Reinforcement Learning 'in the Wild': Modeling
Real-World Data

 We look forward to welcoming you at our workshop on 12th June 2025 at the RLDM conference, Dublin. 

RSVP and join our mailing list here: https://forms.gle/o7onA8soY8uCWA4P7

Reinforcement learning (RL) has been instrumental in formalizing and understanding cognitive processes such as learning and decision-making, elucidating their neural mechanisms in both humans and animals, and uncovering individual differences associated with psychopathology. However, translating findings from the lab to the real world is not trivial, as carefully controlled experiments cannot fully capture the complexity of real-world environments. Recent years have seen a rapid increase in the amount of behavioural data that is passively collected, through smartphones and other digital records. Applying RL models directly to this real-world behaviour offers a unique opportunity to extend laboratory findings to real-world contexts and to develop better RL agents for these environments. 

This workshop will bring together experts from psychology, computer science, neuroscience, and psychiatry to discuss open questions in this emerging field, including what types of real-world data are amenable to RL modelling, what adjustments are required to adapt existing frameworks to the complexity and variability of real-world data, and what opportunities exist for advancing AI agents and mental health research.

The workshop will consist of invited talks, a panel discussion and debate, and a brainstorming activity to develop new project ideas. Together, we hope these sessions will stimulate knowledge sharing, encourage improved practices, and seed future collaborations. A record of the workshop will be compiled into a public resource to ensure accessibility of the ideas for those who could not attend. Through this workshop, we aim to formally establish the field of “real-world RL” and encourage researchers to explore its potential advantages and limitations.