Behavioral correlates of the decision process in a dynamic environment: post-choice latencies reflect relative value...

One characteristic of natural environments is that outcomes vary across time. Animals need to adapt to these environmental changes and adjust their choices accordingly. In this experiment, we investigated the sensitivity with which rats could detect, and adapt to, multiple changes in the environment. Rats chose between two spouts which delivered 5% sucrose rewards with distinct probabilities. Across three phases, reward probabilities changed in size (large or small) and direction (increase or decrease). A discrete trial-structure was used, which allowed the choice process to be decomposed into three distinct response latency measures (choice execution latency, spout sampling duration, and trial-initiation latency). We found that a large decrease in reward probabilities rapidly produced the greatest change in choice proportions. The time taken to execute a choice reflected the differences in reward probabilities across the two spouts in some cases, but also reflected training history. By contrast, the amount of time rats spent responding at reward spouts in anticipation of reward consistently reflected the relative likelihood of reward across the two spouts and not the absolute probability of reward. The latency to initiate the subsequent trial reflected choice evaluation. These three response latencies thus indexed key behavioral correlates of the choice process as it unfolds in time. We discuss how this paradigm can be used to assess the corresponding neural correlates of decision-making.