Where do our motivations come from? What is valuable and what are rewards? How do our bodies related to our motivations? These are the questions that have been at the center of intellectual inquiry throughout human history. Indeed, these are the issues that form the general basis for the proposed M2 internship that will attempt to ground them in neural mechanisms and mathematical principles.
More specifically, this internship is focused on further developing the theory of homeostatically regulated reinforcement learning as a framework for interoceptive learning of motivated behaviors (eLife 2014;3:e04811 doi: 10.7554/eLife.04811)
The theory links internal states of the organism with definitions of rewards and punishments and build a normative framework for individual rationality.
In the project will will explore the hypothesis that rewards come from internal needs that are then integrated into learning of motivated behaviors. Internal state regulation of rewards, in turn, leads to the variability in response and may be the basis for individual motivational characteristics.
The project can take several possible directions:
1. AI and machine learning direction - development of homeostatic reinforcement learning in continuous time and space for "embodied" agents.
2. Development of inverse homeostatic reinforcement learning for inter-individual interactions.
3. Extensions of the theory beyond primary physiological rewards (based on bodily needs) to higher level internal states, such as social, cognitive and financial.
4. Applications to pathologies such as addiction, asocial behaviors as manifestations of altered internal state dynamics.
5. Neural implementations of the homeostatic reinforcement learning, notably taking into account recent data on the involvement of the dopaminergic brain circuits in state dependent primary and social rewards, as well as the active representations of the internal state in the cortex.
This internship will take place at the Mathematics of Neural Circuits team in the LNC2 at ENS and as a part of the Group for Neural Theory at the DEC ENS.
Interactions and with experimental collaborators will possible as a part of the internship.
Candidates with mathematics, physics, engineering and computer science backgrounds are encouraged to apply.