Holzman, & Goldman-Rakic, 1995), especially when demand on executive processes is high (Conklin, Curtis, Calkins, & Iacono, 2005). Deficits of working memory and other executive functions may be related to negative symptoms, psychosocial deficits and poor prognosis (Twamley, Palmer, Jeste, Taylor, & Heaton, 2006) and have been recognized as a target for both pharmacological (O’Grada & Dinan, 2007) and psychological (Greenwood, Landau, & Wykes, 2005) interventions. Because deficits in WM remain even after symptom improvement (Snyder et al., 2008), a better understanding of the nature of WM deficits and their underlying brain mechanisms is needed in order to improve treatment strategies. Emotion recognition, particularly classification of emotional faces, is another domain where deficits have been reported in schizophrenia (Kucharska-Pietura et al., 2005, Pinkham et al., 2008 and Van’t Wout et al., 2007). Several emotion classification studies indicate that emotion processing deficits in schizophrenia are more pronounced for negative facial affects (Koehler et al., 2003 and Mandal et al., 1998). Unlike the working memory deficit, where the correlation with symptoms has been inconsistent, emotion recognition deficits were correlated with social impairment (Hooker & Park, 2002). While dysfunction of the frontal lobes (Perlstein, Carter, Noll, & Cohen, 2001) has been implicated in the visual WM deficits, activation changes in an emotion circuit involving the superior temporal sulcus (Michalopoulou et al., 2008) and limbic areas (Fakra, Salgado-Pineda, Delaveau, Hariri, & Blin, 2008) may underlie the emotion classification deficit. In light of the difficulties of developing remediation strategies for these cognitive deficits, it would be interesting to probe whether patients suffering from schizophrenia are still responsive to implicit emotion cues that may be utilized to obtain cognitive benefits. Such implicit emotion benefits have mainly been reported for threat related stimuli, probably because of their higher evolutionary salience. For example angry faces were preferentially processed in visual search (Fox et al., 2000) and other selective attention tasks. This preference for angry faces correlated with social phobia (Mogg, Philippot, & Bradley, 2004) and anxiety (Mogg, Garner, & Bradley, 2007). We have recently developed a task that probes implicit effects of emotion on WM and showed a benefit for angry faces. Functional magnetic resonance imaging (fMRI) revealed that this angry benefit in face WM was supported by activation of the globus pallidus (Jackson et al., 2008). In the present study we investigated whether schizophrenia patients – although impaired in overall WM performance and emotion classification – would still show the same angry faces benefit in WM as healthy individuals.

Results 3.1. Working memory The ANOVA of the d′ values of the working memory task showed main effects of emotion (F(2,132) = 5.79, p = 0.004) and group (F(1,66) = 15.51, p < 0.001), but no interaction between emotion and group (F(2,132) = 0.83, p = 0.438). Pairwise comparisons revealed that the main effect for emotion was driven by significantly higher performance for angry compared to happy (t(67) = 2.71, p = 0.009) and neutral (t(67) = 3.20, p = 0.002) faces. The effects remained significant after Bonferroni correction (corrected alpha level = 0.016). The difference between happy and neutral faces was non-significant (t(67) = 0.56, p = 0.577). The main effect for group reflected the lower overall WM performance amongst patients than controls ( Fig. 2). Performance of patients (white) and healthy control participants (grey) on the ... Fig. 2. Performance of patients (white) and healthy control participants (grey) on the working memory task for angry, happy and neutral faces, measured in dprime values (y-axis) and their standard errors of the mean. The histogram shows an angry benefit for both groups and impaired performance across conditions for the patient group. Figure options Reaction times were slower in the patient group (main effect of group: F(1,66) = 25.23, p < 0.001), but there was no significant interaction with emotion (F(2,132) = 0.82, p = 0.441) and no main effect of emotion (F(2,132) = 0.19, p = 0.825). 3.2. Emotion classification The ANOVA of the accuracy scores on the emotion classification task showed a main effect of emotion (F(1.62,92.54) = 32.89, p < 0.001), an interaction between the factors emotion and group (F(1.62,92.54) = 4.96, p = 0.014), but no group effect (F(1,57) = 1.001, p = 0.321). The main effect of emotion was driven by better classification accuracy for happy faces compared to angry (t(58) = −8.58, p < 0.001) and neutral (t(56) = 6.87, p < 0.001) faces ( Fig. 3). The interaction between emotion and group could be explained through the impaired classification of angry faces in the schizophrenia group compared to controls (t(57) = −2.30, p = 0.025) ( Fig. 3). Performance of patients (white) and healthy control participants (grey) on the ... Fig. 3. Performance of patients (white) and healthy control participants (grey) on the emotion recognition task for angry, happy and neutral faces, measured in recognition rate (percent correct) (y-axis) and their standard errors of the mean. The histogram shows better recognition for happy compared to angry and neutral faces and impaired recognition of angry faces in the schizophrenia group. Figure options 3.3. Arousal and valence ratings Patients and controls scored almost identically on valence ratings. The ANOVA over the data from the 34 controls and 25 patients showed a main effect of emotion (F(1.47,84.03) = 136.79, p < 0.001). Neither the main effect for group (F(1,57) = 0.04, p = 0.85) nor the interaction between the two factors (F(1.47,84.03) = 0.17, p = 0.77) were significant. As would be expected, valence ratings were significantly more pleasant for happy than angry or neutral faces, and more pleasant for neutral than angry faces (all p's < 0.001). Arousal ratings were again rather similar across groups (34 controls and 25 patients). The ANOVA showed a main effect of emotion (F(1.81,103.03) = 21.22, p < 0.001), but no main effect of group (F(1,57) = 0.75, p = 0.11) and no interaction (F(1.81,103.03) = 0.86, p = 0.29). The main effect of emotion was driven by the high arousal ratings for angry (significantly higher than happy (p = 0.007) and neutral (p < 0.001)) and significantly higher arousal ratings for happy compared to neutral faces (p = 0.001). 3.4. Performance correlation with medication levels Pearson correlation between angry benefit and d′ scores for angry, happy and neutral WM performance and CPZ equivalent in the patient group revealed no significant correlation with medication levels (all r's ≤ 0.1, all p's > 0.1). 3.5. Covariance analysis with NART In order to rule out the possibility of the lower performance of the patient group on the working memory task was caused by differences in psychometric intelligence we computed its covariance with the NART score (the NART was not obtained in 2 patients and 4 healthy participants, as they were not native speakers). The main effect of group remained significant (F(1,59) = 4.70, p = 0.033).