Gaze behaviors during free viewing revealed differences in visual salience processing across four major psychiatric disorders: a mega-analysis study of 1012 individuals

The aberrant salience hypothesis posits that misattribution of salience underlies psychotic symptoms in schizophrenia. This salience involves both emotional arousal and perceptual salience from novelty and sensory features. Neuroimaging studies have shown reduced midbrain signaling in schizophrenia patients related to various salience types. While different domains of salience have distinct neural underpinnings, they may share a common computational framework. Previous research on gaze behavior in psychiatric disorders, particularly schizophrenia, has revealed abnormal scan paths during free viewing, suggesting limitations in the area of focus compared to healthy controls. This study aimed to determine whether visual salience is affected in schizophrenia and if this abnormality is unique to the disorder, expanding upon prior research with a much larger sample size and multiple psychiatric diagnoses.

Numerous studies have documented abnormal gaze patterns in individuals with schizophrenia during free viewing tasks, characterized by narrower focus areas compared to healthy controls. These studies used various stimuli such as pictures, geometric drawings, and other visual elements, consistently pointing to disruptions in visual exploration strategies. The current study builds on this existing body of work by employing a larger, multi-site sample and incorporates a more comprehensive computational model to analyze visual salience.

1012 participants from seven institutes were included: 550 healthy controls and 238, 41, 50, and 133 individuals with schizophrenia, bipolar disorder, major depressive disorder, and autism spectrum disorder, respectively. Participants viewed 20 static images (8 seconds each) across five categories (buildings, scenes, food, portraits, grayscale comics) under free viewing conditions. Eye movements (left eye) were recorded at 1 kHz using an Eyelink 1000 Plus system. Data were preprocessed to segment blinks, saccades, and microsaccades. Visual salience was computed using the Itti-Koch model, a validated computational model of visual attention, assessing salience based on luminance, color, and orientation features. Salience scores represented the average salience at participants' gaze locations. Statistical analyses involved linear regression models within institutes, assessing effect sizes for different salience features. Cross-institute analysis was performed using a general linear model, controlling for age and sex. Post-hoc comparisons with Bonferroni corrections were conducted to determine group differences. Additional correlations were evaluated between salience scores and symptom severity scales (e.g., PANSS), along with medication dosage.

Schizophrenia patients exhibited significantly higher salience scores (full model and individual features) compared to healthy controls. This effect was consistent across individual institutes. The highest effect sizes were observed for orientation salience, followed by color salience. Cross-disorder comparisons revealed significantly higher salience scores in schizophrenia patients compared to those with major depressive disorder and autism spectrum disorder. The order of salience score magnitude was consistent with the clinical severity of psychosis across disorders. No significant correlations were found between salience scores and symptom severity or medication dosage.

The consistent finding of elevated visual salience in schizophrenia supports the aberrant salience hypothesis, suggesting impaired perceptual salience processing. The stronger effect for orientation salience is consistent with previous research on orientation processing deficits in schizophrenia and aligns with potential implications for contour integration difficulties. The cross-disorder comparisons highlight the potential utility of visual salience analysis in differentiating between psychiatric disorders, particularly schizophrenia, based on the severity of altered salience processing. This aligns with the clinical observation of psychosis severity. The results show a consistent pattern across multiple sites, increasing the confidence in the findings.

This large-scale study provides robust evidence for elevated visual salience in schizophrenia patients during free viewing, consistent with the aberrant salience hypothesis. The observed differences in visual salience processing across disorders suggest a potential biomarker for distinguishing schizophrenia from other psychiatric conditions. Future research should investigate the generalizability of these findings to other populations (children, elderly) and explore the relative contributions of illness and medication effects. Examining the relationship between salience processing, symptoms, and treatment response would also be valuable.

The study mainly included adult participants, limiting generalizability to other age groups. The presence of medicated patients necessitates further investigation to disentangle disease and medication effects. The reliance on diagnosis-specific symptom scales in the correlation analyses limits the possibility of exploring transdiagnostic relationships between salience processing and symptoms. Future studies should address these limitations to expand our understanding of visual salience processing in psychiatric disorders.