Guest Talk: Computational modeling tutorial by Sean Devine (McGill University)
Date: 29-02-2024
We hosted a guest talk by Sean Devine, a PhD student visiting from Ross Otto's lab at McGill University, on “The Homestretch: Cognitive Effort Exertion in Proximity to a Goal.”
Abstract: One of the longest-standing theories in psychology is the law of least work, which posits that organisms generally avoid effortful action. However, evidence suggests that effort exertion increases near a goal, even when unrewarded. Despite this idea dating back to the 1930s, its role in cognitive task performance remains underexplored. In this talk, Sean Devine presented six experiments demonstrating how individuals adjust their cognitive effort in response to progress information, reward proximity, and aversive stimuli. His findings provide new insights into how goal gradients influence cognitive effort investment in humans.
Reinforcement Learning (RL) Workshop
Sean Devine led a hands-on RL workshop, introducing participants to the fundamentals of Reinforcement Learning (RL) modeling in cognitive science. The workshop covered:
Basic RL algorithms for learning and decision-making
Maximum likelihood methods for estimating RL parameters
Best practices for parameter estimation (likelihood space exploration, parameter recovery, and predictive checking)
Code examples in Python (no prior experience required)
Resources from the workshop are available on **Sean Devine’s GitHub.
Hierarchical Drift Diffusion Modeling (HDDM) Tutorial
Sean Devine also introduced participants to Hierarchical Drift Diffusion Modeling (HDDM) using the popular HDDM Python package. Topics included:
- What a drift diffusion model is and its applications in cognitive science
- Why HDDM is useful for understanding decision-making and response times
- How to fit variants of HDDM to empirical data
- An overview of the new HSSM package for advanced modeling
![ref1]No prior knowledge of Python was required for this tutorial, making it accessible to a broad audience. All tutorial materials are available on Sean Devine’s GitHub.