Human action selection under threat: computing adaptive behaviour – Dominik Bach (University College London)

November 5, 2021 @ 4:15 pm – 6:00 pm
Zoom meeting
Yasmin Fouani-Eckstein

Behaving appropriately under threat is key to survival. It also poses a range of interesting computational challenges unparalleled in other domains. A suggested solution to these demands is a computing architecture that relies on a multiplicity of tailored algorithms for specific threat scenarios. However, evidence for this suggestion is circumstantial in non-humans, and scarce in humans. In my talk, I will discuss the methodological problem of investigating threat avoidance in humans, and present pharmacological, clinical lesion, and neuroimaging data to support the validity of serious computer games as a way forward. Using different games that emulate a situation of approaching reward under threat (i.e. risky foraging), I triangulate a cognitive-computational algorithm for behavioral control in this scenario. This algorithm appears to be under instrumental and partly model-based control, with approximations of different granularity used in parallel. MEG data suggest some ideas about neural implementation.

Leave a Reply

Your email address will not be published. Required fields are marked *