Encoding of motor-goal options should lead to the

representations of two potential motor goals during the memory period of all PMG trials, irrespective of any choice preferences of the monkeys. Motor-goal preferences were defined by the monkeys’ average 5-FU purchase choice behavior in PMG-NC trials. Since the monkeys had close-to-equal choice preferences for direct and inferred motor goals in the balanced data set, the bimodal selectivity profiles are not suited to dissociate encoding of motor-goal options versus motor-goal preferences. If, on the other hand, the monkeys had a bias in favor of one of the two options, then encoding of motor-goal preferences should lead to neural activities in the memory period of PMG trials that reflect the relative probability of selecting either potential goal in the PMG-NC trials. By using different reward schedules we recorded two data sets, one with balanced choice behavior (see above), and one with strong behavioral choice bias, to dissociate the options and preference Selleckchem Trametinib encoding hypotheses. In the second data set, correct PMG-NC trials were rewarded according to an equal probability reward schedule (EPRS). With the EPRS, in which

a 50% reward probability independent of the choice history was guaranteed (reward probability: 52 ± 5%; p > 0.05 [A], 50 ± 4%; p > 0.05 [S]), both monkeys showed a strong bias in favor of the inferred reach goal (Figure 5A), i.e., most reaches in PMG-NC trials were directed toward the inferred motor goal (85 ± 4.0% monkey A, 63 ± 4.1% monkey S), and only a small fraction toward the direct goal (2.4 ± 0.8% monkey A, 17.8 ± 3.4% monkey S). In the remaining PMG-NC trials (12.6% monkey Chlormezanone A, 19.2% monkey S) the monkeys aborted the trial without reaching, or reached toward one of the orthogonal goals (<1%). This means that both monkeys had a preference for the inferred goal when the transformation rule was unknown, and when either goal

selection was rewarded with equal probability in EPRS sessions (= biased data set). We can only speculate about the reason for the intrinsic bias of both monkeys during the EPRS (Figure S3). The reason behind this behavior is not immediately relevant for the purpose of dissociating options encoding from preference encoding at the neural population level, though. It is sufficient to note that both monkeys consistently had a similarly strong bias over an extended period of time in the EPRS sessions, and little to no bias in the BMRS sessions. If neurons encoded behavioral choice preferences then we would expect encoding of only the inferred motor goal in the PMG trials of the biased data set, in contrast to the encoding of both potential motor goals simultaneously as seen in the balanced data set.