Studying habit learning using a computerized decision making task
Anne Collins, Professor
Psychology
Closed. This professor is continuing with Spring 2024 apprentices on this project; no new apprentices needed for Fall 2024.
Habits are inflexible patterns of behavior that have become deeply ingrained through repetition. When it comes to situations we are likely to encounter often, habitual control can be effective and efficient. When passing by the Campanile en route to class, you might turn left automatically, and eventually arrive at your destination without much cognitive effort. Yet when goals or environmental structure change, habitual control can be inflexible. When walking to the library instead of class during dead week, you might habitually turn left at the Campanile, when the quickest route to the library would take a right turn. In addition to driving such inconvenient slips of action, habits are thought to play an important role in many mental health conditions such as OCD, eating disorders, and addiction.
Despite habits’ centrality to everyday and disordered decision-making, researchers have had difficulty revealing habits as a function of repetition using well-controlled lab experiments in humans. This is thought to be due to the ease with which humans can override their habitual impulses using goal-directed control.
The Computational Cognitive Neuroscience (CCN) Lab is testing whether habitual behavior emerges more readily in humans when multiple decision-making steps are involved. The idea is that humans make high-level, abstract choices (e.g. get to class), then disengage control and execute the more concrete steps of those decisions (e.g. turn left at the Campanile) in a habitual way.
We have developed an online RPP task that seems to successfully elicit habitual behavior using a multi-step design. For this URAP project, we are looking to advance this work in one or more of the following possible directions:
- Collect an in-person dataset and replicate analyses
- Develop and collect follow-up version(s) of the task to test what aspects of the original task allow habitual responding to emerge
- Enhance the task (add animations, sound effects, etc.) to encourage participant engagement and reduce dropout
- Adapt existing computational models to fit to data from alternate versions of the task
Role: The first phase of the project will involve assigned readings from the habit literature, and additional readings depending on the focus of the project (e.g. reinforcement learning, computational modeling). The second phase of the project will involve task development; this will involve modifying existing task code (Javascript, HTML, CSS, PHP) for in-person data collection, implementing follow-up task designs, or adding features to boost task engagement. The third phase of the project will involve data collection and monitoring. The final phase of the project will involve analyzing collected data using R, Python, and/or MATLAB (e.g. replicating analyses, adapting/enhancing and fitting existing computational models). The exact project focus will be determined according to students’ interest and ability.
Qualifications: Applicants should have a strong interest in cognitive science and computational modeling. Some coding experience, or readiness to learn coding, is a plus. Students must be able to work independently.
Day-to-day supervisor for this project: Sarah Oh, Graduate Student
Hours: 6-8 hrs
Social Sciences Education, Cognition & Psychology