Late, not early, stages of Kanizsa shape perception are compromised in schizophrenia
Introduction
Schizophrenia is a devastating psychiatric disorder characterized by delusions, hallucinations, disorganized thought, bizarre behavior, flat affect, and declines in social, academic, and vocational functioning. Recent studies from brain imaging and visual psychophysics have revealed a constellation of visual abnormalities that are not immediately apparent from the clinician׳s armchair, ranging from reduced contrast sensitivity (Slaghuis, 1998) to weaker three-dimensional depth illusions (Emrich, 1989, Keane et al., 2013). In the last 30 years, and especially in the last 10 years (Silverstein and Keane, 2011, Uhlhaas and Silverstein, 2005), it has become increasingly apparent that schizophrenia impairs perceptual organization—the process whereby coherent and persisting object representations are derived from spatiotemporally fragmented retinal images. As examples, when faces or line drawings are shown in degraded fashion, persons with schizophrenia are worse than healthy controls at identifying those stimuli (Doniger, Foxe, Murray, Higgins, & Javitt, 2002); and when presented with a scattered array of oriented elements (Gabors), patients are less adroit at representing a subset of those Gabors as forming a single contour (Silverstein et al., 2009, Silverstein et al., 2012, Silverstein et al., 2000). The deficit can also lead to a paradoxical performance advantage whenever perceptual organization renders the task more difficult, as with size contrast illusions (Silverstein et al., 2013, Tibber et al., 2013, Uhlhaas et al., 2006).
An unresolved issue is why perceptual organization impairment arises in schizophrenia. Does it occur as a result of a dysfunction of lateral connectivity in early visual cortex, as some have maintained (Kéri et al., 2005, Robol et al., 2013, Dakin et al., 2005)? Or does it instead arise from higher order circuits, perhaps from faulty feedback from frontal and parietal regions (Keane, Silverstein et al., 2012)? We investigated this question with a classic “Kanizsa” square stimulus, in which four white sectored circles form a darkened surface bounded by illusory contours. Kanizsa shapes make for nearly ideal test stimuli: the eliciting conditions have been extensively documented since the 1950s (Geisler et al., 2001, Kanizsa, 1955, Kellman and Shipley, 1991, Lesher and Mingolla, 1993) and the neural underpinnings of the process have been investigated non-clinically with a variety of techniques including EEG, fMRI, single-cell recording, and TMS (Lee and Nguyen, 2001, Maertens et al., 2008, Murray et al., 2002, Wokke et al., 2013), revealing a critical role of feedback from LOC and long-range horizontal connections in V1/V2 (Seghier & Vuilleumier, 2006). Moreover, a key component process of illusory shape perception—contour completion (or contour interpolation)—is important in its own right, allowing species throughout the animal kingdom to extract object shape and number (Nieder, 2002).
The neurobiological substrate and time course of Kanizsa shape perception in schizophrenia have been studied with the scalp-recorded electroencephalogram (EEG), but the results have not always been consistent. Spencer et al. (2003) had observers discriminate Kanizsa shapes from featurally similar fragmented configurations, and measured stimulus-locked phase locking, which records EEG phase variance at a fixed duration after stimulus onset. Healthy controls, but not patients, evinced an early evoked gamma band response (72–98 ms) over occipital electrodes when responding to illusory vs. fragmented shapes. In a follow-up study, “response-locked phase locking”—measured backward in time from a subject׳s button press—was greater for the illusory than the fragmented stimulus for both groups. However, the difference arose at a higher frequency for controls than patients (31–44 Hz vs. 22–24 Hz) (Spencer et al., 2004). The reduced oscillation frequency was considered to reflect disrupted early “feature-binding” (though see below).
Foxe, Murray, and Javitt (2005) applied a virtually identical behavioral paradigm as above, but analyzed high density visual evoked potentials (VEPs) rather than oscillations. They found that the N1 component (106–194 ms) was enhanced for the illusory relative to the fragmented stimulus for patients and controls. It was thereby argued that illusory contour formation is intact in schizophrenia. The interpretation is plausible given that: (i) the N1 component reflects ventral stream processing, especially in the lateral occipital complex, a primary locus for illusory contour formation (Doniger et al., 2002, Halgren et al., 2003, Mendola et al., 1999, Murray et al., 2004, Murray et al., 2002); and (ii) the N1 time frame corresponds to the period in which illusory contours form in behavioral and neurophysiological studies (Gold and Shubel, 2006, Guttman and Kellman, 2004, Keane et al., 2007, Lee and Nguyen, 2001, Ringach and Shapley, 1996).
Importantly, both sets of studies left open the possibility of abnormal high-level contributions to Kanizsa shape perception. Foxe et al. (2005) unexpectedly found greater right inferior frontal activation among patients in the time period of the so-called NCl waveform (240–400 ms), an established signature of perceptual closure (Butler et al., 2013). Spencer et al. (2004) discovered that the two strongest clinical correlates of reduced response-locked phase locking were global thought disorder (r=.61) and one of its components, conceptual disorganization (r=.58). These symptom correlates were estimated on the basis of how clearly a patient communicated during a clinical interview (see below) and suggest that Kanizsa shape perception is at least associated with higher order cognition (Silverstein et al., 2013, Uhlhaas et al., 2006). Spencer and colleagues also acknowledge that reduced synchrony over occipital electrodes could be ascribed to an aberrant high-level template matching process in which a configuration is recognized as a target.
Importantly, behavioral results from the above EEG studies could not settle whether perceptual differences exist in schizophrenia. Lower patient accuracy in the discrimination task (as in Spencer et al., 2003, Spencer et al., 2004) could be blamed on generalized deficits—that is, reduced attention or motivation. Normal patient accuracy (as in Foxe et al., 2005) could be explained by ceiling effects, since the task was extremely straightforward and the stimuli so distinct. Therefore, considered jointly, the above EEG studies have not made it clear whether Kanizsa shape perception deficits exist in schizophrenia, let alone the level at which such deficits arise. What is needed, and what we provide here, are the first psychophysical data that directly address this issue.
We probed Kanizsa shape perception with a variation of Ringach and Shapley׳s (1996) “fat/thin” shape discrimination task, which has been extensively employed to understand perceptual development (Hadad, Maurer, & Lewis, 2010), modal and amodal completion (Kellman, Garrigan, Shipley, & Keane, 2007), completion speed (Guttman & Kellman, 2004) and autism (Milne & Scope, 2008), among other issues. On each trial of our experiment, subjects discriminated the orientations of four sectored circles or pac-men (Gold et al., 2000, Gold and Shubel, 2006, Guttman and Kellman, 2004, Keane et al., 2007, Murray et al., 2001, Pillow and Rubin, 2002, Ringach and Shapley, 1996, Zhou et al., 2008). In the illusory condition, the sectored circles jointly formed a Kanizsa square, and subjects decided whether the elements formed a fat or thin shape (see Fig. 1A). In a control (“fragmented”) condition, the sectored circles faced downward to prevent illusory contours, and the task was to discern whether the elements were each rotated left or right. These two conditions have sometimes been described as differing by a geometric property, “relatability”, which governs when elements can and cannot connect via interpolation (Kellman & Shipley, 1991). Half of the trials involved distractor lines, which disrupt illusory contour formation and worsen illusory shape discrimination (Dillenburger and Roe, 2010, Keane et al., 2012, Keane et al., 2013, Ringach and Shapley, 1996, Zhou et al., 2008). Task difficulty depended on pac-man rotational magnitude, with larger rotations making the alternatives easier to distinguish. An adaptive staircase determined the difficulty for a trial and estimated the amount of rotation needed to achieve threshold accuracy (79.7%).
Two metrics were of interest. One is global shape integration, which corresponds to how well subjects distinguish Kanizsa shapes relative to featurally similar fragmented shapes (without distractors). A lower relative threshold in the illusory condition demonstrates an enhanced capacity to take advantage of the Gestalt layout of the stimulus. Also of interest was how well subjects fill-in illusory contours. Filling-in was gauged by how much subjects responded to seemingly irrelevant information (distractor lines) placed near the filled-in paths. The underlying assumption was that the more that subjects fill-in illusory contours, the more that distractors would impair discrimination in the illusory relative to the fragmented condition. This second metric was chosen because others have shown that distractor lines near the edges of Kanizsa shapes worsen illusory shape perception, but exercise little, if any effect when illusory contours are absent (Keane et al., 2012, Keane et al., 2013, Ringach and Shapley, 1996, Zhou et al., 2008). Reverse correlation and other studies have also revealed filling-in by examining behavioral or neural responses to line segments or luminance noise placed near filled-in boundaries (Dillenburger and Roe, 2010, Dresp and Bonnet, 1991, Dresp and Grossberg, 1997, Dresp and Grossberg, 1999, Gold et al., 2000, Gold and Shubel, 2006, Keane et al., 2007, Keane et al., 2013). Certain displays also show phenomenologically that background texture increases contour salience when aligned with the pac-men and degrades contour salience otherwise (Ramachandran, Ruskin, Cobb, Rogers-Ramachandran, & Tyler, 1994).
Our two metrics—global shape integration and filling-in—may seem like they are measuring the same thing, but prior studies suggest otherwise. Murray, Imber, Javitt, and Foxe (2006) employed a fat/thin discrimination task and showed that—within 124–186 ms post-stimulus onset—the response magnitude and scalp topography of the VEP was strongly modulated by the presence of illusory contours but not the accuracy of response. By contrast, a later VEP time period (330–406 ms) depended on response accuracy for illusory configurations. These results were taken to show that boundary completion is automatic and dissociable from shape discrimination. Keane, Lu et al. (2013) utilized a behavioral fat/thin task and reached virtually identical conclusions. Subjects in that study were biased to conceptualize each discriminated stimulus as a single partly visible shape or as a disconnected set of edge elements (group and ungroup strategy, respectively). The elements of the stimulus were geometrically arranged to either allow or prevent illusory contours (relatable and non-relatable conditions, respectively). The “group” strategy enhanced the discrimination of relatable stimuli but not non-relatable stimuli. This provides evidence that how a subject cognitively regards a Kanizsa stimulus can play an important role in how well it is discriminated. At the same time, distractor line effects obtained regardless of strategy, indicating that illusory contours were filled in automatically and independently of cognitive expectation.
We investigated illusory contour and shape completion in the context of a key symptom of schizophrenia, conceptual disorganization, which was evaluated by how a participant communicated during the clinical interview. A conceptually disorganized person׳s speech tends to be long-winded and rambling, with no obvious final objective; it contains logical errors, non-sequiturs and loose associations, so that one topic does not clearly relate to the next; and it is often interrupted with extended pauses, where the individual might completely lose his/her train of thought. In its most severe forms, disorganized speech may be almost completely incomprehensible, producing a veritable “word salad.” There are good reasons to focus on conceptual disorganization. Spencer and colleagues found that it strongly correlated with stimulus- and response-locked phase locking in Kanizsa detection tasks (rs>.57) (2003, 2004). Cross-sectional and longitudinal studies have linked conceptual disorganization with reduced size contrast (Ebbinghaus) illusions and impaired contour integration (Silverstein et al., 2013, Uhlhaas et al., 2006). More generally, thought disorder, a symptom which encompasses conceptual disorganization, has been touted as the “primary defining feature” of schizophrenia in that it unifies an otherwise extremely heterogeneous illness (Andreasen, 1999). Indeed, a disturbance in associative processes was seen as the hallmark of schizophrenia in one of the original formulations of the disorder (Bleuler, 1911/1950).
To summarize, we examined the pac-man rotational magnitude (threshold) needed to reach ~80% accuracy for four different conditions, corresponding to whether or not there were illusory contours (relatability) and whether or not there were distractor lines. Illusory contour formation was assessed by gauging how much distractors increased threshold in the illusory vs. the fragmented conditions; global shape integration was assessed by comparing thresholds for the illusory and fragmented conditions without distractors. If distractor lines increase threshold in the illusory condition with little effect in the fragmented, and if only global shape integration is impaired in the illness, then primarily late integration processing stages would be implicated in schizophrenia. By contrast, if patients are less susceptible to distractor lines and they more poorly distinguish Kanizsa shapes, then early and perhaps also late stages could be impaired. A high-level integration deficit further implies worse global shape integration among conceptually disorganized individuals.
Section snippets
Observers
For all subjects, inclusion/exclusion criteria were: (1) age 18–65; (2) no history of electroconvulsive therapy; (3) absence of neurological, pervasive developmental, or affective disorders; (4) no drug dependence in the last 6 months (as assessed with the Mini International Neuropsychiatric Interview 6.0 (Sheehan et al., 1998); (5) Full Scale IQ≥75, as estimated with a vocabulary test (Zachary, 1991) (cut-off similar to Spencer et al., 2003, Spencer et al., 2004); (6) no brain injury due to
Results
Fig. 2 shows the threshold results for each condition and group; Fig. 3 directly compares the filling-in and global shape integration metrics for those same groups. There were main effects of relatability, distractor, and group (F(1,91)=6.85, p=.01, =.07; F(1,91)=36.42, p<.001, =.286; F(1,91)=11.88, p=.001, =.116). An expected distractor by relatability interaction (F(1,91)=18.417, p<.001, =.168) revealed that the distractors raised threshold more in the illusory than the fragmented
Discussion
We employed a variant of the well-validated fat/thin discrimination task to improve upon prior schizophrenia studies, which have not yet established the existence, let alone neural locus, of Kanizsa shape perception impairments. We found that, regardless of group, distractors raised thresholds more in the illusory relative to fragmented conditions, suggesting that patients fill-in and rely upon illusory contours during Kanizsa shape discrimination. Yet patients simultaneously demonstrated
Acknowledgments
This research was supported by an NIH National Research Service Award (F32MH094102) to BPK, an NIH R01MH084828-01 to SMS, and an American Psychological Foundation award to SMS. We thank Deepthi Mikkilineni for helping with data collection and Michael Gara for suggestions on data analysis. We also thank the staff and patients at Rutgers University Behavioral Health Care for their time and efforts.
References (95)
- et al.
Antipsychotic dose equivalents and dose-years: A standardized method for comparing exposure to different drugs
Biological Psychiatry
(2010) - et al.
Weak suppression of visual context in chronic schizophrenia
Current Biology
(2005) - et al.
Psychophysical evidence for low-level processing of illusory contours and surfaces in the Kanizsa square
Vision Research
(1991) - et al.
Contour integration across polarities and spatial gaps: From local contrast filtering to global grouping
Vision Research
(1997) - et al.
Spatial facilitation by color and luminance edges: Boundary, surface, and attentional factors
Vision Research
(1999) - et al.
Edge co-occurrence in natural images predicts contour grouping performance
Vision Research
(2001) - et al.
Deriving behavioural receptive fields for visually completed contours
Current Biology
(2000) - et al.
Contour interpolation revealed by a dot localization paradigm
Vision Research
(2004) - et al.
The development of contour interpolation: Evidence from subjective contours
Journal of Experimental Child Psychology
(2010) - et al.
Cortical activation to illusory shapes as measured with magnetoencephalography
Neuroimage
(2003)
Magnetoencephalographic responses to illusory figures: Early evoked gamma is affected by processing of stimulus features
International Journal of Psychophysiology
Gamma responses and ERPs in a visual classification task
Clinical Neurophysiology
Classification images reveal spatiotemporal contour interpolation
Vision Research
Is interpolation cognitively encapsulated? Measuring the effects of belief on Kanizsa shape discrimination and illusory contour formation
Cognition
A theory of visual interpolation in object perception
Cognitive Psychology
Bayesian adaptive estimation of psychometric slope and threshold
Vision Research
Synchronous gamma activity: A review and contribution to an integrative neuroscience model of schizophrenia
Brain Research Reviews
Syndromes of schizophrenia and smooth-pursuit eye movement dysfunction
Psychiatry Research
The role of edges and line-ends in illusory contour formation
Vision Research
Reduced facilitation effect of collinear flankers on contrast detection reveals impaired lateral connectivity in the visual cortex of schizophrenia patients
Neuroscience Letters
Perceptual completion across the vertical meridian and the role of early visual cortex
Neuron
Lateral interactions between spatial channels: Suppression and facilitation revealed by lateral masking experiments
Vision Research
On the perception of illusory contours
Vision Research
The genetics of symptom dimensions of schizophrenia: Review and meta-analysis
Schizophrenia Research
Spatial and temporal properties of illusory contours and amodal boundary completion
Vision Research
Functional neuroimaging findings on the human perception of illusory contours
Neuroscience & Biobehavioral Reviews
Perceptual organization, the disorganization syndrome, and context processing in chronic schizophrenia
Schizophrenia Research
Perceptual grouping in disorganized schizophrenia
Psychiatry Research
Visual sensory processing deficits in first-episode patients with Schizophrenia
Schizophrenia Research
A unitary model of schizophrenia: Bleuler׳s “fragmented phrene” as schizencephaly
Archives of General Psychiatry
A cortical mechanism for triggering top-down facilitation in visual object recognition
Journal of Cognitive Neuroscience
The clinical translation of a measure of gain control: The contrast-contrast effect task
Schizophrenia Bulletin
Dementia praecox or the group of schizophrenias
The psychophysics toolbox
Spatial Vision
An event-related potential examination of contour integration deficits in schizophrenia
Frontiers in Psychology
Measurement of premorbid adjustment in chronic schizophrenia
Schizophrenia Bulletin
Texture interactions determine perceived contrast
Proceedings of the National Academy of Sciences of the United States of America
Gamma oscillations and object processing in the infant brain
Science
Influence of parallel and orthogonal real lines on illusory contour perception
Journal of Neurophysiology
Impaired visual object recognition and dorsal/ventral stream interaction in schizophrenia
Archives of General Psychiatry
A three-component-system hypothesis of psychosis. Impairment of binocular depth inversion as an indicator of a functional dysequilibrium
British Journal of Psychiatry Supplement
Filling-in in schizophrenia: A high-density electrical mapping and source-analysis investigation of illusory contour processing
Cerebral Cortex
The computation of occluded contours in visual agnosia: Evidence for early computation prior to shape binding and figure-ground coding
Cognitive Neuropsychology
The spatiotemporal properties of visual completion measured by response classification
Journal of Vision
Sex, symptom, and premorbid social functioning associated with perceptual organization dysfunction in schizophrenia
Frontiers in Psychology
“Margini quasi-percettivi in campi con stimulazione omogonea”
Rivista di Psicologia
Cited by (23)
Systematic review of visual illusions in schizophrenia
2023, Schizophrenia ResearchMagnocellular and parvocellular contributions to brain network dysfunction during learning and memory: Implications for schizophrenia
2022, Journal of Psychiatric ResearchAberrant Cortical Connectivity During Ambiguous Object Recognition Is Associated With Schizophrenia
2021, Biological Psychiatry: Cognitive Neuroscience and NeuroimagingBrain network mechanisms of visual shape completion
2021, NeuroImageContour interpolation: A case study in Modularity of Mind
2018, Cognition
- 1
Equal contributions.
- 2
Current address: University of California San Diego, Center for Behavior Genomics, 9500 Gilman Drive, La Jolla, CA 92093-0603, USA.