Elsevier

Biological Psychology

Volume 58, Issue 3, December 2001, Pages 203-227
Biological Psychology

A generalised deficit can account for problems in facial emotion recognition in schizophrenia

https://doi.org/10.1016/S0301-0511(01)00114-4Get rights and content

Abstract

Neuroimaging research has shown localised brain activation to different facial expressions. This, along with the finding that schizophrenia patients perform poorly in their recognition of negative emotions, has raised the suggestion that patients display an emotion specific impairment. We propose that this asymmetry in performance reflects task difficulty gradations, rather than aberrant processing in neural pathways subserving recognition of specific emotions. A neural network model is presented, which classifies facial expressions on the basis of measurements derived from human faces. After training, the network showed an accuracy pattern closely resembling that of healthy subjects. Lesioning of the network led to an overall decrease in the network's discriminant capacity, with the greatest accuracy decrease to fear, disgust and anger stimuli. This implies that the differential pattern of impairment in schizophrenia patients can be explained without having to postulate impairment of specific processing modules for negative emotion recognition.

Introduction

Since it was first shown that patients with schizophrenia are impaired in categorising facial stimuli of different emotional expressions (Shannon, 1970, Dougherty and Izard, 1974), research interest in face processing abilities in schizophrenia has increased. The reasons for this are twofold. First, it has been suggested that this deficit may have important clinical consequences (e.g. misreading social cues may contribute to social avoidance, delusion formation, etc.). Second, if dissociations exist between patients’ performance on facial emotion processing and other related cognitive functions (i.e. recognition of facial identity; categorisation of other complex visual stimuli, etc.), this could offer a valuable window into the relationship between cognitive and emotional aspects of psychological functioning (Mandal et al., 1998, Leventhal and Scherer, 1987).

Over the last three decades a wealth of data has emerged pertaining to the ability of patients with schizophrenia to identify emotions. This work has focused on the ability to categorise facial stimuli as well as verbal and postural cues of emotional state. The following review is restricted to studies where the categorisation or recognition of facial expression of emotion has been a principal aim of the experimental method. Overall, these studies have shown that schizophrenia patients are impaired in recognising facial emotion. However, debate continues regarding whether the observed impairment represents a generalised performance deficit, a problem in the processing of facial stimuli, or a deficit in emotion recognition (Bryson et al., 1997, Mandal et al., 1998).

Integration of the findings in this area is problematic because differences between studies have often been greater than their similarities (Kerr and Neale, 1993). Differences occur in subject and stimulus-related variables, and methodological approach. Stage of illness, medication status, duration of illness, clinical subtyping, age and composition of clinical and control groups have all varied. Studies have also differed in terms of the emotions that have been examined, both in number and valence. For example, some studies have considered only happy, sad and neutral (Mandal et al., 1999b); others have examined seven universally recognised emotions (Bryson et al., 1997); still others have included emotions such as shame, curiosity, etc. (Walker et al., 1980). Methodological variations across studies have spanned a wide spectrum. These include posed versus natural emotional stimuli, duration of stimulus presentation, still versus motion stimuli, response requirements (e.g. free response, forced-choice, match/non-match), and whether or not a differential deficit design (Chapman and Chapman, 1978) was used. Further variations include use of other neuropsychological measures as covariates and examination of performance differences in relation to emotional valence.

Despite these problems, studies using differential deficit designs and distinct categories of emotion can give a relatively coherent picture of the deficit. The main issue is how this deficit may be interpreted.

Section snippets

Early research in emotion recognition

Early studies established that patients with schizophrenia had a deficit in the recognition of facial emotion (Dougherty and Izard, 1974, Muzekari and Bates, 1977, Shannon, 1970). Although the results were relatively consistent, interpretation is actually difficult. The major criticism has been the failure to adopt a differential deficit design (Chapman and Chapman, 1978). Because there have been no comparisons between the tasks used in these studies and others involving similar or related

Differential deficit design studies

Most of this research has used facial recognition as a control task (Addington and Addington, 1998, Salem et al., 1996, Novic et al., 1984), but some studies have also used age discrimination based on face stimuli (Gessler et al., 1989, Habel et al., 2000, Schneider et al., 1995). At least one study has addressed the hypothesis that the deficit may be an impairment in labelling rather than discrimination (Walker et al., 1984). Emotion matching or emotion discrimination tasks have often been

Selection of an appropriate modelling strategy

Feedforward neural networks, or multi-layered perceptrons using the backpropagation learning algorithm have been widely used on tasks of categorisation (Rumelhart and McClelland, 1986). The basic task fulfilled by such networks is classification through the transformation of an input vector into an output vector on the basis of weighted associations between the activation of processing units. Here it was felt that this type of neural network architecture was suitable as a model of degraded

Results

The threshold for statistical significance was set at the P<0.001 level as a partial control for the number of tests.

Discussion

The simple neural network model presented in this paper is capable of learning to discriminate facial emotions on the basis of 12 measures of salient relationships between configurational features of the face derived from real facial expressions. The network exhibited a similar pattern of relative accuracy to different emotions as that seen in healthy subjects. Moreover, when lesioned, the network's performance was degraded such that the consequence of lesioning on measures of accuracy was most

Acknowledgements

The authors would like to thank the Neuroscience Institute for Schizophrenia and Allied Disorders (NISAD) for their support, and Terry Lewin for his helpful comments regarding data analysis and preparation of the manuscript.

References (66)

  • I. Pilowsky et al.

    Schizophrenia and the response to facial emotions

    Compr. Psychiatry

    (1980)
  • I. Pilowsky et al.

    The classification of facial emotions — a computer based taxonomic approach

    J. Affect. Disord.

    (1994)
  • F. Schneider et al.

    Emotional processing in schizophrenia — neurobehavioural probes in relation to psychopathology

    Schizophr. Res.

    (1995)
  • F. Schneider et al.

    Functional MRI reveals left amygdala activation during emotion

    Psychiatry Res. Neuroimaging

    (1997)
  • M. Tovee

    Face processing: Getting by with a little help from its friends

    Curr. Biol.

    (1998)
  • J. Archer et al.

    Face processing in psychiatric conditions

    Br. J. Clin. Psychol.

    (1992)
  • Archer, J., Hay, D., Young, A., 1994. Movement, face processing and schizophrenia: evidence of a different deficit in...
  • F. Benes et al.

    Reduced neuronal size in posterior hippocampus of schizophrenic patients

    Schizophr. Bull.

    (1991)
  • J. Borod et al.

    Perception of facial emotion in schizophrenia and right-brain damaged patients

    J. Nerv. Ment. Dis.

    (1993)
  • J.D. Cohen et al.

    Context-processing deficits in schizophrenia: Converging evidence from three theoretically motivated cognitive tasks

    J. Abnorm. Psychol.

    (1999)
  • D. Damasio

    Face agnosia and the neural substrates of memory

    Annu. Rev. Neurosci.

    (1990)
  • Dolan, R., Fletcher, P., Morris, J., Kapur, N., Deakin, J., Frith, C., 1996. Neural activation during covert processing...
  • F.E. Dougherty et al.

    Responses of schizophrenics to expressions of the fundamental emotions

    J. Clin. Psychol.

    (1974)
  • Driscoll, M., 1995. A study of facial expression using neural networks. Unpublished Master's dissertation, University...
  • P. Ekman et al.

    Pictures of Facial Affect

    (1976)
  • P. Ekman et al.

    Emotion of the Human Face: Guidelines for Research and an Integration of Findings

    (1972)
  • P. Falkai et al.

    Cell loss in the hippocampus of schizophrenics

    Eur. Arch. Psychiatry Neurol. Sci.

    (1986)
  • T. Feinberg et al.

    Facial discrimination and emotional recognition in schizophrenia and affective disorders

    Arch. Gen. Psychiatry

    (1986)
  • Ferdowsi, S., 1983. The processing of facial information in schizophrenia. Unpublished doctoral thesis, University of...
  • K. Friston

    Schizophrenia and the disconnection hypothesis

    Acta Psychiatr. Scand.

    (1999)
  • S. Gessler et al.

    Schizophrenic inability to judge facial emotion: A controlled study

    Br. J. Clin. Psychol.

    (1989)
  • E. Halgren et al.

    Location of human face-selective cortex with respect to retinotopic areas

    Hum. Brain Mapp.

    (1999)
  • P. Harrison

    The neuropathology of schizophrenia — A critical review of the data and their interpretation

    Brain

    (1999)
  • Cited by (93)

    • Facial recognition during early motherhood: Investigating the persistence of age and affect biases

      2020, Acta Psychologica
      Citation Excerpt :

      The second bias is linked to affect; it has been repeatedly observed that, when compared to sad faces, happy faces are processed more quickly and holistically. This is partially thought to be explained by the unique changes in facial features for happy faces in comparison to other emotional expressions, as happy features tend to have less overlap with other emotions (Johnston et al., 2001) and can also stand alone as a representation of happiness (e.g., a smile; Adolphs, 2002). The happy face advantage follows a larger general trend, in which positive stimuli are more quickly categorized than negative stimuli, even in the case of words or non-social pictures (e.g., Matlin & Stang, 1978).

    • Comprehensive review: Computational modelling of schizophrenia

      2017, Neuroscience and Biobehavioral Reviews
    • Attentional biases in high social anxiety using a flanker task

      2016, Journal of Behavior Therapy and Experimental Psychiatry
    View all citing articles on Scopus
    View full text