Cognitive abnormalities and schizophrenic symptoms.

It is clearly important to construct a model of schizophrenic dysfunction which is able to integrate both its neural and perceptual/cognitive aspects. Although information-processing models appear useful as a means of linking biological and social factors relevant to the disorder, 'there remains a gap between the computer terminology in which attentional theories are couched and the patients's experience of schizophrenia' (Anscombe, 1987, p. 291). There are further difficulties with this approach. First, there is no agreed model of normal cognitive functioning and, as Shallice (1988) notes 'large-scale information theories are very loosely characterized; adding a connection, a constraint, or another subsystem to a model is unfortunately only too easy' (p. 321). Secondly, it is difficult to relate such research to biological models. Thirdly, schizophrenics perform poorly on most tasks; to rule out an explanation in terms of a 'generalized deficit', one requires the demonstration of a differential pattern of performance uncontaminated by task difficulty, or, preferably, schizophrenics' superior performance. In considering schizophrenics' disturbances of perception and cognition, a number of theoretical models have been drawn upon, and these often differ considerably in their assumptions concerning the nature of normal information processing. It is clearly hazardous to attempt to interpret studies in a different framework from those in which they were designed. However, I have argued (Hemsley, 1987) that, in part, models such as those of Broadbent (1971), Posner (1982) and Shiffrin & Schneider (1977), illustrate the way in which the spatial and temporal regularities of past experience influence the processing and, more speculatively, awareness, of current sensory input. A review of research on schizophrenics' cognitive abnormalities employing such models suggests that 'it is a weakening of the influence of stored memories of regularities of previous input on current perception which is basic to the schizophrenic condition' (Hemsley, 1987, p. 182). It is not claimed that the 'memories of past regularities' are not stored, nor that they are inaccessible. Rather the suggestion is that the rapid and automatic assessment of the significance, or lack of significance, of aspects of sensory input (and their implications for action) is impaired. The awareness of redundant information would therefore be affected. It should hence be possible to construct tasks on which schizophrenics perform better than normals due to the latter forming expectancies which are inappropriate to the new context in which a stimulus is presented. A disturbance of this kind is intriguingly reminiscent of earlier descriptions of the ways in which schizophrenics' perceptions and/or thinking are disturbed. As Cutting (1989) points out, Matussek (1952) and Conrad (1958) were among the first to argue that the early stage of schizophrenia could be explained in terms of a breakdown of Gestalt perception, and that such a disturbance could form the basis of delusional perception. Matussek (1952) describes a patient who was aware o f ' a lack of continuity of his perceptions both in space and over time' (cf. failure to make use of spatial and temporal redundancy): 'He saw the environment only in fragments. There was no appreciation of the whole. He saw only details against a meaningless background' (p. 92). Arieti (1966) later used the term perceptual and apperceptual fragmentation for such phenomena. In similar vein, Shakow's experimental work had led him to the conclusion that a schizophrenic 'can't see the wood for the trees and examines each tree with meticulous care' (Shakow, 1950, p. 386). Recent research relevant to Hemsley's (1987) formulation has drawn on animal learning theory in devising experimental paradigms. This has become increasingly 'cognitive' and the setting up of a viable animal model for schizophrenia no longer seems an impossibility. Two behaviourial