The rat frontal orienting field dynamically encodes value for economic decisions under risk

Understanding decision-making under risk is crucial for public health, as excessive or inadequate risk-taking can lead to negative consequences. Expected utility theory, a common economic framework, posits that external rewards are transformed into internal subjective values ('utility'), with the shape of the utility function influencing risk preference. Concave utility functions indicate risk aversion. Previous studies in humans and monkeys have linked expected utility to neural activity in reward and value representation areas, as well as areas involved in orienting decisions like the parietal and frontal cortex. However, causal evidence linking specific brain regions to economic decisions under risk remains limited, particularly in rodents. This study aimed to address this gap by investigating the causal contribution of the frontal orienting field (FOF) and posterior parietal cortex (PPC) in rats performing a risky choice task. The FOF and PPC have been implicated in perceptual decision-making, with perturbations of the FOF consistently impacting perceptual decisions and the PPC exhibiting less reliable effects. It has been argued that the FOF represents animals' current choice, while the PPC encodes momentary evidence. The researchers hypothesized that FOF silencing would cause stimulus-independent biases, while PPC silencing might lead to stimulus-dependent biases.

Studies in humans and monkeys have identified activity related to expected utility in regions associated with reward and value representation, as well as in regions associated with orienting decisions, including the parietal and frontal cortices. However, only a single causal study has directly investigated this relationship, finding that silencing the supplementary eye field in the frontal cortex shifted monkeys to be less risk-seeking. Previous rodent studies investigating the role of FOF and PPC in perceptual decisions have demonstrated that FOF perturbations consistently influence decisions, while PPC perturbations have less reliable effects. The representation in the FOF has been described as 'post-decision', while the PPC directly encodes momentary evidence. This prior work, coupled with existing research on the frontal-parietal network in economic decisions, formed the basis of the study's hypotheses regarding the effects of FOF and PPC silencing on risky choices.

Rats were trained on a risky choice task where they chose between a surebet (small, guaranteed reward) and a lottery (cue-guided magnitude and fixed probability). A model-based quantification of behavior was used, incorporating parameters for utility curvature, decision noise, and choice biases. Pharmacological silencing (muscimol) was used to inactivate the FOF and PPC, followed by optogenetic silencing of the FOF using halorhodopsin. Generalized linear mixed-effects models (GLMMs) were used to analyze the effects of perturbations on risk preference and choice biases. A three-agent mixture model (rational, lottery-preferring, surebet-preferring) was developed to provide deeper insight into the effects of perturbations on decision-making parameters, including utility curvature (p), decision noise (σ), and choice biases (ω). Hamiltonian Monte Carlo sampling of a hierarchical Bayesian model was used for model fitting. A six-node dynamical model was created to explore the potential mechanisms underlying the observed effects of FOF silencing. Finally, single-unit activity was recorded from the FOF of rats performing the task to assess value encoding. Pseudopopulation decoding was used to decode lottery magnitude from population activity.

Bilateral silencing of the FOF resulted in a significant reduction in lottery choices, indicating increased risk aversion. This effect was consistent across multiple animals and inactivation methods (muscimol and optogenetics). Model-based analysis using the three-agent mixture model revealed that FOF inactivation caused a significant decrease in the curvature of the utility function (p), indicating a shift toward a more risk-averse decision strategy. Unilateral FOF silencing produced similar, though smaller, effects. PPC silencing had minimal long-term effects on risk preference. However, a short-lived effect on risk aversion was observed when only early trials were analyzed, suggesting a potential short-term bias effect. Analysis of FOF neural activity revealed value encoding: many neurons showed increased firing rates with higher lottery values, even when controlling for choice. Population decoding of lottery magnitude from FOF activity was highly accurate, demonstrating the clear representation of lottery value in the FOF. The six-node dynamical model suggested that FOF silencing scales down the network's responses to the lottery magnitude, which can explain the observed behavioral shift towards risk aversion. Unilateral PPC inactivation caused a significant ipsilateral bias in 'free choice' trials but not in risky choice trials. This demonstrated that PPC played a short-lived, likely bias-related role in decision-making, but it was not crucial for learning the value of the surebet. The PPC's influence on behavior was also observed in the surebet magnitude change experiment.

The findings challenge the notion that the FOF solely serves a 'post-decision' role in short-term motor planning. The observed value encoding and the stimulus-dependent nature of FOF inactivation effects support the hypothesis that the FOF plays a broader role in representing action values during decision-making under risk. The consistent effects of FOF inactivation suggest it's a critical node in a network for representing the expected utility of options. The short-lived and bias-related role of the PPC suggests a more limited contribution to the encoding of decision variables in the risky choice task, potentially related to the sensory aspect of the stimuli or the maintenance of priors. The study highlights the distinction between FOF and PPC functions in economic decision-making, and suggests that these regions may participate in distinct processes.

This study provides compelling evidence that the FOF plays a crucial role in representing action values for economic decisions under risk in rats. The results demonstrate that the FOF dynamically encodes the value of lottery options and is critical for the integration of probability and value in decision-making. Further research could investigate the interactions between the FOF and other brain regions involved in reward processing and decision-making, and the role of these areas in other forms of decision-making (unexpected uncertainty) and different modalities of stimuli.

One limitation is the potential for muscimol spillover into adjacent brain areas. This could have influenced the observed effects of FOF and PPC inactivations. Another limitation is the potential for bias in the muscimol experiments, due to an uneven counterbalancing of lottery presentation on the left or right. The short-lived effect observed with PPC inactivation could indicate the need for longer duration inactivations or alternative approaches. Lastly, while the study utilized a rigorous model-based approach to behavioral analysis, individual differences in behavior and responsiveness to manipulations may still exist.