We sorted PMG-NC trials according to the choice of the monkey, and computed DMC values of the neural activity in the late memory period separately for the

trials in which the monkeys freely chose the direct and inferred goals. If the low DMC of a neuron with bimodal selectivity was the averaging result of two opposite unimodal selectivity profiles, one for direct-choice trials and the other for the inferred-choice trials, then a low absolute value of the original DMC would be attended with high absolute values of the two choice-selective DMC values for this neuron. This means, preliminary selection encoding would be indicated by a low similarity between the original and Antidiabetic Compound Library chemical structure the choice-selective DMC values across neurons (Figure 4A, bottom). Vice versa, we can reject the selection hypothesis if a neuron in both choice-selective subsets of trials shows a bimodal selectivity pattern, i.e., when low absolute values of the original DMC is attended with low absolute choice-selective DMC values, resulting in a high similarity between original and choice-selective DMC across neurons (Figure 4A, top). The balanced data set in PRR yielded bimodal BMS-354825 solubility dmso selectivity in PMG-NC trials

separately within direct-choice and within inferred-choice trials. The absolute choice-selective DMC values for direct- and inferred-choice trials were highly similar to the absolute original DMC values

(Figure 4B). This can be seen by the fact that the average distance of the data points from the unity line did not significantly differ from zero neither for direct- (pd > 0.05) nor inferred-choice (pi > 0.05) trials. Idoxuridine When—as a control—the method was applied to the DMG data set, in which we know that the monkeys had selected the motor goal already during the memory period, then the choice-selective and original DMCs were highly and significantly dissimilar (pd = 0.0012, pi = 0.00067; see Figure S2). Additional variance tests indicated that it is unlikely that the bimodal selectivity profiles were the consequence of rapid switching between two alternative preliminary selections within the time of a trial (Figure S2). Taken together, the results from the choice-selective analysis of the balanced data set indicated genuine encoding of potential motor goals rather than alternating preliminary selections in PRR. This supports the motor-goal selection hypothesis and argues against the rule-selection hypothesis. Depending on which stage of the decision process a brain area belongs to, encoding of multiple potential motor goals in that area could represent the multiple options offered to the subject (the “menu”), or the competing behavioral goals associated with these options and weighted with the subject’s preference for either choice. Motor-goal options were defined solely by the task.