Adaptive thought and action

Selecting an appropriate response in an uncertain world is central to the core of intelligence. Such adaptive, intelligent behavior is thought to depend on experience-derived internal representations of the environment which are responsive to feedback. However, internal representations of the environment do not, themselves, enable organisms to select and implement actions. Because the world is dynamic and uncertain, it is unlikely that the brain operates like a great stimulus-response switchboard, and, to the contrary, is thought to encode abstract principles or rules of the task at hand which can guide behavior. Such abstract rules allow behavior to extend beyond specific circumstances to general situations while also considering factors such as current motivation and reinforcement history. Thus, the selection of which abstract rule ought to guide behavior in response to how successfully one’s goals were achieved likely underpins the flexibility and adaptability that are central to intelligent behavior.

The use of abstract rules is thought to depend on the dorsolateral prefrontal cortex (DLPFC) because its damage often results in impaired rule use and reduced cognitive flexibility. Nonhuman primate DLPFC neurons have been shown to encode abstract rules for interacting with categories of stimuli. Importantly, these studies found that DLPFC neurons almost exclusively represented abstracted environmental features (categories) and strategies for interacting with those features (rules) and were nearly devoid of activity attributable to purely sensory or memory-related information. These results suggest that neurons in the DLPFC encode an internal representation of the task at hand. However, what is unclear is whether/how environmental change is reflected in the DLPFC’s neuronal model of the task at hand and how changes in those internal representations initiate/sustain cognitive and behavioral change.

Therefore, a major objective of my research is to characterize the neuronal physiology underpinning the process of shifting from one response strategy to another. So far, abstract rule use has been studied only in situations when animals were cued as to which rule to use in an upcoming trial. Although choices are sometimes guided by cues or instructions, in many everyday situations voluntary behavior requires the selection of actions based on the expected value of available options. Currently, it is unknown how the brain utilizes extended action-outcome histories to flexibly determine the optimal way to think and act. Therefore, to explore the process by which integration of recent reinforcement information into existing DLPFC neuronal representations can initiate a shift in which set of DLPFC rule-encoding neurons guide future behavior, I am acutely recording DLPFC neurons from non-human primates (NHPs) as they switch between two general rules in a delayed-match/non-match-to-sample task. This task requires the animals to determine and stably implement one of two possible decision strategies based on reinforcement feedback. Only one decision strategy could yield reinforcement on a given trial. Every so often, the strategy-outcome contingencies reverse such that the animals must think differently in order to solve the next block of trials. By monitoring DLPFC neurons as the animals switch from one strategy to another, my research aims to characterize the neurophysiological motifs underlying adaptive thought and action.

Translating ideas into actions

An idea is only as good as its execution. Given that PFC neurons encode strategies for interacting with the world, how are those strategies behaviorally implemented? In my task, animals report their decisions via eye movements. Therefore, for my task, the key motor structure of interest is the frontal eye field (FEF), a prefrontal area involved in the control of voluntary eye movement. Given that the DLPFC and FEF are adjacent to one another, a logical question is whether rule-related information in the DLPFC is translated into rule-guided eye-movements via its connectivity with the adjacent FEF. Thus, by performing simultaneously electrophysiological recordings in the DLPFC and FEF, a major goal of my research is to determine whether and how DLPFC rule signals are translated into eye-movement signals in the FEF.