Colloquium: Uncovering the driving mechanisms of perceptual multi-stability: follow the eyes!

Dr. Meso is a Research Fellow at the Institut de Neurosciences de La Timone, a Laboratory of the CNRS in Marseille, France, which he joined in 2011. He is part of a small team of researchers who probe the mechanisms underlying visual function (specifically dynamics and probabilistic computations) empirically and theoretically in human and non-human primates. He combines an expertise in Psychophysics with eye movement recordings to ask questions about visual motion processing and links the mathematical and computational analysis of the data to the construction of theoretical models of neural processing. Dr. Meso initially studied Physics (BSc, Imperial College London) and Medical Physics (MSc, UCL) before shifting to Visual Neuroscience, combining psychophysics with modelling to study visual motion integration (PhD, Royal Holloway), where his key published contributions outlined the role of edges and other features in perceived speed estimates and transparency perception. Following his PhD, he joined a leading psychophysics lab at McGill University where he published work using psychophysical sensitivity to characterise unseen hierarchical cortical neural mechanisms mainly processing so called second order visual motion stimulation. He continues to study the role of cortical hierarchy in visual processing in his current role, with a largely multi-disciplinary approach in collaborations with Psychologists, Biologists, Engineers and Mathematicians in France and internationally. The interests and collaborations more recently look at the common attributes of dorsal and ventral stream hierarchical processing, thus extending beyond motion processing to symmetry and shape sensitivity.

Abstract:

Perceptual multi-stability occurs when a sensory input has multiple interpretations and the resolution of this ambiguity by the brain results in dynamic shifting between these alternatives. This phenomenon has largely been studied using bi-stable stimuli such as binocular rivalry, ambiguous figures and motion plaids, which alternate between two alternatives in the form ABABA. As such, switches have been proposed to be driven by neural noise and adaptation - served by mutual inhibition between the representations of the alternatives. In our work, we seek to further characterise the role of these driving mechanisms using a tri-stable barberpole motion stimulus in which the presence of three alternatives, ABC, means that at each perceptual transition, two choices are possible instead of only one, so we can study the changing probabilities of the different classes of transitions across parametric manipulations. In addition, the perceptual space of perceived direction in which this competition occurs is a continuous one which permits us to use smooth eye movements as a dynamic probe of instantaneous perceived direction. This probe in turn can be well modelled by the dynamical systems mathematical framework, which allows one to extract key signatures of the dynamics. We discuss how we use the patterns of participant responses and the fluctuations within the eye movements to dissociate the roles of noise and adaptation inferring with the help of modelling the relative contributions of each, which depend on the strength of the input.