rTMS to right parietal cortex decreases the precision of spatial priors in perceptual decision making

Abstract

AbstractThroughout life human observers make perceptual decisions under uncertainty guided by prior knowledge about the world’s causal structure and properties. According to Bayesian probability theory, optimal decision making relies on integrating prior knowledge with current sensory inputs, weighted by their relative precisions (i.e., inverse of uncertainty). Thus, representing uncertainty is essential for optimal decisions. Although extensive research suggests that human perceptual decision making follows Bayesian principles, the neural underpinnings of priors and their uncertainties remain elusive. In this five-day study, we employed psychophysics, Bayesian causal inference models, and offline low-frequency (i.e., 1Hz) transcranial magnetic stimulation (TMS) to probe the role of right intraparietal sulcus (IPS), a key region for spatial processing, in the integration of prior knowledge with auditory/visual inputs for spatial decisions. Model-free and Bayesian modelling analyses consistently showed a reduction in the precision of observers’ long-term spatial prior and in the influence of their previous spatial choices on their current spatial decisions for right IPS-TMS compared to sham-TMS. In contrast, the causal prior and the auditory/visual uncertainties remained unaffected. The results show that offline IPS-TMS can selectively reduce the precision or influence of observers’ long-term spatial prior and their short-term spatial expectations on perceptual decisions, without affecting their causal prior or sensory uncertainties (i.e., likelihood). Our findings provide causal evidence for the role of parietal cortex, situated at the top of the audiovisual spatial processing hierarchy, in encoding the uncertainty of spatial - but not causal - priors during perceptual decision-making.Significance statementPerceptual decisions under uncertainty are pervasive in everyday life. Substantial evidence suggests that humans perform perceptual decisions near-optimally. They combine sensory inputs with prior knowledge about the signals’ causal structure and locations weighted by their uncertainties. Yet, the neural underpinnings remain elusive. Combining psychophysics, Bayesian models, and offline low-frequency inhibitory transcranial magnetic stimulation (TMS), we provide causal evidence that the parietal cortex is crucial for encoding the uncertainty of spatial - but not causal - priors during perceptual decision-making. Intriguingly, observers relied less on their long-term and short-term prior spatial expectations after parietal-TMS, as compared to sham-TMS. These results provide important insights into the neural substrates of priors and their uncertainties as key ingredients for near-optimal decisions consistent with normative Bayesian principles.