Simple Human Response Times are Governed by Dual Anticipatory Processes with Distinct and Distributed Neural Signatures

Abstract

A major challenge in understanding how expectations guide sensory-motor behavior has been to relate relatively simple algorithmic explanations to their potentially complex neural substrates. We studied this relationship using intracranial recordings from 23 patients with medically refractory epilepsy performing a simple visual-detection task with variable foreperiod delay. We identified two distinct anticipatory influences on behavior by fitting an abstract motor-preparatory model (“rise-to-bound”) to response times (RTs). The first “predictive” process increased reaction speed and premature responses prior to stimulus arrival (“false alarms”), whereas the second “reactive” process increased reaction speed without affecting false alarms. We linked these anticipatory processes to rapidly fluctuating neural-activity patterns across the brain by leveraging the high spatio-temporal resolution and broad coverage of intracranial recordings. Predictive and reactive signals were highly distributed and interspersed throughout the brain but distinguished by their timing (e.g., occurring around response and stimulus onset, respectively), anatomical distribution (e.g., greater representations in MTL and perirolandic areas, respectively), and modulation patterns (e.g., decreased and increased activity during periods of higher expectation, respectively). These results provide new insights into the complex, but identifiable, mappings between specific, behaviorally derived forms of anticipatory information processing and the widespread neural dynamics that implement them.