Abstract
The animals foraging in the wild are required to rapidly detect and acquire valuable objects from their environment. A network of brain regions organized as two hemispheres is involved in processing value information from the visual inputs arising from the contralateral side. The value information processed in these lateralized networks must be compared before an action. The computational processes invol¬ved in this decision-making process are not known. We studied this by recording the behavior of non-human primates performing a value-based choice task in which animals must choose between the two fractal objects presented on the opposite side of the screen on the horizontal axis. We hypothesized that the visual information would be processed separately in the two hemispheres of the brain; hence it can be modeled as independent accumulators. The simplest decision-making model involved these accumulators racing against each other; towards a decision threshold. But it could not predict the observed behavioral performance in the task or the RT distributions. The behavior was well modeled when we implemented an interaction between the racing accumulators. Interestingly we found that the nature of the interaction was highly sensitive to the choice presented to the animal. Equal valued choices resulted in mutual inhibitory interaction between the two hemispheres. In the case of unequal valued choice, the high valued side exerted an inhibitory effect on the opposite side. In contrast, the low-value side exerted an excitatory influence on the opposite side. This suggests that the interactions between the two hemispheres are dynamic and task-sensitive. We validated the dynamic interaction between the two hemispheres through recordings from the superior colliculus neurons showing the biological underpinning of this computational process.