Eye movements reflect impaired face processing in patients with schizophrenia

Abstract

Background: Impaired processing of faces in patients with schizophrenia may underlie aspects of disturbance in their social interaction. This study examined patterns of eye fixation in subjects with schizophrenia and non-psychiatric controls, while processing a high resolution picture of a neutral face and a nonbiological complex geometric stimulus.

Methods: Ten-second sequences of eye movement were recorded video-oculographically (50 samples/sec) while subjects were “free-viewing” the stimuli. An essential element of the study was customized software that ensured stimulus presentation on a video display only after subjects were fixated upon a centre-screen cue, so that all subjects began stimulus processing from the same point.

Results: Compared with the control group, subjects with schizophrenia exhibited reduced scanpath lengths and a tendency toward fewer fixations for the face stimulus. They also showed an initial relative right spatial hemineglect (for the first voluntary fixation) when viewing the Rey figure, but not when viewing the face stimulus. Overall, there were no significant differences between the schizophrenia and control groups in the lateral distribution of subsequent fixations for either stimulus.

Conclusions: Disturbed spatial and temporal patterns of eye movement in some people with schizophrenia may reflect sub-optimal processing of face stimuli, that may predispose these individuals to dysfunctional interpretation of facial communication cues.

Introduction

Numerous studies have examined smooth pursuit eye movement dysfunction in schizophrenia (for example, Lipton et al 1983, Gaebel 1989, Abel et al 1992, Radant and Hommer 1992). There are, however, relatively few studies that have explored eye movements to complex stimuli in subjects with schizophrenia. Eye movement records provide a real-time, relatively high temporal resolution index of visual information processing. A close interplay between the eye and the brain allows localization and recognition of information in the environment. The point of gaze is shifted rapidly, by means of saccadic eye movements, so that the fovea (the small high-acuity central region of the retina) “fixates” upon salient features in the environment. Although the fovea provides the visual system with input of stimulus detail, less detailed information from the parafoveal and peripheral retina is processed in parallel.

A number of studies have examined scanning eye movements and scanpaths (the path traced by successive fixation points, after Noton and Stark [1971]) in schizophrenia. Kurachi et al (1994) noted a decrease in number of fixations and scanpath length in subjects with schizophrenia compared with non-psychiatric controls for the WAIS picture completion test. Gaebel et al (1987) showed a reduction in overall fixation time (hyperscanning) in schizophrenics with prominent positive symptoms and an increase in fixation duration (hyposcanning) associated with negative symptoms, in response to processing a line drawing of a social scene.

The specific disturbances underlying dysfunctional eye movements in schizophrenia are unknown. Dysfunctions have been postulated in the integration of information Schwartz-Place and Gilmore 1980, Wells and Leventhal 1984, Griffith et al 1980, Frith et al 1983, including deficits in temporal integration (Hemsley 1996) short-term visual memory (Knight et al 1985) and spatial working memory Park and Holzman 1992, Park and Holzman 1993. A lateralised dysfunction in the control of attention in schizophrenia has been postulated. Posner et al (1988) demonstrated in a prestimulus cuing task that when contralaterally cued, schizophrenics responded more slowly to targets appearing in the right than in the left visual field, and proposed a difficulty in disengaging visual attention mediated by the left parietal lobe. Harvey et al (1993) reported a relative right sensory hemineglect in a rod centering task, with a correlation between degree of spatial asymmetry and severity of schizophrenia symptoms. Maruff et al (1995), although generally confirming the results of Posner et al (1988), also found a reduction in the degree of right-spatial hemineglect in patients with prominent positive symptoms and in patients on long-term antipsychotic medication. Early et al (1989) proposed a model (based partially on the findings of Posner et al 1988) that predicted an analogous right visual hemineglect in schizophrenia.

Faces, by virtue of their importance in interpersonal communication, seem to be processed at least in part by specific neural networks Geschwind 1979, Sergent et al 1994, Puce et al 1995, Phillips et al 1997, that may ultimately impinge on allocation of visual attention and control of eye movement. Both visual attention and eye movement paradigms have been used to investigate lateral hemispheric differences in face processing in both normals and schizophrenics. Using free-viewing visual attention tasks, David (1993) found that normal right-handers show a significant perceiver bias toward the left hemiface (relative to the viewer) when judging facial expression. This is consistent with the proposal that each hemisphere generates a contralateral attentional bias and the rightward bias of the left hemisphere is stronger (Reuter-Lorenz et al 1990), given that the right hemisphere is principally implicated in face processing (e.g., Silberman and Weingartner 1986). In diagrammatic chimeric face studies, schizophrenics exhibited a reduced left-hemiface attentional bias (compared with controls), possibly mediated by right hemisphere hypofunction David and Cutting 1990, David 1993.

Several eye movement studies, using face stimuli, have also revealed processing differences in schizophrenia. For freely viewed photographic face stimuli Phillips and David (1997a) found that normal controls showed a bias toward the left hemi-face for the first fixation, while schizophrenics showed no such bias (or even a slight preference for the right hemi-face). Similarly, for chimeric faces schizophrenics showed no overall lateralization bias in eye movements compared with controls.

Other eye movement studies of face perception have analyzed complete visual scanpaths rather than the lateralised allocation of fixations in particular. In a previous study, our group (Gordon et al 1992) examined eye movements in response to a neutral face stimulus and found that schizophrenics had reduced fixation durations (compared with non-psychiatric control subjects) for facial features (eyes, nose, mouth) in the first three seconds of processing. Similarly, Phillips and David 1997b, Phillips and David 1998 found that deluded schizophrenics tend to make fewer and longer fixations than controls, and that these fixations are concentrated outside of salient facial features regions. Using facial affect stimuli, Streit et al (1997) observed shorter distances between fixations and longer fixation durations for schizophrenics compared with controls. In the current study, the first in a series of investigations designed to extend this research, we examined the patterns of eye movement and lateral distribution of fixations in response to a neutral face and a complex geometric stimulus in schizophrenia.

A high-resolution colour photograph of a neutral-affect face was selected from a standardized facial stimulus set (Mazurski and Bond 1993), and the Rey Complex Geometric Figure(Rey 1942) was used as a control stimulus. The latter is a well-known standard figure with no overall semantic meaning and is widely used in neuropsychological testing. The experimental paradigm was extended to control for the initial point of gaze, so that all subjects commenced stimulus processing from the same central point. In a previous study with normal subjects (Manor et al 1995), it was found that the circle feature of the Rey figure consistently attracted the foveal gaze in the earliest phase of visual processing, regardless of the left/right orientation of the stimulus. This concurred with earlier studies (e.g., Loftus and Mackworth 1978), that suggested that after stimulus onset, the visual system rapidly obtained a ‘gist’ of a scene and the most relevant feature was processed first.

We hypothesized that, compared with controls, schizophrenics would show an analogous right-spatial hemineglect for the first and subsequent fixations for the geometric stimulus, dwell longer at the initial (controlled) fixation point before shifting the gaze to the next locus of fixation (i.e., longer “gist” duration), and make fewer, shorter fixations with reduced scanpath lengths. For the face stimulus, we expected that schizophrenics would not show a leftward bias for the first or subsequent fixations, and fixations would be fewer in number, of shorter duration and less extensively distributed (shorter scanpath lengths).