Dominik Bach: Critical intelligence: investigating human escape in virtual reality

Abstract

 All animals including humans have to cope with immediate threat to survive and reproduce. Many species employ complex and sophisticated defensive behaviors. Rapid decisions between these actions, without much leeway for cognitive or motor errors, poses a formidable computational problem that is substantially different from the challenges encountered in classical decision-making experiments. However, theories of human defensive behaviour are largely based on extrapolation across species, imagined, or third-person view scenarios.

 Here, I present the development of a virtual reality to investigate human escape, in which participants can run for shelter to evade various pre-historically relevant threats. Data from three experiments (N1 = 29, N2 = 30, N3 = 56) challenge a view that escape behaviour is instinctive or hard-wired. Instead, the underlying algorithm appears goal-directed and exhibits planning properties. Escape decisions are based on a detailed identification of threat identity and predicted trajectory, and are pre-planned and dynamically updated as the environment changes. In contrast, information-seeking behaviour might rely on simpler computations. Several aspects of behaviour are predicted by participants' stable personality traits.

 Finally, we characterise human escape in terms of a typical movement sequence that has no equivalent in cross-species models of defensive behaviour. Taken together, these findings provide novel insights into human escape, and highlight a need for suitable experimental paradigms to investigate the computational mechanisms underlying complex, real-life behaviour in humans.

About the speaker and his research

Dominik Bach

Host

Marlene Bartos

Whole program

NeuroForum