Valence-partitioned learning signals drive choice behavior and phenomenal subjective experience in humans
L. Paul Sands, Angela Jiang, Rachel E. Jones, Jonathan D. Trattner, Kenneth T. Kishida
bioRxiv Preprint Server March 17, 2023 preprint DOI: 10.1101/2023.03.17.533213 via bioRxiv
Summary
A hypothesized neurocomputational model, valence-partitioned reinforcement learning (VPRL), maintains separate tracks for appetitive and aversive information, generating independent reward and punishment learning signals. This model predicts changes in human choice behavior, subjective experience, and brain activity in regions including the ventral striatum and ventromedial prefrontal cortex during introspection. The findings suggest that valence-partitioned reinforcement learning provides a neurocomputational basis for investigating mechanisms that may drive conscious experience.
Study at a glance
| Characteristics | Theoretical or philosophical paper |
|---|---|
| Citations | 4 |
| Key finding | A valence-partitioned reinforcement learning model, which maintains separate appetitive and aversive information tracks, predicts dynamic changes in human choice behavior, subjective experience, and BOLD-imaging responses in a network of regions converging on the ventral striatum and ventromedial prefrontal cortex. |
Abstract
How the human brain generates conscious phenomenal experience is a fundamental problem. In particular, it is unknown how variable and dynamic changes in subjective affect are driven by interactions with objective phenomena. We hypothesize a neurocomputational mechanism that generates valence-specific learning signals associated with ‘what it is like’ to be rewarded or punished. Our hypothesized model maintains a partition between appetitive and aversive information while generating independent and parallel reward and punishment learning signals. This valence-partitioned reinforcement learning (VPRL) model and its associated learning signals are shown to predict dynamic changes in 1) human choice behavior, 2) phenomenal subjective experience, and 3) BOLD-imaging responses that implicate a network of regions that process appetitive and aversive information that converge on the ventral striatum and ventromedial prefrontal cortex during moments of introspection. Our results demonstrate the utility of valence-partitioned reinforcement learning as a neurocomputational basis for investigating mechanisms that may drive conscious experience.