Thesis defense Jeroen Atsma (Donders series 274)

8 May 2017

Promotor: prof. dr. P. Medendorp, copromotor: dr. F. Maij

Dynamic sensorimotor integration: implications for movement cancellation and visual stability

Every saccadic eye movement that we make rapidly shifts the image of the world on our retina. Yet, we generally perceive our visual world to be stable. How is this possible? The aim of this thesis is to obtain a better understanding of the processes and mechanisms involved in visual stability.

It has been suggested that the brain anticipates the retinal image shifts by remapping the neural image. Neurophysiological evidence suggests that this remapping occurs before saccade initiation. Behaviorally, a signature of the remapping process is seen in the location errors of visual stimuli that are presented at the moment of saccade execution. Would these errors also occur when a planned saccade is withheld just before its execution? Findings in this thesis indicate that visual stability is not compromised with planned, but abruptly cancelled, saccades. This means that saccade execution is a prerequisite for remapping, which rejects various models and constrains neural substrates.

This conclusion was thus based on planned saccades that never happened. For the analysis, we had to roughly estimate when the cancelled saccades would have happened, on the basis of previous performance. While this worked out well, in a subsequent study we explored a technique that may offer a way to determine exactly when movement cancellation is occurring and what the underlying reaction time would have been. We performed an experiment using whole-arm reaching movements and recorded (intra)muscular activity using electromyography (EMG). A reason why we studied arm muscles instead of eye muscles is that actions develop slower because of the arm's inertia. Also, because of the inertia, an active braking pulse is required to halt a developing movement, which could be picked up with EMG. Indeed, we found traces in the EMG signal that revealed the moment of movement initiation and cancelation, even at instances when no overt movement was made. Having established this, future experiments could be devised that exploit this neuromuscular marker to understand perception–action relationships in health and disease.

In the final study we return to the question of visual stability. When we look around we make 'snapshots' of the visual world. Somehow these snapshots are integrated into one coherent representation. It is known that during saccades people often fail to detect whether visual objects remain stable or move. How are the pre-saccadic and post-saccadic images combined in the computations for visual stability? Findings in this thesis indicate that the brain exploits a Bayesian inference process with two causal structures: the object remained stationary or the object moved. Interestingly, instead of choosing one structure, each combined representation reflected a weighted average of both. This demonstrates that the brain actually follows a statistically optimal strategy: rely on each structure (object stationary or object moved) as much as there is evidence for that structure. This provides insight in how the brain deals with uncertainty regarding visual stability.

In a concluding chapter, the concept of 'a point of no return' in movement cancelation, a potential function of early neural remapping, and the generalizability of the Bayesian inference model are discussed, together with suggestions for future research.