Regretful choices: neural representations of choice in the orbitofrontal cortex

Abstract

In order to successfully acquire reward under many different circumstances, a decision maker must learn to expect a particular outcome in a specific situation. The ability to predict an expected outcome simplifies the decision process and enables a decision maker to accurately choose between multiple options without having to experience each option independently. The orbitofrontal cortex represents specific outcomes and aids in the selection of actions to acquire specific outcomes without experiencing the direct action-outcome sequence. Without the orbitofrontal cortex, humans, non-human primates, and rats all exhibit an inability to modify their actions to changing reward conditions. In addition, an intact orbitofrontal cortex is required to correctly identify cues paired with and predictive of reward. Without the orbitofrontal cortex the ability to distinguish between differing expectations collapses. The ability to create counterfactuals, a representation of the alternative would-have-been received outcome, is thought to aid in the process of simulating the expected outcomes of a situation. The representation of counterfactuals has been found in the orbitofrontal cortex in humans and non-human primates. While there is some evidence that rats can represent the counterfactual, the only direct neural evidence is presented in this thesis. Without the representation of a counterfactual, it is impossible to experience regret. If the orbitofrontal cortex is homologous across species and if rats can represent the counterfactual, can rats represent the counterfactual during regret inducing situations? Chapters 1 and 2 of this thesis examine the homology of the orbitofrontal cortex comparing the structure of OFC in humans, rats and non-human primates. These chapters also summarize the current hypotheses regarding Orbitofrontal cortex function. Orbitofrontal cortex has been shown to be largely homologous among humans, non-human primates and rats. The third chapter of this thesis introduces the concept of regret and counterfactuals and explains the etymology, psychology and economics that describe how regret and counterfactuals can be studied.Chapter 4 of this thesis shows that rats are capable of representing the counterfactual. One key aspect of orbitofrontal function is in the representation of counterfactuals and regret. Counterfactuals are defined as the alternative, would-have-been option. Rats performing on a multiple T maze, stopped at a decision point and looked both directions before continuing to reward. During this pause, orbitofrontal neurons represented reward after ventral striatal neurons represented reward. The reward representation was general. Once rats arrived at the reward sites, orbitofrontal neurons reliably represented the reward and continued to do so for every reward encounter. However, when rats arrived at reward locations that were not active, orbitofrontal neurons represented the other reward site, representing the counterfactual. Chapters 5 and 6 of this thesis expand on the necessity of counterfactuals to decision making and shows that rats are capable of experiencing regret on an economic foraging task. Rats showed distinct economic preferences on a circular foraging task. Rats waited longer for flavors of reward that they preferred and spent less time waiting for rewards they did not prefer. Neural responses reliably differentiated between reward flavors and the zones associated with the reward flavors. When a rat left a preferred reward early, without receiving reward, then encountered a non-preferred reward with a longer wait, the rat's behavior matched the economic definition of regret. Regret occurs when a decision maker selects between two options and the option finally received is less valuable than what the alternative option would have produced. Importantly this can be differentiated from disappointment, where an option received is less than expected even though it was not the fault of the decision maker. Counterfactuals are necessary for the experience of regret. Humans without an orbitofrontal cortex do not experience regret. However, regret has traditionally been known as a human experience. Economic definitions of regret make it possible to measure regret in rodents. During these regret instances neural ensembles in the orbitofrontal cortex represented the missed, previous action. These representations agree with psychological accounts of regret that state decision makers regret the action that led to outcome more than outcome. In the final chapter, chapter 7, the role of orbitofrontal cortex is re-examined in the context of regret and counterfactuals. Specifically, the previously introduced hypotheses from Chapter 2 are reviewed under the framework introduced by counterfactuals and regret.