Gain control plays an important role in our perception of contrast and motion in that it allows sensory subsystems to maximize the response-difference arising from different stimuli. Several methods have been used for assessing contrast detection in schizophrenia. First, patients with schizophrenia show decreased contrast sensitivity (i.e., need more contrast to detect a grating) across a range of grating-sizes in behavioral studies (19,20
). Second, patients show reduced amplitude responses to simple visual stimuli with steady-state or transient electrophysiological techniques (21,22
), indicating deficits in contrast gain control within the early visual system.
Stimulus response properties of M- and P-neurons overlap significantly, making differentiation difficult, particularly in behavioral studies. Nevertheless, features that bias stimuli toward the M-pathway include high temporal frequency, low spatial frequency, low absolute luminance, and low contrast. Although behavioral studies have found contrast sensitivity deficits across spatial frequencies, often thresholds are relatively low (e.g., < 10% contrast; [20
]), limiting P-pathway involvement. In one study in which thresholds were higher (e.g., > 16% contrast), relative preservation at high spatial frequencies was observed (22
). Similarly, larger contrast sensitivity deficits were found when stimuli were presented dynamically rather than statically, also suggesting greater M-pathway, than P-pathway, impairment (19
In steady-state evoked potential studies, stimuli have been biased toward M- versus P-pathways with different standing levels of luminance contrast (“pedestals”). Under such conditions, differential M- versus P-pathway biased responses have been observed (22
) (B and C). To the extent that P-pathway dysfunction occurs, patient curves show decreased gain at low luminance contrast and a lower plateau, indicating decreased signal amplification, as in the M-pathway. The decreased slope at low contrast and decreased plateau in patients closely resembles results seen after microinfusion of an NDMA antagonist into cat LGN and visual cortex (16,17
) (A and B), consistent with glutamatergic theories of schizophrenia (23–25
A third approach uses an illusion in which the contrast of a small textured disk appears reduced when presented within a high-contrast surround compared with when it is presented in isolation (26
) (). Note that stimuli used in this study were presented greatly above their contrast detection threshold. Patients with schizophrenia were much less susceptible to the illusion, with 12 of 15 patients being more accurate (less biased) than the most accurate control (27
). These results are consistent with decreased center-surround antagonism and hence decreased contrast gain control in schizophrenia patients. Gain control in this illusion might be due to short-range lateral interactions (e.g., γ-aminobutyric acid [GABA]-ergic projections).
Figure 5 The “contrast-contrast” illusion reveals contrast gain control deficits in schizophrenia (reprinted from Curr Biol, 15, Dakin S, Carlin P, Hemsley D, Weak suppression of visual context in chronic schizophrenia, R822–824, copyright (more ...)
A large number of studies have reported motion processing deficits in schizophrenia (28–32
). Motion is signaled by direction-sensitive cells in V1 and then pooled by MT neurons with: 1) larger receptive fields, and 2) center-surround antagonism (as a likely substrate for gain control). A recent study (33
) provides evidence for decreased gain control in schizophrenia in a motion discrimination task. Whereas center-surround antagonism in control subjects resulted in reduced ability to perceive motion of a high-contrast stimulus as its size increased, patients with schizophrenia did not show this reduction in motion perception. Importantly, like Dakin et al.
), these authors find that a disruptive context has less influence on patients than on controls, arguing against nonspecific deficits or lack of attention as an underlying cause of differences. Increased center-surround antagonism, indicative of increased gain control, has also been found in motion studies in schizophrenia (34
Significant correlations between impaired motion perception and M-pathway dysfunction also point to motion processing deficits in schizophrenia resulting from impaired gain control (28
). Patients with schizophrenia show preferential M-pathway dysfunction (21,22,28,35–38
), although deficits have also been observed in parvocellular processing (19,20
). The M-pathway has several properties (speed of processing, low spatial resolution) that make it a suitable physiological substrate for gain control (39
). The P-pathway also exhibits nonlinear gain characteristics, although less so than the M-pathway. Mechanisms of gain control dysfunction include NMDA and GABA-ergic dysfunction. Indeed, NMDA dysfunction seems to be linked to gain control in the M-pathway. Other neurotransmitters (e.g., 40
), which are also implicated in schizophrenia, also modulate visual processing. For example, dopamine deficiency has been linked to impaired perceptual and electrophysiological response to contrast signals including those presented in a center surround paradigm (41,42
). A recent neurophysiological study suggests that nicotine increases gain control in the visual cortex (43
). This might be important in understanding “self-medication” with smoking and strengthens the hypothesis of weak gain control in schizophrenia. It is a challenge to understand and reconcile the involvement of different types of neurotransmitters in visual perception. It is also unclear whether perceptual deficits exhibited by people with schizophrenia for the processing of transient (moving/flickering) stimuli arise from intrinsic dorsal stream dysfunction or from aberrant M-pathway input (21,44
In summary, gain control studies in schizophrenia clearly show that patients have difficulty modulating neuronal responses to take advantage of the surrounding context. There is also evidence that gain control deficits, seen in contrast detection and M-pathway deficits, are important in predicting outcome (22,45
), and are related to higher-level problems in perceptual organization (28,46
) and to symptomatology (20,47–51
Integration in Schizophrenia
Visual integration deficits are seen in contrast, contour, form, and motion processing in schizophrenia. For example, in the last 10 years the connectivity supporting the integration of orientation across space (into extended visual contours) has been studied psychophysically with so-called “flank facilitation” paradigms (52
). Here one measures the detectability of a low-contrast oriented target in the presence of two similar higher-contrast flanking patches arranged so the triplet forms an elongated contour. With some target-flank separations control subjects find it easier to detect the central element when the flanks are present than when they are absent (facilitation). Patients with schizophrenia do not exhibit such a difference, suggesting a failure in ability to integrate the collinear flankers (53
). This would seem to implicate weaker interactions between orientation detectors possibly mediated by abnormal long-range horizontal connectivity in V1.
There are numerous examples of poor form processing in schizophrenia that would seem to directly implicate integration deficits. These include deficits in object recognition, grouping, perceptual closure, face processing, and reading (54–62
). Classic studies show that there is less influence of global on local processing (54,58
). Indeed, patients perform better than control subjects under conditions when global integration would normally interfere with responses to individual elements (54,56,58
). A number of studies have used a psychophysically rigorous contour integration paradigm (63
). This task examines the ability to perceive a contour made up of separate elements within a background of noise elements. Both the contour segments, and background noise elements are small oriented Gabor elements, which are designed to be well-matched to the spatial frequency processing characteristics of orientation-selective simple cells in primary visual cortex (V1); therefore they are ideal for the examination of these features and their integration. Embedded contours constructed from such elements cannot be detected by purely local feature detectors or by the known types of orientation-tuned neurons with large receptive fields (e.g., 64
); their detection requires the integration of local orientation measurements (). Deficits in contour integration have been extensively documented in schizophrenia (57,65–67
). This is thought to result from decreased NMDA-modulated lateral excitation among the spatial filters signaling these elements and the consequent reduction in synchronization of this neural activity ([68
]; see also for reviews [69,70
]). Simpler Gestalt tasks, involving perception of basic shapes with nonfragmented contours, are not affected in schizophrenia, however (71
Figure 6 Performance of the schizophrenia group (dashed line) and healthy control group (solid line) across six conditions of contour element jitter manipulation. The subject's task was to indicate, with a two-button response device, on each trial, whether the (more ...)
Interactions between dorsal and ventral streams and frontal cortex provide one model for how form integration deficits might arise in schizophrenia. Processing is substantially faster via the dorsal stream, which would permit it to prime ventral stream areas (72–74
). A fundamental role of the M system/dorsal stream might be to produce a low-resolution template of the visual scene that influences perceptual processes, such as categorization of natural images, object recognition, and perceptual grouping in the ventral occipito-temporal cortex, by allowing P pathway fine-detailed input to be used more effectively (3,75–80
). With a perceptual closure paradigm Doniger et al.
) found that patients had impaired ability to recognize fragmented pictures. Patients also had decreased amplitude of the dorsal stream-generated P100-evoked potential component, which occurred earlier in time than impairment in the ventral stream-generated closure negativity (Ncl
) component associated with object recognition. Initial P input to the ventral stream was normal as indicated by an intact N1 component. Thus, the impaired behavioral closure and decreased Ncl
seem to be due to lack of interactions between dorsal and ventral stream areas leading to decreased priming of ventral stream. This provides an example of integration deficits due to lack of recurrent interactions in schizophrenia.
As discussed in the preceding text, numerous studies have shown motion processing deficits in schizophrenia (28,30–32,81
). Whereas gain control is involved in motion deficits (e.g., 33), processing of motion also clearly involves integration, because motion is signaled by direction-sensitive cells in V1 whose responses are then pooled by MT neurons with larger receptive fields to signal complex motion.
In summary, there are numerous examples of integration deficits in schizophrenia. Impairments in visual integration have been linked to increases in disorganized symptoms (57,66,67
), poorer premorbid social functioning (82
), presence of childhood trauma in schizophrenia (83
), and illness severity and chronicity (84