The dynamics of neuronal population signalling during the temporal flow of perceptual events.

When we walk along a busy street against the flow of people, looking for someone we hope to meet, we face a flood of visual inputs. In this situation, the brain mechanisms underlying visual processing are engaged continuously and for an unpredictable length of time. They must analyse incoming sensory information continuously to evaluate, initiate and guide motor actions at all times (walking, avoiding obstacles, scanning faces, etc). In contrast, most of our knowledge of the neuronal basis of visual processing is based on simple ‘laboratory’ situations: discrete trials with predictable start (cue), a fixed stimulus, end (another cue) and motor action (one of a few known alternative responses). One of the next major challenges for systems neuroscience will be to incorporate in our experimental paradigms some aspects of ‘normal vision’ such as the continuous integration of information over time and the ongoing evaluation for motor actions. My current proposal builds onto the well-defined experimental framework of perceptual decision-making, but rather than treating perception and behaviour as a sequence of discrete, finite episodes, each culminating in a decision, new experimental paradigms will probe how the brain engages in active, continuous monitoring of the dynamically changing flow of information. Previous work by myself and others has shown that neurons in extrastriate visual area V5/MT of primates can control 3D and motion components of a complex perceptual experience. Undertaking high-dimensional recordings from many neurons simultaneously in this well-described area of the visual system of awake behaving primates, I propose to investigate the broader questions of how neurons interact dynamically in space and time in order to shape visual perception and decision-making. This project has four parts. Firstly, in order to probe the role of cooperativity in neuronal circuits for visual perception, I will introduce unpredictable dynamic changes in visual stimuli and investigate the temporal relationship between these stimulus changes and percept-related neuronal activity and interactions. Do dynamical responses provide evidence for hysteresis in state-dependent neuronal interactions? Secondly, as a visual 3D-motion percept emerges, we will track the interactions between task-relevant neurons across functional subdomains like columns in real time. As a bistable stimulus is viewed over time (seconds), we will investigate the relationship between changes in neuronal interactions and the reported percept. Thirdly, we will test whether neuronal response patterns obtained with simple motion and 3D stimuli predict responses to more complex visual stimuli (such as biological motion and 3D motion patterns embedded in movie sequences). Lastly, we will employ the empirical data obtained from these high-dimensional recordings to challenge neuro-computational models of network dynamics for perceptual decisions and collaborate on their construction.