Abstract
In perceptual decisions, subjects infer hidden states of the environment based on noisy sensory information. Here we show that both choice and its associated confidence are explained by a Bayesian framework based on partially observable Markov decision processes (POMDPs). We test our model on monkeys performing a direction-discrimination task with post-decision wagering, demonstrating that the model explains objective accuracy and predicts subjective confidence. Further, we show that the model replicates well-known discrepancies of confidence and accuracy, including the hard-easy effect, opposing effects of stimulus variability on confidence and accuracy, dependence of confidence ratings on simultaneous or sequential reports of choice and confidence, apparent difference between choice and confidence sensitivity, and seemingly disproportionate influence of choice-congruent evidence on confidence. These effects may not be signatures of sub-optimal inference or discrepant computational processes for choice and confidence. Rather, they arise in Bayesian inference with incomplete knowledge of the environment.
Highlights
In perceptual decisions, subjects infer hidden states of the environment based on noisy sensory information
We show that a Bayesian framework with optimal inference but incomplete knowledge about the environment can explain choice accuracy, confidence, and their discrepancies in experimental measurements
Our model extends Partially Observable Markov Decision Processes (POMDPs)[23], which assume that subjects optimize a reward function by adjusting their beliefs about stimulus identity and the best choice based on two factors: sensory observations and prior knowledge about environmental states[24,25,26], which are learned from past experience
Summary
Subjects infer hidden states of the environment based on noisy sensory information. We show that the model replicates well-known discrepancies of confidence and accuracy, including the hard-easy effect, opposing effects of stimulus variability on confidence and accuracy, dependence of confidence ratings on simultaneous or sequential reports of choice and confidence, apparent difference between choice and confidence sensitivity, and seemingly disproportionate influence of choicecongruent evidence on confidence These effects may not be signatures of sub-optimal inference or discrepant computational processes for choice and confidence. These discrepancies have been occasionally interpreted as evidence for suboptimality of the decision-making process or for disparate processes for computing choice and confidence Contrary to those interpretations, we show that a Bayesian framework with optimal inference but incomplete knowledge about the environment can explain choice accuracy, confidence, and their discrepancies in experimental measurements. We demonstrate the precision of our predictions about choice confidence by testing them on monkeys performing a direction discrimination task with post-decision wagering[2], where both choice accuracy and confidence were measured
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have