Probabilistic Representation in Human Visual Cortex Reflects Uncertainty in Serial Decisions

Abstract

How does the brain represent the reliability of its sensory evidence? Here, we test whether sensory uncertainty is encoded in cortical population activity as the width of a probability distribution, a hypothesis that lies at the heart of Bayesian models of neural coding. We probe the neural representation of uncertainty by capitalizing on a well-known behavioral bias called serial dependence. Human observers of either sex reported the orientation of stimuli presented in sequence, while activity in visual cortex was measured with fMRI. We decoded probability distributions from population-level activity and found that serial dependence effects in behavior are consistent with a statistically advantageous sensory integration strategy, in which uncertain sensory information is given less weight. More fundamentally, our results suggest that probability distributions decoded from human visual cortex reflect the sensory uncertainty that observers rely on in their decisions, providing critical evidence for Bayesian theories of perception.

A Bayesian and efficient observer model explains concurrent attractive and repulsive history biases in visual perception

“…Furthermore, it exhibits a long exponential decay, with stimuli experienced minutes into the past still exerting repulsive effects. Moreover, we find that this pattern of concurrent attractive and repulsive biases cannot be explained by a Bayesian ideal observer model, which exploits the temporal stability of the environment (van Bergen and Jehee, 2019). We present a novel ideal observer model, in which repulsive biases arise due to efficient encoding (Stocker and Simoncelli, 2006;Wei and Stocker, 2015), and attractive biases are due to Bayesian decoding of sensory information.…”

“…Previous studies proposed that attractive biases in perceptual estimates arise from a probabilistically optimal strategy of decoding sensory information into a perceptual decision (Cicchini et al, 2018;van Bergen and Jehee, 2019). Attractive serial dependencies towards the previous stimulus orientation are well captured by such an observer model that estimates the current stimulus orientation by integrating a noisy sensory measurement of the current stimulus with a prior prediction about the upcoming stimulus orientation in a probabilistically optimal, i.e.…”

“…Previous studies proposed that attractive biases in perceptual estimates towards the previous stimulus arise from a probabilistically optimal strategy of decoding sensory information into a perceptual decision (Cicchini et al, 2018;Kalm and Norris, 2018;van Bergen and Jehee, 2019). In a stable environment, integrating current noisy sensory measurements with information about previous stimuli generally results in more accurate perceptual estimates, compared to relying on noisy sensory measurements alone.…”

“…Like repulsive adaptation, attractive serial dependence has now been reported for a large variety of visual features, such as orientation (Cicchini et al, 2017;Czoschke et al, 2019;Fritsche et al, 2017), numerosity (Cicchini et al, 2014;Corbett et al, 2011;Fornaciai and Park, 2018) and spatial location (Bliss et al, 2017;Manassi et al, 2018;Papadimitriou et al, 2016). This phenomenon may represent a (Bayes-)optimal strategy of decoding the sensory information into a perceptual decision (Cicchini et al, 2018;van Bergen and Jehee, 2019). Effectively, attractive serial dependence may smooth perceptual representations over time, in order to promote visual stability in the face of disruptive factors, such as eye blinks and external or internal noise.…”

“…A fundamental question of human visual perception is how our brains efficiently and accurately process the flood of visual information that continuously bombards our retinae. Importantly, our visual environment exhibits strong temporal regularities (Dong and Atick, 1995;Schwartz et al, 2007;Simoncelli and Olshausen, 2001); for instance, in natural viewing behavior, orientation information tends to be preserved across successive time points and thus stable over short time scales (Felsen et al, 2005;van Bergen and Jehee, 2019). Our brains could potentially exploit this temporal stability and leverage information from the recent past to optimize processing of new visual input.…”