Abstract
Perceptual decisions in the presence of decision-irrelevant sensory information require a selection of decision-relevant sensory evidence. To characterize the mechanism that is responsible for separating decision-relevant from irrelevant sensory information we asked human subjects to make judgments about one of two simultaneously present motion components in a random dot stimulus. Subjects were able to ignore the decision-irrelevant component to a large degree, but their decisions were still influenced by the irrelevant sensory information. Computational modeling revealed that this influence was not simply the consequence of subjects forgetting at times which stimulus component they had been instructed to base their decision on. Instead, residual irrelevant information always seems to be leaking through, and the decision process is captured by a net sensory evidence signal being accumulated to a decision threshold. This net sensory evidence is a linear combination of decision-relevant and irrelevant sensory information. The selection process is therefore well-described by a strong linear gain modulation, which, in our experiment, resulted in the relevant sensory evidence having at least 10 times more impact on the decision than the irrelevant evidence.
Highlights
Perceptual decision-making is the process of making a discrete choice based on available sensory information
We used a motion-only version of the task, which has the advantage that decision-relevant and irrelevant information are represented by sensory neurons with the same properties, and quantifying the neural representation of the sensory evidence is relatively straightforward, as neural responses to random-dot motion stimuli have been studied in detail (Britten et al, 1993)
For stimuli with both motion components having non-zero coherence we evaluated a function of the form e(αrel·crel + αirr·cirr ) p(correct choice) = 1 + e(αrel·crel + αirr·cirr) based on the idea that choices could be made on the basis of a combination of relevant and irrelevant sensory evidence. crel is the strength of the relevant motion component, cirr is the strength of the irrelevant component, with a negative value indicating an incongruent trial, and αrel and αirr describe how much impact the relevant and the irrelevant sensory evidence have on the choice
Summary
Perceptual decision-making is the process of making a discrete choice based on available sensory information. Sasaki and Uka used a random-dot motion stimulus that contained depth information, and monkeys had to perform either a motion direction or depth discrimination (Sasaki and Uka, 2009). We used a motion-only version of the task, which has the advantage that decision-relevant and irrelevant information are represented by sensory neurons with the same properties, and quantifying the neural representation of the sensory evidence is relatively straightforward, as neural responses to random-dot motion stimuli have been studied in detail (Britten et al, 1993). Choices were well-captured by a logistic function of a linear combination of the decision-relevant and irrelevant coherences, suggesting that the selection process might be described as a linear gain modulation of the sensory evidence. We further show that incomplete suppression of the decision-irrelevant sensory information is required to explain the decision behavior and that lapses in remembering which stimulus component had been cued alone are insufficient to account for the observed behavior
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call