Unlocking the information pathways of the brain using machine-based modelling (and a little bit of a magic)

Main BBSRC strategic theme: Understanding the rules of life
Secondary strategic BBSRC theme: Transformative technologies

The proposed doctoral research will combine the latest advancements in machine learning with methods from the comparative science of magic to model how potential networks of information processing and prediction in the brain might result in observable behaviours present in biological organisms.

Unlocking the secrets of the brain is often considered one of the grand scientific challenges of our time. One aspect of this challenge is uncovering how the brain processes the vast quantities of sensorimotor information endemic to embodiment to inform cognition and produce appropriate reactions to dynamic environmental conditions (Friston, 2010). Leading theories postulate that significant "shortcuts" are used to facilitate such processing, through which predictions relating to the changing environment are formed using prior experiences and updated through sensorimotor feedback (Friston 2010; Yon and Frith, 2021). It is believed that many neuropsychiatric and neurodevelopmental disorders may result from errors in the formation of these elegant pathways, such as OCD, schizophrenia and dyslexia (Harding et al., 2024; Sterzer et al., 2018; Sigurdardottir et al., 2017). However, such these pathways remain a cognitive "black box", limiting potential treatment options.

The comparative psychology and biology of magic has recently offered early insights into the role of embodied information in the perceptions and expectations surrounding physical actions performed by dexterous agents via the use of cognitive illusions as a powerful violation of expectation paradigm (Garcia-Pelegrin et al., 2021, 2023, 2024). These investigations have explored cross-species differences in prediction errors relating to manual action expectations in primate species with varying hand morphologies using sleight of hand magic tricks that rely on distinct biomechanical processes....