When navigating a familiar environment, your behavior may vary depending on your goals. For example, you might take a detour to check out a restaurant you have been curious about, but if you're in a rush, you will opt for the shortest route. Our research focuses on how the brain's neural population, particularly in regions associated with goal-directed behavior (the orbitofrontal cortex, OFC) and navigation (the retrosplenial cortex, RSC), orchestrates the transition between competing goals during navigation. To explore this, we designed a virtual-reality maze task for two Rhesus monkeys, where they had to search for fixed reward locations. Throughout the task, we recorded neural signals from the RSC and OFC. Using a hidden Markov model (HMM), we identified two distinct behavioral strategies the monkeys used during navigation: one aimed at maximizing information collection (surveying state) and the other focused on maximizing rewards (deliberation state). We hypothesize that each goal would function like separate fixed-point attractors, and a mixture of each behavioral goal shown in choice level would exhibit mixed effects of two distinct fixed-point attractors. However, instead of two separate attractors in the flow field of neural dynamics, we found the composition of two different dynamics for a single fixed-point attractor explained the two seemingly distinct goals. Altogether, our results show that control of a mixture of dynamics underlies the seemingly distinct action policy.