CHOICE-PREDICTIVE BIASES IN DORSAL PREMOTOR CORTEX AND PARIETAL REACH REGION DURING MULTI-ATTRIBUTE VALUE-BASED DECISIONS

Many previous studies have shown that neural activity in prefrontal areas reflects information pertinent to ongoing decisions, suggesting that decision-making is part of the frontal lobe’s executive functions. However, decision-related neural modulation is also observed in sensorimotor areas. Does this imply that these areas are all causally involved in the decision processes? Here, we investigated the role of dorsal premotor (PMd) cortex and parietal reach region (PRR) using a multi-attribute value-based decision task. Two male rhesus monkeys were presented with reach targets whose reward value was indicated by two attributes - its brightness (”bottom-up” feature) and its orientation (“top-down” feature). We predicted that the latency of choice-predictive activities in PMd and PRR would reflect their causal role, especially when the attributes are in conflict. A population analysis of early-tuned cells showed that 1) both PMd and PRR predicted choices before action execution, 2) PMd often preceded PRR, and 3) conflict resolution between equally-valued choices was biased by baseline activity in PMd, but not PRR. However, a neural space analysis on the entire population showed that 4) tuning dimension trajectories of bottom-up choices deviated from a baseline condition earlier in PRR than in PMd, whereas the opposite was true for top-down choices. These results suggest that value-based decisions are made at the individual attribute levels in a distributed decision network that includes PMd and PRR, as opposed to at an abstract, fully integrated value space exclusive to the prefrontal cortex.