Transient synchrony and dynamical representation of behavioral goals of the prefrontal cortex

Behavioral planning requires organizing actions by integrating perceived or memorized information to achieve goals. Recent studies have suggested that the underlying neural mechanisms involve updating representation of goals for action in associative cortices such as the prefrontal cortex [1]. Although the underlying neural mechanisms are still unknown, we assume that functional linking of neurons would contribute to this transformation of behavioral goals. Thus, we investigated the relation of synchronous firing of neurons to the transformation of goal representation by recording neurons from the dorsolateral prefrontal cortex (DLPFC), while the monkeys performed a path-planning task [2] that requires them to plan immediate goals of actions to achieve final goals. Two monkeys were trained to perform a path-planning task (Fig.1a) that required them to move a cursor to a goal in a lattice-like display. After the cursor emerged in the center of the lattice (Start Display), a goal was presented in a corner (Final Goal Display). The Delay 1 period was followed by the Delay 2 period, in which a part of the path in the lattice was blocked that disabled the cursor to move through the path. Then, a GO signal was provided to allow the monkey to move the cursor for one check of the lattice. To dissociate arm movements and cursor movements, the monkeys to perform with three different arm-cursor assignments, which were changed every 48 trials. Neuronal pairs that were recorded simultaneously during more than two arm-cursor assignment blocks (> 96 trials) were included in the dataset. The analysis for task-related modulation of synchronous firing was based on the time-resolved cross-correlation method (Fig.1b) [3]. This method can estimate neuronal synchrony well, because it can exclude the influence of firing rate change in and among trials by using instantaneous firing rate (IFR) for the predictor. Fig.1b shows an example of the smoothed time-resolved cross-correlogram. In this example, weak and strong increase in co-firing rate of the neuronal pair is seen at Final Goal Display and Delay 2 period respectively, while synchronized firing can be recognized at Delay 1 period without accompanying co-firing rate increase. We selected DLPFC neurons showing significant synchrony and goal-related activity with gradual shift of representation from final to immediate goals before initiation of the action. Many of the DLPFC neurons were found to show transient enhancement of synchrony without firing-rate increases. Furthermore, such enhancement was nearly coincident with the timing of shift in their goal representations. These results suggest that transient synchrony plays an important role in the transforming process of goal representations during behavioral planning.