ABSTRACT
Atypical performance of people with schizophrenia (PSZ) in visual working memory (VWM) has long been attributed to decreased capacity compared with healthy control subjects (HCS). This notion, however, largely ignores the effects of other VWM components, such as precision, on behavioral performance. Here, we measured the performance of 60 PSZ and 61 HCS in a classical delay-estimation task and disentangled the contribution of various VWM components by thoroughly comparing several influential computational models of VWM. Surprisingly, none of the models suggest group differences in memory capacity and in memory resources across set size levels–two diagnostic features of VWM. Notably, we find that the model assuming variable precision (VP) across items and trials is the best model to explain the performance of both groups, indicating the two groups employ the qualitatively same internal process in VWM. According to the VP model, PSZ only exhibit abnormally larger variability of allocating memory resources rather than resources per se. These results directly challenge the widely accepted decreased-capacity theory and establish the first time to our knowledge that the elevated resource allocation variability is the major determinant of the atypical VWM behavior in schizophrenia. Finally, we show that individual differences in the resource allocation variability predict variation of symptom severity in PSZ, highlighting its functional relevance to schizophrenic pathology. Taken together, our findings provide a novel account for the VWM deficits in schizophrenia and shed a new light on the utility of generative computational models to characterize mechanisms of mental deficits in clinical neuroscience.
SIGNIFICANCE STATEMENT Working memory deficits in schizophrenia have long been thought to arise from abnormally decreased capacity. We revisit this topic by leveraging the most recent advances in research on visual working memory. Combining a classical visual working memory paradigm and generative computational modeling, we challenge the conventional decreased-capacity theory and demonstrate that the atypically larger variability of distributing memory resources across items and trials accounts for behavioral deficits observed in schizophrenia. The current study offers a new perspective for future studies aiming for characterizing the diagnostic pathology of schizophrenia.
Authors’ contributions
RZ, YK &YZ developed research idea and study concepts; RZ, YK & YZ design the research; XR and LZ collected the data; RZ performed the data analyses and modeling; RZ & YZ wrote the manuscript.
Acknowledgement
We thank Zheng Ma, Ting Qian and Haojiang Ying for their invaluable comments on the manuscript. This work was supported by the National Social Science Foundation of China (17ZDA323), the National Key Fundamental Research Program of China (2013CB329501), the Major Program of Science and Technology Commission Shanghai Municipal (17JC1404100), the Fundamental Research Funds for the Central Universities (2018ECNU-QKT015), and the NYUECNU Institute of Brain and Cognitive Science at NYU (YK).
Footnotes
↵* co-senior author