71. Visual Working Memory Deficits in Schizophrenia: A Case of “Noisy” Perceptual Processing?

Abstract

Abstract Working memory (WM) deficits are a core feature of schizophrenia (SZ) with prognostic significance. Past work indicates that abnormal perceptual encoding plays an important role in WM deficits, but the mechanism of this association is unclear. In this study, we tested the hypothesis that ‘noisy’ perception in SZ cascades to imprecise WM representations by using a human-observer model to estimate internal noise (IN) during visual perception and examining its relation with visual WM precision. Individuals with SZ and demographically-matched healthy controls (HC) completed a perceptual discrimination task to estimate IN levels and an analog visual WM task to examine how IN relates to WM precision. The perceptual discrimination task involved distinguishing orientations of briefly presented gratings (1 cycle/°; tilted ±45° from vertical) embedded in varying levels of external noise (0%–21%). Contrast thresholds were estimated, and IN was modeled from task performance with the perceptual template model. The WM task required reproducing remembered orientations of high-contrast gratings (same size and spatial frequency as in the discrimination task) with a manual dial at a 1s delay. WM precision was computed as the concentration of the von Mises distribution, fit from subjects’ orientation errors. Estimated IN during perceptual discrimination was significantly increased in SZ compared to HC. WM precision was reduced in SZ at every set size. IN was negatively correlated with WM precision for SZ and HC. IN was also negatively correlated with IQ, while WM precision was positively correlated with IQ. We found evidence of elevated IN levels during visual perception in SZ, which were inversely related to their visual WM precision. Results support a model of ‘noisy’ perception in SZ that may underlie broad cognitive impairment. While future work must identify the specific neural drivers of IN level, we posit that it might broadly reflect a functional imbalance of excitatory/inhibitory neurotransmission in related cortical networks. Findings underscore the link between perception and WM encoding in SZ and offer a novel computational strategy for identifying common pathophysiological mechanisms of SZ cognitive dysfunction.