Mentor: Dr. Björn Herrmann (Co-Supervisor: Dr. Harrison Ritz)
Assistant Professor, Affiliated Scientist
Project Description
Our lives bombard us with countless tasks that we must navigate on the way to our goals. Previous research has explored how our brains switch between different tasks, finding that people reset to a “neutral” brain state between tasks. This same strategy emerged in artificial neural networks trained on a similar experiment, suggesting that this resetting process may be optimal in some way. A limitation of these experiments is that they explicitly told participants which task to perform on each trial. In everyday life, people often need to link together consecutive tasks (e.g., finding the Netflix watchlist before deciding what show to pick).
We will explore how ‘optimal’ task switching strategies change under different forms of instruction by training recurrent neural networks (RNNs) to perform cognitive psychology experiments. We will examine how RNN strategies differ when RNNs are explicitly told which task to perform (e.g., “task A” vs. “task B”) compared to when they are told the transition between tasks (e.g., “switch” vs. “repeat”). You will learn how to design cognitive experiments for neural networks, how to train RNNs using the PyTorch Python package, and how to interpret dynamic computations in trained RNNs. This research will generate predictions for subsequent neuroimaging experiments on how people switch between tasks. Understanding these neural computations will deepen our field’s knowledge of how cognitive flexibility may work in the brain, and how these processes may differ in people with ADHD or depression.
Mentorship Statement
This research project will be led by Harrison Ritz, a postdoc fellow at The Rotman Research Institute. Harrison is a big fan of summer internship programs, having previously worked as a mentor in diversity-promoting summer programs at multiple universities. He is a responsive scientific mentor, accommodating mentees’ unique interests while scaffolding their skills and responsibilities as their confidence grows. In addition to learning about experimental design, computational modelling, and data science, mentees will develop their ‘academic trade skills’ in project planning, time management, and scientific communication. Summer internships are a lot of fun, and are a great opportunity for folks to learn about the day-to-day experience of working in a research lab. We hope that you apply!