Exploring the nature of facial affect processing deficits in schizophrenia

Abstract

Schizophrenia has been associated with deficits in facial affect processing, especially negative emotions. However, the exact nature of the deficit remains unclear. The aim of the present study was to investigate whether schizophrenia patients have problems in automatic allocation of attention as well as in controlled evaluation of facial affect. Thirty-seven patients with schizophrenia were compared with 41 control subjects on incidental facial affect processing (gender decision of faces with a fearful, angry, happy, disgusted, and neutral expression) and degraded facial affect labeling (labeling of fearful, angry, happy, and neutral faces). The groups were matched on estimates of verbal and performance intelligence (National Adult Reading Test; Raven's Matrices), general face recognition ability (Benton Face Recognition), and other demographic variables. The results showed that patients with schizophrenia as well as control subjects demonstrate the normal threat-related interference during incidental facial affect processing. Conversely, on controlled evaluation patients were specifically worse in the labeling of fearful faces. In particular, patients with high levels of negative symptoms may be characterized by deficits in labeling fear. We suggest that patients with schizophrenia show no evidence of deficits in the automatic allocation of attention resources to fearful (threat-indicating) faces, but have a deficit in the controlled processing of facial emotions that may be specific for fearful faces.

Introduction

From the first descriptions of schizophrenia the apparent emotional disturbances, such as emotional flattening or blunted affect are regarded as a core deficit of schizophrenia (Bleuler, 1911). In addition, more recent research demonstrated deficits in the processing of emotional material including faces, pictures, verbal information, and other symbolic information in schizophrenia. Indeed, the deficits in facial affect recognition are one of the main findings of the last decade and are now widely acknowledged (Schneider et al., 1995, Mueser et al., 1996, Wölwer et al., 1996, Addington and Addington, 1998, Habel et al., 2000, Köhler et al., 2000, Penn et al., 2000, Pinkham et al., 2003). Moreover, a growing body of research indicates that the emotion processing deficits in schizophrenia are stronger for the recognition of negative facial affect (Mandal et al., 1998, Edwards et al., 2002, Köhler et al., 2003), such as fearful faces (Gaebel and Wölwer, 1992, Archer et al., 1994). These facial affect-processing deficits may be linked to amygdala disturbances in schizophrenia. The amygdala, as a key area for the processing of negative emotional information (Morris et al., 1996) seems deficient both structurally and functional in schizophrenia, which might result in emotion processing deficits (Gur et al., 2002).

Yet, the exact nature of the facial emotion-processing deficit in schizophrenia remains unclear. Currently, the prevalent idea seems to be that schizophrenia patients have general deficits in emotional processing (Shayegan and Stahl, 2005). That is, no differentiation is made between different processes or components, with the exception of valence, i.e. the recognition of negative emotions has been reported to be more affected than positive emotions (Mandal et al., 1998). However, automatic and controlled evaluative processing of affect have been distinguished as important components of emotional processing (LeDoux, 1996, Cunningham et al., 2003). Automatic processes are thought to be primarily involved in the generation of quick evaluative judgements. For instance, valence detection on a positive–negative dimension is an automatic evaluative process often used in priming paradigms (Hermans et al., 1994, Rossell et al., 2001, Klauer and Musch, 2003). In turn, controlled processes are involved in processing more complex information, such as explicit emotion recognition or labeling (Gorno-Tempini et al., 2001, Winston et al., 2003). Notably, there is also evidence for a differential neural basis for these two distinct kinds of emotional processing. Whereas automatic emotional processing of facial expressions has been shown to depend primarily on right hemispheric resources (Hartikainen et al., 2000) including the right amygdala (Markowitsch, 1998, Gläscher and Adolphs, 2003), the labeling of facial emotions has been shown to depend on the left hemisphere (Young et al., 1993, Stone et al., 1996). To our knowledge, the difference between automatic and controlled facial affect processing has not yet been contrasted in a single study.

Previous research on automatic processing of schizophrenia resulted in mixed findings. For instance, schizophrenia patients employed abnormal visual scan paths when viewing faces (Phillips and David, 1997). In addition, patients with schizophrenia showed a reduced left perceptual bias in response to emotional faces (Gooding et al., 2001). On the other hand, Hoschel and Irle (2001) reported stronger emotional priming in schizophrenia when primed with an unpleasant face, reflecting a stronger influence of automatically processed emotional stimuli on judgements. These findings were extended by Suslow et al. (2003a) who demonstrated that in an affective priming task schizophrenia patients were more sensitive to subliminal negative facial affect than controls and perceive positive facial affect as aversive or unpleasant. In addition, Suslow et al. (2005) showed in a sequential affective priming task based on faces that schizophrenia patients without affective negative symptoms showed comparable priming effects as controls, whereas patients suffering from flat affect or anhedonia manifested only a prime effect due to negative facial valence. Finally, Kring et al. (1999) used electromyography to measure facial responsivity to affective faces and found that schizophrenia patients showed greater corrugator reactivity than healthy controls. Moreover, support for augmented automatic affect processing in schizophrenia comes from studies that demonstrated the presence of a threat-related bias, i.e. a predisposition for threatening stimuli already present in the early stages of processing, in people with persecutory delusions (Blackwood et al., 2001). This is consistent with a recent model of emotion processing in schizophrenia (Aleman and Kahn, 2005), in which a neural basis was proposed for the paradox of impaired explicit emotion perception and expression in the face of normal or heightened levels of emotional reactivity. Following this line of reasoning patients with schizophrenia could show normal or increased automatic attention to facial affect, but show deficits in controlled processing. However, to our knowledge there are no studies that investigate both processes together in schizophrenia on a behavioral level.

The aim of the present study was to further investigate the nature of the facial affect processing deficits in schizophrenia by examining incidental attention allocation processes, i.e. Stroop-like interference of facial affect and controlled evaluation of facial affect. Automatic allocation of attention to facial affect (Vuilleumier, 2005) was tested with an incidental facial affect-processing task in which participants made a gender judgement concerning affective faces. We hypothesized that patients with schizophrenia would show a normal incidental processing of threat-related facial affect, i.e. expected interference for fearful faces compared to neutral faces. Controlled evaluation of facial affect was measured using degraded, low-pass filtered pictures of affective faces with different emotional intensities since low spatial frequency faces might convey global information for coarse emotional recognition. We hypothesized that schizophrenia would be associated with difficulties in the labeling of threat-related faces, in particular fearful faces (Edwards et al., 2001).

Both involuntary attention allocation to faces and controlled evaluation of facial affect were tested in a highly educated, stable patient sample with low levels of symptomatology to eliminate confounding effects, such as acute phase of the illness or severe symptoms to the maximum.