project to beta-test HSSM & investigate history biases in decision-making: looking for student!

70 views

Skip to first unread message

Anne Urai

unread,

Oct 12, 2023, 9:01:34 AM10/12/23

to hddm-users

Modelling decision-making across species using likelihood-approximation networks

Project opening with PIs Anne Urai (Leiden, NL) and Alexander Fengler (Brown, USA)

Background

Human observers’ previous choices consistently bias their subsequent evidence accumulation (Urai et al., 2019). Using a publicly available dataset where hundreds of mice make perceptual decisions (International Brain Laboratory et al. 2021; int-brain-lab.github.io/iblenv), we recently replicated this pattern (Urai & IBL, 2023, 2023.ccneuro.org/proceedings/0000544.pdf).

This decision-making strategy, which robustly captures individual differences across tasks, thus seems to be exhibited across mammalian species - thereby providing insights into the shared neural circuit mechanisms of cognition.

Project description

The aim of the current project is two-fold.

First, we'd like to replicate the findings above with the state-of-the-art HSSM package (github.com/lnccbrown/HSSM) recently developed by Alexander Fengler. This package is designed for modern Bayesian statistical inference for cognitive process models. In this project, you will re-implement the previous code (github.com/anne-urai/mouse_history_ddm) in the new package, and beta-test and develop HSSM in the process.

The second goal is to further validate the existing model fits: testing whether extensions of the basic Drift Diffusion Model (e.g. collapsing bounds, suggesting increasing urgency in decision making over time) need to be added, and examine to what extent mouse decision-making matches the previously observed patterns in humans.

What we're looking for

- You are interested in cognitive modelling of behaviour

- You have some experience with Python coding (or at least experience with one other programming language, that you think will allow you to transfer to Python)

- You could be based anywhere and work remotely (of course, you are very welcome to join our labs in-person in Leiden, NL or Providence, USA).

- If you are a student at Leiden University or Brown University (e.g. cognitive neuroscience, AI) this project would be very well suited as a thesis or internship.

In short, this project is ideal for a (cognitive) computational neuroscientist looking for a hands-on coding project that will have an outcome with great relevance for cross-species comparisons in decision-making research.

What we offer

You will learn about Python coding, theoretical frameworks of evidence accumulation, the statistical machinery for Bayesian parameter estimation of such evidence accumulation models, and get hands-on experience working with the large decision-making dataset of the International Brain Laboratory.

Anne Urai is a cognitive neuroscientist at Leiden University, The Netherlands. Read more at anneurai.net and anne-urai.github.io/lab_wiki/Working_with_me.html.

Alexander Fengler is a Postdoctoral Research Associate at Brown University, USA. Read more at alexanderfengler.github.io