What contribution can theory and research in "information processing" make to our understanding of schizophrenia?

Schizophrenia affects approximately 1% of the world's population, making it the most common psychosis. It remains is the fourth leading cause of morbidity in both women and men in the United States.
"Schizophrenia is probably the most misused psychological term in existence" ( Carlson, 1995). Schizophrenia is diagnosed when a person suffers two or more of the following symptoms during one month: delusions, hallucinations, disorganised speech, grossly disorganised or catatonic behaviour and negative symptoms (e.g. poverty of speech, social withdrawal). The onset of the illness generally occurs at a later age in women (23-35 years) than in men (18-25 years). Subtypes of schizophrenia have been identified based on which symptoms are predominant. They are Disorganised, Paranoid, Catatonic,

Residual and Undifferentiated Types.
The current diagnostic guidelines are a very blunt tool. The Undifferentiated Type criterion is attacked as being a hold-all for patients awaiting the creation of a new diagnostic label. More extremely, there are those who contend that emotional withdrawal, primitive disinhibition, catatonia and delusions of persecution are all observed in teenagers. They would claim that this 'syndrome' is merely an artificial psychiatric construct.

More seriously, family studies have shown that many relatives of schizophrenics are schizotypal but do not have positive symptoms. Hence there is a large spectrum of symptoms and much argument about which to include or exclude. Cognitive performance scales are not decisive since some individuals (high schizotypals) without positive symptoms can perform similarly to diagnosed schizophrenics.

The pathological approach to schizophrenia fragmented this syndrome into numerous positive symptoms, deficits and lesions. A more cognitive approach presumes that schizotypal phenomena occur toward the abnormal end of the information processing spectrum. The information processing proposals that will be discussed are reduced latent inhibition, impaired semantic memory and theory of mind.

It is widely accepted that attention is focused on a very small selection of objects or events within our sensory scope to the exclusion of others. Therefore some processes must exist to filter out extraneous information. Cognitive studies have shown that sensory input firstly leads to an activation of related information. Inhibition then follows, with delay, (hence latent inhibition) which restricts the information reaching consciousness to relevant information only. For example the word 'cow' might activate 'animal', 'milk' and 'intimidate', a process called semantic activation. However latent inhibition may filter out the 'intimidate' association if the context was clearly 'animal' related. Kraeplin (1913) found that his patients schizophrenic patients found it impossible to keep their attention fixed on any one thing for any length of time. It appears that normal thoughts are disrupted by a clamour of irrelevant thoughts and perceptions. This has led to the idea that an inability to filter out extraneous information (latent inhibition) may be at work in schizophrenia.

Cognitive tests like the lexical decision tasks - LDT (visual) and dichotic studies (auditory tasks) have been used to test the role of latent inhibition (LI) and semantic activation in schizophrenia.
Manshrek et al (1988) used an LDT to test the theory of increased semantic activation. He found that schizophrenics with thought disorder had faster response times than non-thought disorder schizophrenics. Schizophrenics were faster at the task than normal and non-psychotic controls. Because his study was limited to a very small group of patients, Kwapil et al. attempted to replicate his findings using a larger sample. Their findings were similar: thought disorder schizophrenics showed semantic priming effects. However it is crucial to note that both studies used medicated patients.

En face, the reduced latent inhibition model (leading to semantic over-activation) seems to account for incoherence and thought disorder symptoms seen by schizophrenics. However, it has been challenged by contentions that reduced latent inhibition could be an effect of anti-psychotic medications administered to schizophrenics.

A double-blind study by Beech et al. (1990) found that medication (chlorpromazine) did have a normalising effect on latent inhibition. In fact medication facilitated LI in some cases. However the dosage of chlorpromazine used (25mg) was much lower than clinical dosages. More importantly, this study used a non-schizophrenic sample.

Williams et al. (1998) offer much needed clarification. Their experiment sought to determine the effect of Haloperidol on auditory LI. Anti-psychotic naïve schizophrenics and schizophrenics on medication for only two weeks prior were both given the same auditory task. It was found that the naïve group had intact auditory inhibition. The medicated group showed decreased auditory LI. From this one might conclude that medication did indeed cause reduced LI. However, these results must be contrasted with healthy controls.
Healthy volunteers on haloperidol showed reduced LI while healthy volunteers on saline did not. But only a high dose of haloperidol (1 mg) reduced LI. A lower dose of 0.5mg did not affect LI. This non-linearity between drug dosage and LI has been confirmed in other human and animal studies.

These findings by Williams et al are crucial. They debase the LI explanation of thought disorder. In fact they might be an indictment the iatrogenic nature of psychotropic drugs. These findings demonstrate the importance of testing medicated as well as un-medicated schizophrenics, non-psychotic patients and healthy controls, before reaching any conclusions. This principle should be applied equally in cognitive, post-mortem, biochemical and imaging investigations.

It raises some crucial questions. Given that acute medication caused reduced LI, why do chronically medicated patients exhibit normalised or even facilitated LI? Perhaps there is a measure of desensitisation to some drug effects. Furthermore, it seems the effects of anti-psychotics are dose-related and task-specific. After all Beech found that anti-psychotics increased LI in a visual task while Williams found the opposite effect in an auditory task.

How might we explain contradictory findings of increased semantic activation (Manshrek et al.) and semantic memory impairments (Aloia et al.) both observed in chronically medicated patients?
Aloia et al (1998) found consistent deficits on various tests of semantic processing in schizophrenics compared to normal controls. This is in direct contradiction to previous findings of semantic facilitation by Manshrek et al and others. Moreover, verbal fluency scores could be used to differentiate between schizophrenic groups with high (low scores) or low levels of thought disorder. Aloia et al infer from this an underlying disorganisation of the semantic system which manifested as both thought disorder and limited verbal fluency. Thought disorder schizophrenics also scored poorly on the Peabody Picture Vocabulary tests. This shows that they may have deficient information about the meaning of words. Perhaps pre-existing language limitations in the sample group may exacerbate thought disorder once illness begins. However, there is no evidence that the 'meaning' of words is lost during the course of illness.

The schizophrenic patients used in this study were mental hospital inpatients and received neuroleptic medications at the time of study. In a follow-up study Aloia et al (1998) compared semantic priming in high thought disorder schizophrenics, low thought disorder schizophrenics and normal controls. They divided word pairs into three categories based on low, medium and high degrees of association. In normal subjects priming was best for highly associated word pairs and insignificant for low-association pairs. This pattern was repeated in patients with low thought disorder. However, the high thought disorder group did not show priming at any level of association. They actually inhibited responses to high and medium associations. Hence they were less likely to recall closely related words. These results coincide with connectionist models of semantic disorganisation in schizophrenia.

Nestor et al. (1998) found that normal subjects exhibit best recall for words with a small number of associations combined with a high degree of associative strength (high connectivity small-network size). They showed worse recall for low-connectivity small-network size, then high-connectivity large-network size and lastly low-connectivity large network size. Hence the increasing size of networks impaired recall while increasing connectivity promoted recall.

Schizophrenics, however, showed an enhanced effect of connectivity but a similar effect of network size. They remembered more words of high-connectivity large network size than low-connectivity small network size. Nestor et al. suggest that this shows over-activation of strongly connected networks in schizophrenics. This over-activation may cause associative intrusions on consciousness.
Can increased semantic priming and impaired semantic memory co-exist?

Hodges, Salmon and Butters (1992) devised a comprehensive battery of tests dependent of semantic memory. McKenna et al. (1996) used this well validated battery to compare semantic memory between non-elderly schizophrenics with chronic severe illness, normal individuals and patients with Alzheimer's disease. The chronic schizophrenics performed worse than normal subjects in almost all tests, with the level of impairment sometimes approaching that of the Alzheimer's patients. McKenna et al further showed that the semantic memory impairment was disproportionate to overall intellectual impairment.

Evidence for semantic facilitation, semantic disorganisation or even semantic dysfunction has all been presented. How might this conflicting information be resolved? Considering only lexical tasks, one might conclude that thought disorder is best explained by over-activation of associative networks. However over-activation may lead to the awareness of far-flung associations to the exclusion of nearer associations.
Using a connectionist metaphor, the mind gains a global vision of the London tube system that allows awareness of the endpoints of various lines e.g. Edgware road. However the mind is unable to attend to finer details e.g. the stations on the Piccadilly line within zone one. This situation might be equally expressed as semantic memory impairments and semantic facilitation depending on what test paradigm used. Implicit in my proposal is the idea that conscious awareness can only hold finite amounts if information.

Can the reduced latent inhibition model explain reality distortions? Claridge proposes that delusions and hallucinations might be triggered by real sensory input. In non-schizophrenics, latent inhibition filters out most sensory input such that it only relevant information broaches consciousness. However in schizophrenics this irrelevant input intrudes upon consciousness. Since the inner narrative must be sensibly continuos, attempts to fit irrelevant input into the cognitive schemata results in delusion. The finding that paranoid delusional schizophrenics score higher on intelligence scales than non-paranoid schizophrenics (Posner and Boies, 1971) may support this view, if intelligence is a measure of the ability to maintain a coherent inner narrative.

The latent inhibition explanation of reality distortions is based on the assumption that preconscious screening occurs after sensory input is represented iconically. A cohesive model of relevant information must first be formed preconciously after which irrelevant information is identified and actively inhibited or passively decayed. Is irrelevant information inhibited at the graphemic, phonemic or semantic stage? Semantic priming has been activated by subliminal stimuli (Evans 1992) using Subliminal Activation Without Cognitive Identification (SAWCI) methods. This indicates that preconscious processing can reach the semantic level before inhibition takes place. Perhaps SAWCI can be modified to identify the level of preconscious representation of other sensory inputs e.g. objects, faces, spoken words.
On balance, LI impairments do not fully account for all positive symptoms of schizophrenia though they do contribute to our understanding of the Disorganised Type.

Claridge proposes that memory processes could be disrupted by anomalous sensory processing. Many patients report that onset of illness coincides with sensory aberrations. Patients recount feeling that they were changed somehow, or that people looked different, or the world seemed implacably different.

Perhaps an insult to the sensory system, the result of which is that sensory input now looks and feels different from before. The subject might respond with anxiety and begin to study objects, faces, etc, closely. The people in his/her environment may respond to scrutiny with avoidance or hostility. To the subject, these people who looked different now seem unfriendly. Anxiety creates an attention bias towards all stimuli perceived as threatening. Claridge extrapolates that this may lead to delusions and hallucinations.

There is undoubtedly an environmental role in augmenting the distress of schizophrenics. However Claridge's proposition is, in my view, weak because it does not account for the inability of subjects to disconfim paranoid or delusional experiences. It explains away hallucinations as mere imaginings.

Furthermore, people who suffer sensory insults often compensate quickly for any deficits e.g. proposagnostics become more sensitive to voices and behaviours than to faces.
Maher would contend that delusions are not disconfirmed because often they reduce the patient's confusion and anxiety. They become the key to explaining everything including anomalous sensory experiences. Yet Maher does not adequately account for auditory hallucinations? Are they imagined, like imaginary playmates, to keep anxious, lonely schizophrenics company? Or are controlling hallucinations really a pretext to carry out actions that might otherwise be unacceptable to the ego?

Reality distortions seem to be best accommodated by 'theory of mind' propositions. Firth has proposed a model explaining how deficiencies in theory of mind (ToM) can cause symptoms of schizophrenia.
Having a 'theory of mind' or the ability to 'mentalise' is the capacity to make representations of one's thoughts and beliefs. It is also the awareness that others have beliefs and some ability to infer what they might be. ToM has been studied greatly in relation to autism and is now providing insights into schizophrenia.

Dennett, in his analysis of how one might come to predict the behaviour of a complex system (e.g. a chess program) proposes that a strategy of 'intentionality' is most successful. Predicting the program behaviour works if beliefs and intentions are ascribed to it based on what an intelligent human might do in the same situation.

Evolutionary psychologist would argue that human beings have an innate ability to predict others' behaviour, which is essential for social interactions. Byrne and Whiten (1988) proposed the 'Machiavellian Intelligence' hypothesis, which argues that social primates are distinguished from other groups by their ability to develop and manipulate social knowledge about each other.
This social intelligence, it is argued, is deficient in autism. Because the cardinal features of autism are strikingly similar to negative features of schizophrenia, this has sparked efforts to explain schizophrenia through a ToM model.

Schizophrenics, when asked to describe pictures of people, describe their physical appearance but fail to make inferences about their mental state (Pilowsky and Basset, 1980. Allen, 1984.). This suggests that schizophrenics are unaware of others mental states which may lead to negative signs e.g. flattening of affect and social withdrawal. Frith further proposes that schizophrenics are unaware of their own beliefs and intentions. This may explaining the impoverishment of will that occurs.

Incoherence can also be explained in part by ToM. Rochester and Martin (1979) showed that one characteristic of incoherent speech is lack of referents. Pronouns are used without any reference to who or what the protagonist might refer to. Schizophrenics (and autistic children) make the assumption that if they know something then everybody must know it. As a result, an attempt to explain 'they', 'he' etc is rarely made.

Unlike autistic children who are born with ToM deficiencies, schizophrenics have had a period of normal ToM. The onset of illness may result in reduced ability to mentalise, but the desire to understand the actions and intentions of others remains. Attempts to mentalise may result in wrong inferences because the tools used to make accurate inferences are faulty. This may be the origin of positive symptoms.
Hallucinations may arise due to dissociation of awareness of own or others' intentions. A sense of derealisation is felt because one becomes unaware of their own intentions or goals. Delusions of control may follow, since one's actions cannot be related to any apparent intentions. Paranoia may be a consequence of the belief that others mean harm or are actively hiding their intentions from the schizophrenic.

Frith proposes a possible neurophysiological mechanism for such dissociation. Lesion studies suggest that representations of perception (in posterior structures e.g. primary sensory cortices) are stored separately from awareness of perceptions (frontal cortex). A neurophysiological insult may cause dissociation between the auditory perception of a voice and the conscious awareness of it. The subject hears the voice but is unable to consciously relate it to the immediate environment. This model is flawed because it depends on external triggers for hallucinations. Hallucinations are known to occur when the schizophrenic is in total isolation.

Brothers (1990, 1992) elaborates on the neurophysiological basis behind ToM. He argues strongly that there is a distributed brain system dedicated to social cognition. He identifies the principal components as the superior temporal sulcus (STS), the amygdala and the orbitofrontal cortex (OFC).

Single cell recordings from macaque monkeys have implicated STS in face processing functions, including identity perceptions. The amygdala is connected to a variety of cortical and subcortical structures. It is seen as an interface between information processing of the cortex and regulatory functions of the hypothalamus and the brainstem. The STS is connected to the amygdala, which in turn is richly connected to the OFC. Hence it may provide an interface between the functions of the two. Broks sites the case of a woman who showed impairments in identifying emotional facial expressions after partial bilateral amygdala surgery.

The OFC's function is thought to be regulation of social interactions. Brothers cites a case reported by others of a man who underwent surgery for resection of an orbitofrontal meningioma. After surgery, his intelligence remained normal. However he showed dramatic losses in the ability to judge social situations appropriately.

ToM theory applies well to Paranoid Type schizophrenia. Semantic memory and LI theories apply well to Disorganised Type schizophrenia. Can a unifying model emerge to explain both aberrant ToM and impaired semantic memory and LI? Other unanswered questions are the latency in onset of illness and the episodic nature of positive symptoms.

Information processing theory has yet to offer a unified explanation for the diverse nature of symptoms seen in schizophrenia. Yet schizophrenia research is distinguished by the bounty of possible models. Cerebral lateralisation, Right hemisphere dysfunction, working memory impairments, early or late neurodegeneration, neurodevelopment insult…The profusion of explanations reflects the sheer span of schizophrenia and schizotypal syndromes. The syndrome may be a cluster of disorders with dissimilar aetiology but similar symptomatology. Alternatively it may only be a matter of time and scientific sophistication before one root pathogenic factor is identified. Some persistent enigmas remain unexplained. What happens between the 15-20 years that leads to onset of illness? Why is onset even later in women? High-risk studies may provide insights. The onset of illness may be related to the end of callosal myelogenetic cycle that occurs approximately by the fifteenth year. Other changes occurring between the ages of 10 and 16 are a reduction of cerebral oxygen metabolism, a reduction in delta amplitude of EEG, a reduction in neuroanatomical plasticity and surge in adrenal function. These possible onset factors are the topic of much research interest and may yet provide a breakthrough.

Yet another first class essay. I particularly like the way you introduce each paragraph with a sentence that kind of acts like a flag, giving a quick view of what is to come.

I also like the way you bring your own examples to bear and draw your own conclusions in your own words. Just be careful with this, some examiners will be of a conservative bent.

Remember that:

1. The Latent Inhibition theory is supposed to be a low-level (perceptual) theory, but it is possible to argue that it is actually a high-level (cognitive) theory because of a) the fact that expectation can be argued to affect performance and b) because it appears to have something to do with the Central Coherence of deficits founding people with autism.

2. When describing theories and experiments in general, do try to include a sentence or two about what they actually did (it is a good general rule to always read the methods section when reading a paper - - I rarely read anything other than that actually).

3. When speaking about ToM, it is probably a good idea - although we didn't do it in the tutorial (my fault) - to briefly discuss whether it is useful to compare two such different disorders.

4. Here are some additional refs (from Boer, J.A. den, Westenberg, H.G.M. and Praag, H.M. van (Eds.) Advances in the neurobiology of schizophrenia. Chichester, Wiley, 1994):

Cognitive fragmentation
Perceptual dysfunction can be measured by cognitive tasks, but it is also among the most typical subjective experiences of the early stages of schizophrenia (Cutting and Dunne, 1989). There are many patients who describe a fragmentation of their sensory experiences:
There were a lot of meaningless details around me (...) I lacked the overall view. I saw fragments only (...) In fact, I am already wrong when I say that I saw everything, since these things presented themselves differently from otherwise. They were not included in a large context, but they were meaningless details (Matussek, 1952).

Sometimes the visual world quite literally 'decomposes'. Disintegration may also be apparent in a lack of perspective, or in a disconnection between external percepts and a corresponding inner feeling. Moving around changes the environment and makes it particularly difficult to integrate details into context:
Everything is in bits. You put the picture up bit by bit into your head. It's like a photograph that's torn in bits and put together again. You have to absorb it again. If you move it's frightening. The picture you had in your head is still there but it's broken up. If I move there's a new picture that I have to put together again (McGhie en Chapman, 1961).

No wonder patients tend to keep things frozen, to fixate their eyes, to stop moving. This might be one cause of catatonic symptoms:
I am in search of immobility (....) I aim at tranquillity and motionlessness. I also tend to stop life around me. That's why I like durable objects (...), things that stay forever, that never change (Minkowski, 1953).
Different kinds of sensory input are no longer interconnected. The fragmentation of experiences that are self-evident and automatic under normal conditions is a major cause of feelings of derealization:

My perception of the world seemed to sharpen the sense of the strangeness of things. In the silence and immensity, each object was cut off by a knife, detached in the emptiness, in the boundlessness, spaced of from other things (Sechehaie, 1950).

The experience of time, although not among the conventional symptoms of schizophrenia, is often affected by the integrative deficit as well. Time is a subjective experience of continuity. During a psychosis it may disintegrate into a 'slide presentation':

Words like 'before' and 'after' no longer had any meaning. There was only 'now' (White, 1979).
Hemsley (1987) has pointed out that schizophrenic patients seem to suffer from a weakening of the influence of stored memories of regularities of previous input on current perception. The result is that they make less use of redundancy. Because of their failure to effectively use spatial and temporal regularities in perceptual input, there is a breakdown of 'Gestalt' perception. Hemsley's 'memories of regularities', of course, are cognitive schemata: knowledge representations which normally lend structure to perceptual experience.

Perceptual organization
Perceptual input is not automatically organized according to expectancies and rules (schemata) based on earlier experiences. Interestingly, this formulation allows for tasks where schizophrenic patients perform better than control groups. Such findings may provide powerful arguments against the generalized deficit hypothesis of schizophrenic cognitive dysfunction. One would expect superior performance in situations where reliance on rules and schemata is inappropriate to the task requirements.

There are several examples of tasks that fulfil these requirements. Carter et al. (1993) found that, although schizophrenic patients were generally slower than control subjects in the Stroop colour-word interference task, they actually showed a greater facilitation for congruent words (e.g. naming the colour of the word 'red' written in red). They concluded that (undifferentiated) schizophrenic patients profited from the processing of irrelevant information.

Brennan en Hemsley (1984) examined the phenomenon of illusory correlation. This is the experience of correlation between events which in reality are uncorrelated. They presented a series of word-pairs, some of which had strong associative connections. Normal subjects overestimated the co-occurrence of word-pairs with strong associative connections, whereas (non-paranoid) schizophrenic patients performed more accurately. They were less influenced by expectations based on their knowledge of associations between words.

Schwartz Place en Gilmore (1980) described a visual task on which schizophrenic patients were superior to control subjects. Their results were replicated by Wells and Leventhal (1984). This task consisted of a display containing a number of lines. Subjects had to report this number. Complexity was varied: all lines of one orientation, horizontal and vertical lines in mixed order, or both line orientations grouped together. The performance of controls deteriorated as the organization of the arrays became more complex. They see wholes instead of details. Schizophrenic patients were not affected by the level of perceptual organization and performed significantly better in the most complex condition. This suggests that they do not profit from Gestalt grouping principles, which made their performance on this task superior to that of normals.

These studies support the view that (part of) the schizophrenic cognitive dysfunction is a failure to integrate details in a Gestalt fashion. When the automatic top-down processes of perceptual integration are inefficient, images and messages have to be constructed via bottom-up processes (John and Hemsley, 1992). This makes these patients generally slower in cognitive tasks, but it affects their social competence as well and it seems likely that it is a major cause of schizophrenic social dysfunction (Morison, Bellack and Mueser, 1988). Corrigan and coworkers (Corrigan and Green, 1993; Corrigan, Wallace and Green, 1992) have argued that deficits in social schema processing and a relative insensitivity to abstract cues may explain the limited repertoire of social skills of many schizophrenic patients.

Face recognition may serve as an example. It is dependent on the integration of many details. Schizophrenic patients have difficulty detecting or recognizing emotions displayed in photographs (Feinberg et al., 1986). Frith et al. (1983) confirmed that in sorting human faces, schizophrenic patients did not use Gestalt principles in the way control subjects did. In everyday life, recognition of affective features is even more difficult, since it requires very rapid scanning of facial expressions often lasting just tenths of a second:

But I perceived a statue, a figure of ice which smiled at me. And this smile, showing her white teeth, frightened me. For I saw the individual features of her face, separated from each other: the teeth, then the nose, then the cheeks, then one eye and the other. Perhaps it was this independence of each part that inspired such fear and prevented my recognizing her even though I knew who she was (Sechehaie, 1950).
Other cognitive impairments, like memory and learning problems, contribute to the difficulties patients experience in the transactions of everyday life. Goldberg et al. (1993), for instance, found significant associations between measures of memory and global levels of social and vocational functioning. The perceptual organization problems represent a more general dysfunction affecting input as well as output mechanisms. At the motor side similar dysfunctions can be identified. Two notable examples are speech disorders and eye movement dysfunction. Both refer to a failure to use contextual information and to take advantage of redundancy (Manschreck, 1986). McGrath (1991) has described how speech disorders relate to a planning failure. Harrow, Lanin-Kettering and Miller (1989) have examined how impaired perspective (the ability to recognize in a global fashion which behaviours are appropriate for a given communicative context) plays a role in thought disorder.

Hommer et al. (1991) demonstrated the contextual deficit in eye movements. They examined the performance on a task that required the use of an internal representation to control saccadic eye movements. Without visual feedback, patients with first rank symptoms were also less likely to make error corrections on a motor task (Malenka et al., 1986; Frith and Done, 1989). Frith and Done qualified this dysfunction as a disorder in the central monitoring of action, because the ability to make such corrections depends on an internal representation of the response that has just been made.

Stored information
Cognitive failures are discernible in various memory functions (Gold et al., 1992). Schizophrenic patients' poor performance on tests of verbal memory has been demonstrated repeatedly, and similar deficits have been found in visual memory. Recall performance improves with structured tasks (Calev, 1984b). A differential deficit on recall as compared with recognition tasks suggests that the verbal memory deficits are related to poor organization in the encoding phase, like impaired chunking (Calev, 1984a). Speed et al. (1991) demonstrated a failure to use inherent structure of verbal material to improve recall. Manschreck et al. (1991) reported reduced primacy performance (recall of the initial items of a sequentially presented set) in association with this decreased context associated gain in recall. These deficits are ascribed to poor organizational skills. Memory items appear poorly organized into larger cognitive structures that normally facilitate their retrieval.

This may not be the sole explanation and other factors probably contribute to the memory deficits, e.g. deficits of mental effort and motivation. This is accordance with the findings of Tamlyn et al. (1992), who concluded that memory impairment was specifically associated with two different classes of symptoms: formal thought disorder and negative symptoms. The organizational deficit seems the primary cause and other factors, related to effort and motivation (and negative symptoms), may be secondary.
Memory is a heterogeneous concept. The traditional memory functions discussed are called associative, because information is acquired by the repeated contiguity between stimuli and responses or consequences. The term working memory denotes a type of memory that is active for only short periods (often a few seconds). Working memory is the process for the proper use of acquired knowledge: the regulation of behaviour by representational knowledge, that is, by cognitive schemata.

Cognitive tasks like the CPT show that schizophrenic patients have specific problems with the efficacy of their working memory, but there are also specific working-memory tasks (Goldman-Rakic, 1991; Park and Holzman, 1992). The best-known of these tasks is the Wisconsin Card Sorting Test (WCST; Van der Does and Van den Bosch, 1992). In this task, the subject has to sort a deck of cards. Each card has to be matched to a set of reference cards on the basis of a criterion selected by the experimenter. The experimenter only says 'right' or 'wrong' and the subject tries to get as many correct matches as possible. After a number of correct matches, the sorting principle shifts without warning.

The WCST requires that the subject uses representational memory, since the cards contain no information about the correct response. Schizophrenic patients often perseverate on incorrect responses. This does not seem to result from a lack of knowledge, but from a failure to use this knowledge to change and direct action. This deficit is not found in all cases, but when present it is a pervasive deficit (Goldberg et al., 1987). Nonetheless, with appropriate cognitive instruction performance is remediable to some extent, at least in some patients (Bellack et al., 1990).

Conventional memory tasks are effortful: they address explicit memory. Subjects have to recall or recognize intentionally. Implicit memory refers to priming tasks, in which remembering is automatic and the subject is unaware that his memory is being tested. Priming refers to an improved facility for detecting or identifying perceptual stimuli based on recent experience with them. For instance, a primary exposure enhances completion of incomplete words or recognition of degraded pictures. Although some deficits have been observed, lexical and semantic priming effects are not disturbed to a significant degree in schizophrenic patients (Gold et al., 1992; Schmand et al., 1992). Most kinds of automatic learning seem to function within normal limits.

However, there are strong indications that the influence of context is insufficient. The result is that implicit knowledge is not incorporated into larger cognitive structures. This corresponds with the genesis of explicit memory deficits. In this respect, two behavioural phenomena have attracted the interest of schizophrenia researchers: latent inhibition and the Kamin blocking effect. Both are instances of the influence of implicit memory traces on current perception.

The phenomenon of latent inhibition (LI) refers to a decrement in associability of a stimulus previously pre-exposed without being followed by an event of consequence. In the first stage of the experiment a stimulus is repeatedly presented. In the second stage this pre-exposed stimulus is paired with reinforcement in a standard learning procedure. Subjects who are stimulus pre-exposed learn a new association more slowly than subjects who did not receive the pre-exposure. In other words, LI reflects a process of learning not to attend to, or to ignore, irrelevant stimuli. The pre-exposure stimulus becomes predictable and redundant, because initially it has no consequence.

LI has been found disrupted in acute schizophrenic patients, but not in chronic patients (Baruch, Hemsley and Gray, 1988; Lubow et al., 1987). Although this phenomenon has a superficial resemblance to habituation, there is a significant difference: LI is context-dependent, whereas habituation is not. Schizophrenic patients show a failure to relate specific associations to the context in which they occur (Hemsley, 1992).

The Kamin blocking effect involves a pre-exposure phase when the subject learns an association between two stimuli. In the second phase, he is presented with a pairing between a compound stimulus (encompassing one of the previous stimuli and a new one) and the second stimulus from the first phase. When, in the third phase, the relationship between the added stimulus-component and the second stimulus is tested, pre-exposed subjects show less learning than controls who received a neutral first phase condition. This blocking effect is supposed to be the result of not paying attention to the new stimulus in the second phase, because it seems redundant. It adds no new information to the association between the first two stimuli. Like LI, the blocking effect was also reduced in acute, but not in chronic, schizophrenic patients (Jones, Gray and Hemsley, 1992).

The results obtained with LI and blocking tasks are interesting for the same reason why those obtained with the perceptual organization test are interesting: the possibility that the findings are a nonspecific consequence of generally poor cognitive performance is eliminated.

It seems likely that the differential response in chronic schizophrenic patients is determined by the phase of their illness. Future studies should aim at detecting the influence of medication and of symptomatological correlates, e.g. positive versus negative symptoms. Animal studies also suggest this direction. They point to the role of dopaminergic neurotransmission, since the administration of amphetamine disrupts both LI and the blocking effect. It is also of interest that LI and blocking are both abolished by damage to the hippocampus, since hippocampal pathology is considered of etiological significance in schizophrenia (Feldon and Weiner, 1992).