Application of a unifying reward-prediction error (RPE)-based framework to explain underlying dynamic dopaminergic activity in timing tasks

摘要: This manuscript is intended as a theoretical companion to Hamilos et al., 2020, in which we examined the role of dopaminergic neurons (DANs) self-timed movements. In that study, recorded DAN signals mice trained initiate licking movement after delay following start-timing cue. both before cue and during timing interval predicted onset, up seconds itself. particular, "ramped up" from time movement. On given trial, slope ramping was predictive when would occur, with steep associated early shallow late movement, reminiscent ramp-to-threshold process. Ramping were recently proposed framework temporal-difference learning under resolved state uncertainty (Mikhael 2019; Mikhael & Gershman, 2014). Here, show an adapted version al.9s model recapitulates signaling observed our task. We also applied results reported recent temporal bisection categorized intervals relatively short or long compared criterion (Soares 2016). The successfully relative amplitude dynamic These combined suggest common neural mechanism broadly underlies behavior: trial-by-trial variation rate internal "pacemaker," manifested reflect stretching compression derivative subjective value function veridical time. this view, faster pacemaking high signaling, whereas slower low levels signaling.