Abstract
Oscillatory dynamics are widespread in animal behaviour. For example, fish periodically increase their swimming speed resulting in a burst-and-glide movement pattern. Following recent studies, we propose a potential functional interpretation to this behaviour by noting that a similar neuro-inspired oscillatory system has favourable decision-making properties. By coupling via non-linear feedbacks a slow variable describing responsiveness to inputs to a set of faster decision variables, a decision-making system can go through periodic cycles of "exploration" (gathering evidence) and "exploitation" (executing a decision), which emerge intrinsically from the system's dynamics. In this way, decisions are made constantly "on the go", and an individual flexibly and adaptively adjusts to changes in the decisions' landscape. Furthermore, the decision-making system exhibits favourable properties such as spontaneous decisions in the absence of evidence, symmetry-breaking, and is constantly brought back to a state of ultra-sensitivity to the options' qualities. Such properties can be co-opted to develop a bio-inspired control design for effectively guiding the spatial navigation of an artificial agent. In this project, I am expanding a ring attractor model by adding neuro-inspired oscillatory behaviour. The aim is to use this model to potentially describe burst-and-glide movement in fish, and efficiently guide the navigation of a robot through its environment.
About the speaker
I am a third-year PhD student in animal collective behaviour at Swansea University (Wales), currently working on primates. I am interested in how individual differences like social dominance rank effect collective movement and activity synchronization. During my research visit at CASCB, I am developing a model of burst and glide movement in fish with the aims of describing repeated decision-making and developing a bio-inspired control design for the navigation of an artificial agent.