Neural correlates of motor memory with multiple time scales in sensorimotor adaptation

Recent behavioral studies with computational models have revealed motor memory with multiple time scales in sensorimotor adaptation [1]. However, little has been studied about underlying neural correlates of the motor memory. For this, we designed an even-related fMRI experiment where 21 subjects (15 males and 6 females) learned opposing visuomotor rotation (40 ̊ or -40 ̊) tasks in alternating blocks by which we could prevent premature adaptation within few trials. Then, the traces of motor memory activation with 30 different time scales, logarithmically scaled from 2 s to 92.6 h, were estimated using a computational model as described below [1]. To account for dual adaptation, we defined the state of motor memory with a time constant, τk as a vector with two internal states. The motor adaptation at each trial, n is given by the sum of the 30 states of memory. The update equation of the memory state with a time constant, τk is given as