Human online adaptation to changes in prior probability

Abstract

Optimal sensory decision-making requires the combination of uncertain sensory signals with prior expectations. The effect of prior probability is often described as a shift in the decision criterion. Can observers track sudden changes in probability? To answer this question, we used a change-point detection paradigm that is frequently used to examine behavior in changing environments. In a pair of orientation-categorization tasks, we investigated the effects of changing probabilities on decision-making. In both tasks, category probability was updated using a sample-and-hold procedure. We developed an ideal Bayesian change-point detection model in which the observer marginalizes over both the current run length (i.e., time since last change) and the current category probability. We compared this model to various alternative models that correspond to different strategies -from approximately Bayesian to simple heuristics -that the observers may have adopted to update their beliefs about probabilities. We find that probability is estimated following an exponential averaging model with a bias towards equal priors, consistent with a conservative bias. The mechanism underlying change of decision criterion is a combination of on-line estimation of prior probability and a stable, long-term equal-probability prior, thus operating at two very different timescales. Author summaryWe demonstrate how people learn and adapt to changes to the probability of occurrence of one of two categories on decision-making under uncertainty. The study combined psychophysical behavioral tasks with computational modeling. We used two behavioral tasks: a typical forced-choice categorization task as well as one in which the observer specified the decision criterion to use on each trial before the stimulus was displayed. We formulated an ideal Bayesian change-point detection model and compared it to several alternative models. We found that the data are best fit by a model that estimates category probability based on recently observed exemplars with a bias towards equal probability. Our results suggest that the brain takes multiple relevant time scales into account when setting category expectations. Introduction 1 Sensory decision-making involves making decisions under uncertainty. Furthermore, 2 optimal sensory decision-making requires the combination of uncertain sensory signals 3 PLOS 1/24 with prior expectations. Perceptual models of decision-making often incorporate prior 4 expectations to describe human behavior. In Bayesian models, priors are combined 5 with likelihoods to compute a posterior [1]. In signal detection theory, the effect of 6 unequal probabilities (signal present vs. absent) is a shift of the decision criterion [2]. 7 The effects of prior probability on the decision criterion have been observed in 8 detection [2-4], line tilt [5], numerosity estimation [6, 7], recognition memory [8], and 9 perceptual categorization [9] tasks, among others. These studies generally use explicit 10 priors, assume a fixed effect, and treat lear...

Bayesian transfer in a complex spatial localization task

“…The authors concluded that people can approximate an optimal Bayesian observer by using a switching heuristic that forgoes multiplying prior and sensory likelihood. In another study, Norton, Acerbi, Ma, and Landy (2018) compared subjects’ behavior to the “optimal” strategy, as well as several other heuristic models. The model fit showed that participants consistently computed the probability of a stimulus as belonging to one of two categories as a weighted average of the previous category types, giving more weight to those seen more recently; subjects’ responses also showed a bias toward seeing each prior category equally often (i.e.…”

Humans adapt their anticipatory eye movements to the volatility of visual motion properties

“…In order to quantitatively evaluate the algorithm and following a similar strategy as [63], we computed an overall cost C as the negative log-likelihood (in bits) of the predicted probability bias, knowing the true probability and averaged over all T trials:…”

Using the past to estimate sensory uncertainty

“…showing that learning about a hidden variable such as observers' priors can be accounted for by an exponential averaging model (Norton et al, 2019).…”

“…In the fixed-effects analysis (see Table 2 for details), the Bayesian learner was substantially better than the instantaneous learner across all three experiments, but outperformed the exponential learner reliably only in the sinusoidal sequence. Likewise, the random-effects analysis based on hierarchical Bayesian model selection (Penny et al, 2010, Rigoux et al, 2014) showed a protected exceedance probability that was substantially greater for the Bayesian learner (Sin, RW2) or the exponential learner (RW1, RW2) than for the instantaneous learner ( Figure 4F). However, the direct comparison between the Bayesian and the exponential learner did not provide consistent results across experiments.…”

“…Models were fit individually to observers' data by sampling from the posterior over parameters for each observer (Table 2). We compared the three models in a fixed and random effects analysis (Penny et al, 2010, Rigoux et al, 2014 using the Watanabe-Akaike information criterion (WAIC) as appropriate for evaluating model samples (Gelman et al, 2014) (i.e., a low WAIC indicates a better model, a difference greater than 10 is considered very strong evidence for a model). In the fixed-effects analysis (see Table 2 for details), the Bayesian learner was substantially better than the instantaneous learner across all three experiments, but outperformed the exponential learner reliably only in the sinusoidal sequence.…”