More formally we might state this as: What are the evolutionary origins of reward value? What role does reward valuation play in our homeostasis? How are homeostatic states and reward values computed by the brain? And what function does this all ultimately serve?
The framework we work from makes one key assumption from which several other theoretical predictions can be made: The principal role of reward valuation is to shape behavior toward optimizing homeostatic states of the body, which in turn optimizes long-run survival. The work we do is thus partly theoretical and partly empirical.
We seek to derive theories of reward that are grounded on fundamental principles of homeostatic dynamics and their evolution. We draw on concepts, tools, and results, from a diversity of fields, including economics, ecology, physiology, and computational biology. And we use these to build computational models of reward and homeostasis.
From these models we derive falsifiable predictions for behavioral, physiological, and neural responses. In human subjects, we triangulate between computational modeling, fMRI, physiological monitoring, and economic behavior. And through ongoing collaborations with the Crick Institute, we are testing parallel predictions using opto- and pharmaco-genetics.
We are engaged in several ongoing theoretical projects. The first addresses how biological agents behaviorally optimise their own body temperature using reinforcement learning algorithms operating on homeostatic drive functions derived from physiological ecology data. The second formulates a general homeostatic theory of reward that conjectures how fundamental features of economic behavior can be unified into a single explanatory framework. The third attempts to formulate a generalized model of homeostatic dynamics, and to derive reward functions from first principles of fitness maximization.
Experimentally, we focus primarily on variables such as glucose, hydration, and temperature. We take the strategy of manipulating central homeostatic states, to probe how these states are sensed, and how such states modulate reward computation.
As such we are principally interested in homeostatic-reward interfaces in the brain, in particular the interface between the hypothalamus and the midbrain. Future work aims to test for reward-homeostasis dysfunctions in psychiatric and metabolic disorders.
In parallel we are pursuing several empirical projects that follow from this theoretical framework. First, we are currently acquiring functional imaging data of reward responses to the oral consumption of glucose and the modulation of these responses by blood glucose. Second, we are collaborating on parallel experiments in animal models that will test the same experimental protocol using calcium imaging, opto- and pharmaco-genetics in hypothalamic-midbrain networks. Third, following explicit predictions from homeostatic theory that contradict economic theory, we are testing risk preferences for costs, and their covariance with structural and connectivity brain data. Finally, we are testing reward responses to foods as a function of homeostatic state and metabolic interventions in gastric bypass patients. Finally with the physics group at DRCMR we are implementing and developing selective excitation imaging sequences to probe the hypothalamic-midbrain regions that implement the interface between homeostasis and reward.