Simultaneous EEG-fMRI Study Reveals Separate Bias Mechanisms Modulated by Stimulus Evidence During Rapid Face Processing

摘要

Perceptual choice is affected not only by the stimulus evidence present in the decision alternatives, but also by the propensity to choose one alternative over another. In the drift-diffusion model (DDM), such bias can be expressed either as a change in drift rate or a change in starting point of the decision process. Here, we collected simultaneous EEG and fMRI data while human subjects were performing a three-choice (Face vs. Car vs. House) visual categorization task. We spatially and temporally dissociated the neural correlates underlying the face decision bias as a function of stimulus evidence. Firstly, by fitting the DDM, we quantitatively showed that the change in drift rate best accounted for the decision bias towards faces when sensory evidence was abundant, whereas the shift in starting point best explained the bias effect when inadequate sensory evidence was present. Secondly, we used the EEG single-trial variability to temporally identify brain regions modulated by the two sets of subject-wise bias parameters in the fMRI analysis. Imaging results showed a double dissociation of the bias effects in space and in time as the level of sensory evidence changed: bias in drift rate correlated only with an early sensory network while bias in starting point activated a distributed late decision-related network.