). The meta-learning training phase consisted of 6 cycles of 10 Null trials followed by five pairs of E and R trials . +5° rotation was used in E. Since the learner received the reward feedback in R for multiple trials in the training phase, the updated memory after the first R trial is potentially influenced by the accumulation of a reward effect.
The experiments comprised four blocks. The first block contained a Probe phase followed by a meta-learning training phase and another Probe phase. The remaining blocks consisted of a meta-learning training phase followed by a Probe phase. 30 Null trials and one-minute breaks were inserted before and after each Probe phase. The score was removed in R trials in the first block to measure baseline behavior without reward feedback.as follows.