SEARCH

SEARCH BY CITATION

Keywords:

  • cognition;
  • perception;
  • psychosis;
  • risk;
  • schizophrenia

Abstract

  1. Top of page
  2. Abstract
  3. MODEL OF CORTICAL HIERARCHY
  4. SYMPTOMS
  5. CONCLUSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES

Conventional wisdom has not laid out a clear and uniform profile of schizophrenia as a unitary entity. One of the key first steps in elucidating the neurobiology of this entity would be to characterize the essential and common elements in the group of entities called schizophrenia. Kraepelin in his introduction notes ‘the conviction seems to be more and more gaining ground that dementia praecox on the whole represents, a well characterized form of disease, and that we are justified in regarding the majority of the clinical pictures which are brought together here as the expression of a single morbid process, though outwardly they often diverge very far from one another’. But what is that single morbid process? We suggest that just as the uniform defect in all types of cancer is impaired regulation of cell proliferation, the primary defect in the group of entities called schizophrenia is persistent defective hierarchical temporal processing. This manifests in the form of chronic memory-prediction errors or deficits in learning-dependent predictive perception. These deficits account for the symptoms that present as reality distortion (delusions, thought disorder and hallucinations). This constellation of symptoms corresponds with the profile of most patients currently diagnosed as suffering from schizophrenia. In this paper we describe how these deficits can lead to the various symptoms of schizophrenia.

ALTHOUGH THE DIAGNOSIS of schizophrenia has become more reliable with the establishment of observable diagnostic criteria, the conceptualization of the disorder has not varied considerably since the term was initially proposed by Bleuler and discussed by Kraepelin as dementia praecox. Like these original descriptions, the current concept of schizophrenia is heterogeneous. Endicott1 compared the frequencies and reliabilities of the following criteria sets for the diagnosis of schizophrenia: the New Haven Schizophrenia Index: the Carpenter, Strauss, Bartko (4-, 5-, 6-, and 7-item) system; DSM-III; Research Diagnostic Criteria (RDC; full and chronic); the Feighner system; and the 1975 criteria of Taylor and Abrams. The systems, equally reliable, varied sevenfold in their rates of diagnosing schizophrenia. Oulis et al.2 investigated the lifetime fulfillment of five sub-criteria of the primary diagnostic criterion in inpatients with a definite diagnosis of DSM-IV schizophrenic disorder. Among the diagnostic features, only delusions were almost universal, and only hallucinations and negative symptoms occurred in more than half of all the cases. In addition, 60% of the cases were accounted for by only four of the patterns. The substantial clinical heterogeneity of the DSM-IV category of schizophrenic disorders with respect to their diagnostically characteristic features is traced to the polythetic character of its five sub-criteria.

At this point in time, conventional wisdom has not laid out a clear and uniform profile of schizophrenia as a unitary entity. One of the key first steps in elucidating the neurobiology of this entity would be to characterize the essential and common elements in the group of entities called schizophrenia. Kraepelin, in his introduction, notes ‘the conviction seems to be more and more gaining ground that dementia praecox on the whole represents, a well characterized form of disease, and that we are justified in regarding the majority of the clinical pictures which are brought together here as the expression of a single morbid process, though outwardly they often diverge very far from one another’.3 But what is that single morbid process?

We have suggested that just as the uniform defect in all types of cancer is impaired regulation of cell proliferation, the primary defect in a group of entities called schizophrenia is persistent defective hierarchical temporal processing seen as chronic memory-prediction errors or deficits in learning-dependent predictive perception that present as reality distortion (delusions, thought disorder and hallucinations).4 This constellation of symptoms corresponds with the profile of most patients currently diagnosed as suffering from schizophrenia. In this paper we extend our concept to describe how these deficits can lead to the various symptoms of schizophrenia.

As Kraepelin states, ‘patients often perceive much better than one would expect from their behavior.’ But ‘the trustworthiness of perception is decidedly decreased.’3 While patients with schizophrenia do not have overt perceptual deficits that render them unable to see, hear or touch in a way that is identifiable to a clinician, we propose that a chronic impairment in learning-dependent predictive perception, leads to the cluster of symptoms that Kraepelin identified as dementia praecox and Bleuler referred to as the schizophrenias.

MODEL OF CORTICAL HIERARCHY

  1. Top of page
  2. Abstract
  3. MODEL OF CORTICAL HIERARCHY
  4. SYMPTOMS
  5. CONCLUSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES

Our model,4–6 described previously, is a variation of the Hierarchical Temporal Memory model of the neocortex described by Hawkins and Beardslee.7 In this model, the operation of the neocortex is portrayed as a basic unit – the cortical micro column – that is organized in a structured hierarchy with each unit using the same learning and inference algorithm. Our model postulates that the neocortex, along with the hippocampus and thalamus, builds a model of the world using a spatial-temporal hierarchy that is constrained by the evolutionary structure of the organism (see Fig. 1). The constraints are manifold and include the sensory organ structure and the environment. The model build consists of the sum of the pre-existent evolutionary configuration and a person's invariant constructs (i.e. memories that capture the essence of an experience while allowing for some deviation in the particulars). This model is then updated to accommodate new experiences within the framework of its pre-existing structure. As we interpret the world, higher cortical areas constantly compare current circumstances to invariant memory stores to form prefigurations about the next moment of experience (memory prediction) or as posited by Purves8 about the likelihood of the successful behavior or response. These prefigurations are organized hierarchically.

image

Figure 1. Columnar circuitry underlying memory-prediction processes. Columns in this hypothetical section of auditory cortex are tuned to specific musical intervals. The circuitry in this area of cortex allows for prediction of the specific upcoming note of a familiar song, even when this song is played in a different key than previously experienced. (i) Layer (L) 6 cells in a higher cortical region project an invariant ‘hypothesis’ of the identity of the song (or musical phrase) broadly throughout L1 of this area of cortex. (ii) Information regarding the last musical interval experienced is communicated broadly throughout L1 via thalamocortical loop input. Cells in L2 and L3 have extensive dendritic arbors in L1 and learn to anticipate the next musical interval based on the invariant hypothesis and last interval experienced. This initial prediction is invariant to the specific notes involved. Thus, in this example, all columns representing the interval of a fourth are primed (indicated by 45-degree hatch). (iii) Bottom-up input to L4 signals the specific current note, activating all columns with ‘C’ as the first tone in the interval (indicated by 315-degree hatch). Processing of this information is facilitated by accurate prediction that primed this column for bottom-up stimulation. (iv) The convergence of the invariant prediction of a 4th and bottom-up signal representing the note C create a specific prediction that the next note will be an F (indicated by cross-hatch). This prediction is communicated broadly to L1 of the lower cortical area by the diffuse projection from the output cell in L6. (Reprinted with permission from Kraus et al. 20095).

Download figure to PowerPoint

From a top-down orientation, they are organized by a series of priming levels to be activated. From the bottom-up orientation, they are organized by an upward cascade of sensory signals that set the stage for perception. The overall model therefore posits that the nature of the output from a given area of cortex depends on coincidence of top-down signals with the patterns of bottom-up input it is receiving at any given time. When a person is in a new situation experiencing stimuli that do not clearly fit any top-down hypotheses derived from previous experience, a given area of cortex relays the details of the patterns it receives to higher cortical areas; that is, the signals are passed on to the next highest layer and this pattern extends till a match is achieved. However, when the bottom-up signals are successfully predicted, the cortical area no longer transmits the details of the bottom-up signals it is receiving, but instead a condensed signal, which serves as a ‘label’ for the pattern of activity, is transmitted to the next higher cortical area in the hierarchy. The next cortical area is then able to begin detecting higher-level patterns in the sequence of ‘labels’ received from below. In this way, as a situation becomes more familiar, the representations of a given level of analysis are shifted to lower cortical areas, freeing higher areas for detection of higher level patterns.

Perception, far from being an objective translation of reality, is shaped both by a perceiver's knowledge and by his or her past experience with particular stimuli. Bottom-up sensory signals are constantly being interpreted in light of top-down predictions, which serve to speed processing and bias perception towards likely percepts. For instance, perception of ambiguous phonemes is significantly influenced by the lexical context in which they are experienced.9 This biasing of sensation by prediction has also been observed as reduced variance in functional imaging signals. When dichoptic-color masking is employed to manipulate the conscious awareness of images of houses and faces, images within subjects' conscious awareness produced temporal lobe activation patterns that were more stable across presentations than activation produced by images that did not reach consciousness.10 The authors hypothesize that conscious perception may be represented in a more discrete form than unconscious sensations and that recurrent networks activated by conscious perception may stabilize activation patterns. Goldstone11 defined perceptual learning as ‘relatively long-lasting changes to an organism's perceptual system that improve its ability to respond to its environment and are caused by this environment’. The deficit that arises when this processing is damaged is a deficit in the ability to predict perception on the basis of what has been learned about the perceptual world up to that moment in time.

What happens in schizophrenia?

We have proposed that schizophrenia, just like cancer, is a variety of entities sharing a common underlying defect. Just as impaired regulation of cell proliferation underlies all types of cancer, schizophrenia is a group of entities characterized by chronic reality distortion secondary to disruption of the circuitry underlying hierarchical temporal memory prediction (or learning-dependent predictive perception). In a previous paper we described how the cellular components of the cortical micro column are linked and how dysfunctions arise in psychosis.4 We suggested that deficits in structure and function, particularly in higher-order hierarchical units (including the prefrontal cortex hippocampus and thalamus), are the key pathophysiological cause that leads to the illness. We have previously suggested that this deficit can arise from a variety of causes: environmental, genetic and social.4 In this paper we extend our concept to discuss how this deficit leads to the key symptom features seen in schizophrenia.

Evidence for deficits of learning-dependent predictive perception in schizophrenia

Part of the test of the hypothesis of an impairment in learning-dependent predictive perception in schizophrenia would be a manifestation of difficulties in perception that depend upon higher-order processing. In a classic example, when fragmented images of common objects are presented from the most degraded to the most complete, schizophrenia patients require more complete stimuli than healthy controls before they are able to identify the images.12,13 Patients with schizophrenia perform better than control subjects in a task in which subjects are shown degraded stimuli of known objects and later asked to draw the stimuli from memory as accurately as possible.14 Healthy subjects tended to fill in the ‘missing’ portions of the figures, but patients more accurately reproduced the figure in their drawings. These results are consistent with a decreased influence of invariant memories of the familiar objects and a relative weighting of low-level processing in perception and memory in schizophrenia.

Binocular depth inversion occurs when viewing ‘hollow’ versions of common objects created by switching the images (particularly faces) typically viewed by each eye.15,16 This illusion reflects higher-order invariant memories overriding bottom-up sensory information of unlikely stimuli.17 Patients with schizophrenia more frequently report perceiving the correct stimuli than do control subjects. Patients meeting criteria for an initial prodromal state for schizophrenia obtained scores higher than healthy controls, but lower than schizophrenia patients.18

Mismatch negativity provides another example of learning-dependent predictive perception deficits in schizophrenia. Schizophrenic patients have decreased mismatch negativity (MMN) to auditory oddball stimuli.19 MMN is elicited when a regular series of tones is interrupted by a tone that deviates. The elicited MMN is largely unaffected by attention20 and the amplitude of the MMN is inversely related to the deviant probability.21 MMN appears to reflect a pre-attentive process of predicting upcoming sensory input based on regularities extracted from previous experience22 and the decreased MMN seen in individuals with schizophrenia suggests a deficit in these processes.

A common deficit in schizophrenia patients and their first-degree relatives is a decrease in smooth pursuit gain and subsequent catch-up saccades with eye tracking.23 In contrast, when a target changes directions unpredictably, patients with schizophrenia track better than controls during the brief period around the change. This result suggests that tracking in schizophrenia patients relies more heavily on lower-order retinal error signals, while controls rely more heavily on higher-order predictive mechanisms to set pursuit gain.24 Schizophrenia patients also exhibit decreased predictive pursuit relative to healthy controls when retinal error signals are eliminated by temporarily masking the pursuit target25,26 or by stabilizing the target on the fovea.27 Very recent work on the process of learning to perceive motion based upon direction and speed of visual stimuli suggest that healthy controls generalize features of movement from direction to speed and vice-versa, while patients with schizophrenia do not (Yue Chen et al., unpublished data). These results again are consistent with a learning-dependent predictive perception deficit in schizophrenia in that learning the movement of visual stimuli does not result in a generalized predictive perceptual ability.

Learning-dependent predictive perception deficits can lead to distortion of reality in two key ways: not identifying incoherencies in the environment or misinterpreting ambiguous perceptions leading to internally biased and inaccurate representations.

Reality distortion

Recently, Sorkin et al. evaluated the ability of patients with schizophrenia to assess incoherence in the external world via a head-mounted display delivered in a virtual reality environment.28 Apart from the deliberately planted incoherencies, the virtual environment resembled the real world. In this task, the subject is instructed that whenever the path of the subject traverses an incoherent event, the subject should detect and report the incoherency. Schizophrenia patients performed very poorly on this task and could be powerfully discriminated from healthy controls. In addition, poorer performance as measured by the audio-visual incoherency detection was associated with hallucination severity, which is rare in studies of the relation between cognition and clinical symptoms.29 This distortion is consistent with our model (Fig. 2). The higher-order hierarchical temporal memory deficit increases the difficulty of the detection of incoherencies.

image

Figure 2. Failure of incoherency identification. Individuals with schizophrenia are impaired in their ability to identify incoherencies in a virtual environment.28 We propose that identification of incoherencies is aided by automatic allocation of attentional resources to stimuli that violate prediction. In this example, the subject sees the image of a man playing a guitar on a doorstep, which (i) activates invariant memories related to guitars. (ii) Activation of memory stores trigger predictions that the sound of guitar music will be heard. (iii) When instead the sound of a trumpet is processed by the auditory system, and (iv) later perceived, (v) the mismatch between predicted and actual cortical activation attracts attentional resources, aiding causal learning. Because these processes are weakened in schizophrenia, the recruitment of attentional resources is haphazard and incoherencies frequently go undetected. (inline image) Learning-dependent predictive perceptual process in healthy individuals (inline image) Aberrant processes in schizophrenia (inline image) Weakened processes in schizophrenia.

Download figure to PowerPoint

Frequently, stimuli in the real world arrive to our sensory systems in degraded and ambiguous states. It has been proposed that a major function of the cortex is to use current contextual clues and invariant memories to guide perception towards the most likely interpretation of these suboptimal signals.7 We propose that deficiencies in the processes of learning-dependent predictive perception in schizophrenia result in perceptions that are less contextually guided and influenced more by internal biases (Fig. 3). When sensory signals are degraded to such an extent as to become meaningless (uninterpretable), patients with psychosis nonetheless tend to hear spurious messages from the noise. This deficit in predictive perception may be the genesis of auditory hallucinations and may play a role in development of delusions. In our model, due to learning and perceptual deficits, patients with schizophrenia have not established a comprehensive library of accurate percepts and expectations. As a result, they fill in the gaps of missing information with internally based, idiosyncratic guesses that are often inaccurate. A key point in the model is that the ongoing march of time is relentless; if an individual misses important information at one moment, that moment is forever unperceived and that information is never learned. Thus the individual can never recover; the previous moment's failure leads to an inability to process information in the next moment, on and on throughout life.

image

Figure 3. Failure of invariant memories to influence interpretation of ambiguous stimuli. When subjects are asked to identify a word made ambiguous by deletion of a single letter, their answers are biased by presentation of a preceding auditory passage. However, this effect is diminished in individuals with schizophrenia.30 In this example, we propose that (i) the immediate past perception regarding a mine tunnel activates invariant memories related to mines. (ii) Activation of memory stores trigger predictions that the next moment of experience will be related to mines. (iii) When ambiguous stimuli are presented, (iv) perception is biased toward the predictions. This phenomenon is muted in subjects with schizophrenia due to a decreased influence of invariant memories on perceptual processes. (inline image) Learning-dependent predictive perceptual process in healthy individuals (inline image) Aberrant processes in schizophrenia (inline image) Weakened processes in schizophrenia.

Download figure to PowerPoint

This deficit is particularly devastating in the early phases of life when healthy children are storing huge libraries of implicit information about the nature of reality. If those memory stores are insufficient and filled with misattributes and misperceptions, then the ability to understand and experience the complex nature of reality in adult life is challenged. Recent work from the Dunedin Longitudinal Study suggests that at a very young age, those individuals who will eventually develop schizophrenia have a verbal deficit; as they develop, a visual/processing speed deficit emerges as well, and this processing deficit is related to the presence of the earlier verbal deficit.31 Thus, as healthy children grow and scaffold their previously developed abilities into more complex skills, the presence of one deficit early on makes the child who will develop schizophrenia vulnerable to other deficits, and may be a key factor in the eventual onset of psychosis.

Structural evidence for a learning-dependent prediction deficit in schizophrenia

A number of structural deficits in the multimodal association areas, hippocampus and thalamus are seen in patients with schizophrenia. An extensive literature32–39 documents consistent morphometric changes in schizophrenia. Besides whole-brain volume reduction in patients of about 3% in gray matter,35,36 volume reductions are particularly large in frontotemporal regions. The prefrontal cortex, superior temporal gyrus, hippocampus, thalamus and amygdala are reduced by a greater amount than the whole brain.32–38 In addition there are consistent cytoarchitectonic data. Pyramidal neurons in PFC deep layer 3 play a central role in both thalamocortical and corticocortical circuitry. The somal volume estimates in subjects with schizophrenia are significantly decreased.40 This decrease is associated with a shift in the distribution of somal volumes toward smaller sizes. Pyramidal cell somal volume is reduced in layer 3 of auditory association cortex, but not in layer 5, in subjects with schizophrenia.41,42 Decrease in dendritic spine density on dorsolateral prefrontal cortex layer 3 pyramidal cells in schizophrenia is consistent with the hypothesis that the number of cortical and/or thalamic excitatory inputs to these neurons is altered in schizophrenia. A significant decrease in the number of both primary (29%) and secondary (46%) basilar dendrites on pyramidal neurons in layer 5 have also been demonstrated.43 Similarly, in layer 3 there is also a decrease in both primary (17%) and secondary (15%) basilar dendrites in patients with schizophrenia.43 These changes are in accordance with our hypothesis that higher hierarchical levels are impaired. Additional data also show evidence for our hypothesis. Top-down facilitation is triggered by magnocellular information projected early and rapidly to the orbitofrontal cortex. Neuroimaging data show that stimuli designed to bias processing toward the magnocellular pathway differentially activate the orbitofrontal cortex.44 In addition, patients with schizophrenia demonstrate the expected impairment in functioning of the magnocellular visual pathway.45

SYMPTOMS

  1. Top of page
  2. Abstract
  3. MODEL OF CORTICAL HIERARCHY
  4. SYMPTOMS
  5. CONCLUSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES

A list of the major symptoms in schizophrenia and the mechanisms by which failures in learning-dependent predictive perception may lead to them, can be found in Table 1.

Table 1.  Impairments in learning-dependent predictive perception underlying symptoms of schizophrenia
Nature of predictionDeficit in schizophreniaPossible symptoms
Predictions based on invariant memories lead to accurate, timely perceptionDue to unreliable predictions, sensory analysis is divorced from context and relies more heavily on laborious bottom-up processes, or perception may conform to internally based, inaccurate predictionsHallucinations and delusions
Patterns of sensory stimuli that have been consistently maintained are likely to continueA decreased contextual window, exhibited by impairments in mismatch negativity and smooth pursuit eye movements, and a tendency to jump to conclusionsDelusions
Expected effects of one's own thoughts and actions are incorporated into prediction of upcoming sensationEffects of one's own thoughts and actions are not accounted for in prediction of upcoming sensationDepersonalization symptoms, Schneiderian delusions
Semantic associations gleaned from past experience guide perception and thoughtSemantic associations are weakened, leading to impaired language production and comprehensionFormal thought disorder
Extensive past experience with facial and vocal analysis directs attention to features likely to indicate emotional stateBreakdown in learning-dependent predictive perception leads individuals with schizophrenia to be unable to attend to effective indicators of emotional stateSocial impairment, negative symptoms

Delusions and thought disorder

The key feature in almost all patients with schizophrenia is delusion. We suggest that an impairment in learning-dependent predictive perception is the morbid change that leads to conclusion-jumping and then becomes the basis for delusions and hallucinations.

Failures in learning-dependent predictive perception beget conclusion-jumping

Patients with schizophrenia demonstrate an increased tendency to jump to conclusions. This tendency is demonstrated in a task in which subjects must determine whether marbles are being drawn from one of two jars: one containing 85% black, 15% white, and the other vice-versa.46 The progression of previously drawn marbles is visible. Patients generally reach conclusions with fewer marble draws and express more certainty than normal controls. These characteristics tend to be more pronounced in delusional than non-delusional schizophrenic patients.46 Interestingly, even when disconfirmatory evidence is presented, delusional patients over-adjust their assessment of which jar is being drawn from compared to normal controls.46,47 These findings suggest that delusional schizophrenic patients exhibit a bias towards recent events in their decision-making process, consistent with a failure of past events to guide expectation. The shrunken contextual window and tendency to jump to conclusions seen in schizophrenia may contribute to the robustness of delusions in the face of subsequent evidence and reasoned arguments.48,49

Based on these and other similar studies we can hypothesize that in the early stages of the disease, the formation and storage of invariant representations at higher hierarchical levels is insufficient. The higher levels do not provide enough input to lower levels for solving the nature of stimuli, and the lower levels do not provide sufficient perceptual details to enable a sufficient establishment of perceptual context. Thus, simple, concrete information needs to be sent repeatedly to higher hierarchical levels for more effortful interpretation. This reduction in the correct identification of percepts, combined with real-world information-processing demands, affords the opportunity for arbitrary internally generated interpretations of reality to intrude upon perception and thought. A repeated inability to perceive correctly may lead to an accumulation of inaccurate but internally meaningful perceptions that could then build upon one another into incorrect beliefs. This failed process may be at the core of the development of a tendency for hallucinations and delusions. Context-based perceptions of real objects and real events are reduced in favor of an interpretation of reality that is individually determined and disconnected from the experiences and beliefs shared by others. This we hypothesize is the mechanism behind the development of delusions and hallucinations.

Perception

Identification of letters is typically a nearly effortless task for healthy individuals and schizophrenia patients' tendency to report internally generated sequences suggests that the normal automaticity in this task is severely compromised. Such a decrease in the automaticity of perception appears to be a fundamental aspect of the perceptual profile of schizophrenia. Although schizophrenia patients are able to use semantic meaning to improve their performance on a visual recognition task, this improvement is less robust than in healthy controls and comes at the expense of increased reaction time (while healthy controls exhibit an improvement in reaction time concurrent with improved accuracy to semantically meaningful stimuli).50 These findings suggest that the integration of top-down predictions with bottom-up sensory signals is deficient in individuals with schizophrenia, forcing them to derive semantic meaning via more effortful construction of bottom-up information. In everyday life, sensory signals are often noisy, thus the ability to integrate top-down prediction with multiple bottom-up sensory streams increases the likelihood of accurate perception. Individuals with schizophrenia exhibit a decreased ability to use visual speech cues to assist in the interpretation of noisy speech even though they performed similarly to healthy controls when lip-reading and audio perception were measured separately.51 Of course, Gestalt psychologists focused for decades on the inability of patients with schizophrenia to integrate perceptions sufficiently to create a completely accurate percept. Conrad52 and Matussek53 theorized that these perceptual abnormalities led to the florid symptoms of schizophrenia. Although we focus on and discuss higher-order processing deficits in schizophrenia, recent evidence points to multiple hierarchical deficits in perception of vision and sound. For example the work of Javitt et al.45 clearly shows that lower-order deficits also occur in schizophrenia. These deficits place greater demands on higher-order processing and tend to worsen and enhance the problems with higher-order memory deficits.

Thought disorder

Besides delusional content, the difficulty with higher-order processing invariably expresses itself with organization of thought and speech. This deficit has been demonstrated experimentally in the Cloze paradigm, in which subjects are asked to discuss a given topic and then their transcribed speech is subjected to deletion of every fifth word. Healthy raters were less accurate in filling in the missing words from passages generated by patients with schizophrenia than those generated by healthy controls, suggesting that typical associations are reduced in individuals with schizophrenia.54

Speech and language

Although we have focused on internal models for perception we posit that there exist similar internal repertoires for motor behavior based on the same principles. Internal repertoires in the brain enable smooth sequential motor behaviors, from visuospatial movements to articulation of speech. At the same time, these motor behaviors also support aspects of perception, such as stabilization of the retinal image and disambiguation of phonological information, thus switching between forward and inverse modes. The goal of these models is to minimize the resulting error signal through adaptive mechanisms. As Indefrey and Levelt55 point out, spoken language ‘constantly operates a dual system, perceiving and producing utterances. These systems not only alternate, but in many cases they partially or wholly operate in concert.’ Further, spatial processing and real-time speech processing make use of the same internal model structures. Thus, the result of learning-dependent predictive perception may be the poor train of thought, neologisms, perseveration and other language disturbances that are seen in schizophrenia. These speech patterns appear to reflect the failure of invariant predictions to guide thought in semantically and syntactically meaningful directions. Without higher-level predictions guiding the train of thought from one moment to the next, the schizophrenic patient often fills speech with lower-level associations, such as clanging or echolalia.

Hallucinations

Auditory hallucinations are one of the most frequent symptoms of early psychosis. Often the voices are attributed to the external world. Hallucinations occur in other sensory modalities as well. Kapur56 notes that this process recedes early in the course of antipsychotic treatment response as internally generated stimuli become less salient.

An impairment in learning-dependent predictive perception may be the genesis of auditory hallucinations as evidenced by the tendency of patients with psychosis to hear spurious messages amongst auditory noise. This impairment reduces the automaticity with which the world is perceived and understood, leading the individual to develop and then heavily weight internally generated interpretations, which may significantly color perception. That is, they try to increase the salience of auditory noise. This phenomenon is seen in a very simple task. When presented with multi-speaker babble consisting of 12 independent streams of speech and given the task of repeating any words or phrases that they perceive, patients with early phase psychosis report longer word strings than healthy controls.57 As learning-dependent predictive perception is hypothesized to be impaired prior to the onset of frank psychosis, the ‘babble’ task may be sensitive to changes in which meaning is assigned to ambiguous stimuli. In fact, this task has recently provided pilot data suggesting that at-risk subjects who later convert to schizophrenia spectrum disorders report longer word strings in contrast to subjects who fail to convert, suggesting a greater propensity to increase the salience of auditory information perceived from background noise.58 Multiple measures of learning-dependent predictive perception have been found to be correlated with auditory hallucinations, including poor temporal context discrimination (remembering to which of two lists a word belonged), an increased tendency to falsely recognize words that were not present in the lists, and misattributing the items to another source.59–61 Similarly, bipolar and schizophrenic subjects who had experienced auditory hallucinations did not exhibit a decrease in ticklishness to self-produced tactile stimulation compared to externally produced stimulation, indicating that perceptual prediction based on motor activity was impaired in these individuals.62

Hallucinations are also linked to response bias (reflecting the tendency to make false detections) in signal detection paradigms. In a task in which participants with schizophrenia were required to detect an acoustic signal randomly presented against a noise background, those with auditory hallucinations were similar in their perceptual sensitivity to those without hallucinations, but differed in their response bias in that they were more willing to believe that the signal was present.63

Invariant concept of a face is weakened

The analysis of faces is an essential function for human beings and is a key component for building relationships. Humans perceive faces even from suboptimal stimuli. This tendency is diminished in schizophrenic individuals. Schizophrenia patients demonstrate impairment in indicating the presence of a briefly flashed, degraded face, but respond indistinguishably from controls in trials in which the face is vertically flipped and scrambled. This indicates that the poor performance is not due to generalized visual processing deficits.64 Further, when schizophrenic patients are presented with a depth-inverted face stimulus (a ‘hollow mask’), they are more likely to perceive the veridical hollow face than are control subjects.18 Similarly, scores on a binocular depth inversion test correlate with severity of psychosis as measured by the Brief Psychiatric Rating Scale.65 These findings suggest that there is a learning-dependent predictive perception deficit in facial features that make up a face in schizophrenic subjects relative to healthy controls. Top-down processes exert less guidance of sensory input towards perception of faces, leading to difficulties for patients in recognizing faces and decoding meaning from facial expressions. Schizophrenia patients are impaired in these fundamental processes. Schizophrenia patients exhibit difficulty identifying faces as their own, familiar or unfamiliar66 and display deficits in facial recognition and face-matching tasks.67 Superficial differences between photographs of identical faces, such as lighting conditions and visual angle, more greatly hamper patients' ability to match a target face to a selection of sample faces.68–70 These results again support a greater influence of lower-level visual properties compared to invariant higher-level predictions in face-processing in schizophrenia.

The relative weighting of percepts towards low-level features in schizophrenia likely underlies the deficits in facial emotion perception associated with the illness.71 Additionally, a failure to integrate facial and vocal information in schizophrenia further disrupts emotional perception.72 These impairments in processing of emotional cues underlie the inability to engage with other people in a manner that is mutually rewarding, which lies at the core of the functional disability of people with schizophrenia, leading to various difficulties, such as extreme social isolation and very high rates of unemployment. Such social isolation and reduced necessity for utilizing cognitive skills may further alienate individuals with schizophrenia, and may be associated with the chronic course of the illness.

Negative symptoms

Much of social interaction involves the interplay of internally based viewpoints and commonly shared experience and belief. If the ability to develop accurate percepts of people and a typical internal model of the world is disrupted, one of the most fundamental expected changes would be difficulty engaging in social interaction and volition. Correlation of MMN activity with social cognition within a sample of schizophrenia patients suggests that a deficit in learning-dependent predictive perception may be underlying the pervasive social deficits associated with schizophrenia.73

Neurocognitive deficits

Executive function, verbal and visual working memory, verbal and visual memory, attention, visuospatial ability and speed of processing deficits are seen in many patients with schizophrenia.74,75 Impairment of learning-dependent predictive perception in schizophrenic patients represents a unifying deficit underlying the defined cognitive deficits associated with the disease. Prediction based on previous experience serves to speed perception, thought and action by priming cortical columns expected to be driven by bottom-up input. Additionally, prediction serves to direct attention to sensory features of expected importance, or to stimuli that violate predictions. Because elements of the sensory world are less contextually integrated in schizophrenia, the predictive signal that cascades from higher to lower cortical regions is incomplete. More of the sensory world is unpredictable for someone with schizophrenia, the salience of items is developed more randomly, sensations compete for limited attentional resources, and the formation of long-term memories is inadequate. Impairments of learning-dependent predictive perception thus impair performance on a broad range of tasks that depend on key cognitive abilities, such as processing speed, long-term memory, and attention.

CONCLUSION

  1. Top of page
  2. Abstract
  3. MODEL OF CORTICAL HIERARCHY
  4. SYMPTOMS
  5. CONCLUSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES

In this paper we extend our model to describe how impairment in learning-dependent predictive perception is the underlying basis of schizophrenia and can lead to the development and expression of the classical features of schizophrenia. The model that we have proposed lends itself to further exploration and testing. When validated, it may serve as a basis for a phenotype to explore the various aspects of the phenomenology of schizophrenia. It can be the basis for developing tests for learning-dependent predictive perception. A battery of such tests can become the basis of identifying at-risk patients and potential future long-term studies of at-risk subjects.

ACKNOWLEDGMENTS

  1. Top of page
  2. Abstract
  3. MODEL OF CORTICAL HIERARCHY
  4. SYMPTOMS
  5. CONCLUSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES

This work was supported by the National Medical Research Council Translational and Clinical Research Program NMRC/TCR/003/2008, Singapore. Dr Krishnan reports that he has holdings in Orexigen and indirect holdings and consultancy in Cenerx. He has consulted for Eisai in the past 12 months. Mr Kraus reports no competing interests. Dr Keefe reports that in the past 12 months he has received investigator-initiated research funding support from the National Institute of Mental Health, GlaxoSmithKline, Novartis, Department of Veteran's Affairs, and the Singapore National Medical Research Council. He also reports in the past 12 months having received honoraria, served as a consultant, or advisory board member for Abbott, BiolineRx, BMS, Cypress Bioscience, Eli Lilly, EnVivo, Lundbeck, Merck, Memory, Pfizer, Roche, Shire, Solvay, Sunovion, and Takeda. Dr Keefe receives royalties from the Brief Assessment of Cognition in Schizophrenia (BACS) testing battery and the MATRICS Battery (BACS Symbol Coding). He is also a shareholder in NeuroCog Trials, Inc.

REFERENCES

  1. Top of page
  2. Abstract
  3. MODEL OF CORTICAL HIERARCHY
  4. SYMPTOMS
  5. CONCLUSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES