Local field potentials in macaque dorsal premotor cortex are modulated by the manipulation of acquired ordinal information in a transitive inference task

Transitive reasoning provides us with the ability to decide among alternatives by connecting previously learned knowledge on an abstract mental representation. We experimentally investigated the neuronal correlates of this ability by recording, through a 96- channels array, the neuronal activity of the dorsal premotor cortex (PMd) of two male macaque monkeys performing a computer-controlled version of the Transitive Inference task (TI). Monkeys had first to acquire, by trial and error, the rank order of pairs of adjacent items as, A>B, B>C, C>D, D>E, E>F (learning phase) of a set of ordered items as A>B>C>D>E>F, then they had to infer the ordinal relation between items never paired in the previous phase as BvsD or CvsE (test phase). In the test phase novel pairs were randomly intermingled with pairs of adjacent items presented during the learning. Both, a symbolic distance (SDe) and a serial position (SPe) effect typically characterize the behavioral outcome in the test phase. The SDe shapes the behavior by increased performance and shorter reaction times (RTs) between items with different ranks than those with similar ranks (BvsE easier than BvsC), while the SPe characterizes the performance as decreased accuracy and longer RTs in comparing pairs of items in middle positions of the list than extreme ones (AvsB and EvsF is easier than CvsD). These behavioral effects are hypothesized to emerge, after the completion of the learning phase, once items adjacent ranks are arranged on a spatially organized mental schema. In the test phase, this schema is explored by inward scanning of the list, starting from the extreme items. In line with the hypothesized cognitive mechanisms, here we observed that the behavior of both monkeys was significantly shaped by the SDe and SPe. Accordingly, this behavioral modulation was reflected in the PMd average power of local field potential obtained through time frequency analyses, confirming an involvement of this brain area in the manipulation of the acquired mental schema. Interestingly, a reflection of the SPe and the corresponding modulation of it was not observed during the learning phase, suggesting these effects emerge only after the acquisition of a mental schema. Overall, our results are in line with the hypothesis that PMd processes the information for completing this task and that a learning process models the neuronal activity for accomplishing the task’s demands.