Commentary: Probabilistic Representation in Human Visual Cortex Reflects Uncertainty in Serial Decisions

In this commentary, I highlight two points implied by the main findings of van Bergen and Jehee's article: (1) that uncertainty from the encoded stimulus is stored in short-term memory and (2) that knowledge about different statistical environmental regularities might affect how the brain integrate past information with current decisions. While the first should inform the development of models of short-term memory storage, specially those concerning short-term synaptic plasticity, a candidate mechanism to explain serial dependence (Barbosa et al., 2020), the second has important implications for how we understand the accumulation of prior information and how it affects current behavior. For downstream areas to have access to previous trial probability distribution of stimulus values, the brain must keep it in short-term memory. van Bergen and Jehee's (2019) results imply that the brain is able to maintain a representation of uncertainty in short-term memory. [...]future probability population coding models should account for uncertainty representation through short-term synaptic plasticity, either as a distributed representation of synaptic weights in the network or as a summary statistics extracted from the population activity and later used in the integration of previous and current trial information.