Factorial variation of saccade vigor with dual decision processes

Abstract

Canonical stochastic models of decision-making treats decision and action as independent and sequential processes. However, studies involving limb movements consistently show that movement duration and kinematics are influenced by the quality of evidence. We tested whether saccade velocity varies with the quality of evidence in monkeys performing a visual search GO/NOGO task in which singleton elongation cued the GO/NOGO stimulus-response rule and the location of a color singleton specified saccade endpoint. We factorially manipulated the efficiency of stimulus-response cue discrimination by varying the elongation of the singleton and the efficiency of singleton localization by varying the color similarity between the singleton and distractors. The effectiveness of the manipulations was revealed by the response times on correct trials that were separately modified by the singleton localizability and stimulus-response cue discriminability, by the incidence of localization and response selection errors with separately modified error response times. Saccade velocity was higher on correct relative to error trials and was inversely proportional to response time. Saccade velocity was separately modified by singleton localizability and stimulus-response cue discriminability. Distinct patterns of error rates and saccade velocity across monkeys indicated individual differences in decision-making strategies. These findings demonstrate that the process selecting endpoints can influence both the timing and dynamics of saccadic eye movements. Incorporating saccade vigor can provide valuable constraints on biologically plausible decision models and help address the persistent challenge of model mimicry.