NEURAL MECHANISMS OF EVIDENCE ACCUMULATION IN CONTINUOUS AND CATEGORICAL PERCEPTUAL DECISION MAKING

Real-world actions, such as sailing, require both categorical choices (e.g., whether to jibe or tack) and continuous estimation (e.g., steering towards a target) while monitoring the environment. While both processes rely on evidence accumulation, it remains unknown whether they share the same neural mechanisms. Here, we address this question by investigating the integration of sequential visual information towards a categorical or continuous choice. Participants (N=48) estimated the average position of seven checkerboard patches briefly presented along a circle, responding via either a directed mouse click (continuous task), or a button press to indicate one of two targets (categorical task). To quantify the subjects’ performance, we modeled evidence integration as the vector sum of stimulus directions, incorporating systematic biases and noise. Fitting the model to single-trial data, we identified: (i) a temporal bias, where early stimuli were overweighted (primacy) in most participants; (ii) an anisotropic spatial bias favoring horizontal over vertical locations; and (iii) a surprise-dependent bias, involving the systematic up-weighting of stimuli that deviated from the running average. Notably, these biases differed between tasks, suggesting distinct underlying neural mechanisms. Furthermore, we found that the centro-parietal positivity (CPP), an EEG signature of evidence accumulation, was modulated by stimulus deviation in the continuous task, providing a neural correlate for the observed surprise-dependent up-weighting. In ongoing work, we examine EEG data from both tasks to further elucidate the neural basis of errors in continuous versus categorical choices.