Saccadic adaptation and reinforcement: one and the same? – AdRenf
Ocular saccades are rapid shifts in gaze direction that human use to perceive the fine details of their visual environment. The plasticity of these eye movements is traditionally studied with the “double-step paradigm”, in which a visual target steps during a saccade made towards this target. After several repetitions, the distance between the new target position and the saccade endpoint is progressively modified so that the eye lands closer and closer to the target new position. To explain this phenomenon, called “saccadic adaptation”, classical cognitive theories invoke internal mechanisms specialized in the minimization of visual error. Recent findings of C. Paeye (the project coordinator) showed, however, that reinforcement learning, via procedures involving auditory rewards in the absence of visual errors, can induce similar changes in saccadic amplitude.
Nevertheless, similarities in data do not prove that reinforcement and conventional double-step adaptations are one and the same process, i.e. that saccadic adaptation is driven by reinforcement. The first objective of this proposal is to lift this barrier by implementing more direct tests.
The first of our tasks aims to induce saccadic adaptation via reinforcement procedures and test whether well-established principles of conventional double-step adaptation also apply to reinforcement adaptation. More specifically we will investigate: 1) the transfer of adaptation to different saccadic vectors and 2) to various types of saccades, 3) the retention of adaptation, and 4) the necessity of feedback temporal contiguity. The last study of this task will investigate more directly the possible interplay between visual error minimization mechanisms and reinforcement processes. To do so, we will induce saccadic adaptation using, at the same time, double-step and reinforcement trials (interleaved trials) in various proportions, and evaluate their respective ability to drive saccadic adaptation.
Our second objective (Task 2) is to investigate and compare perceptual consequences of saccadic adaptation induced by reinforcement and conventional double-step paradigms. We will investigate the perception of 1) object location (only studied in relation to conventional double-step adaptation so far), 2) object size and 3) presentation duration, as well as, 4) pitch in the auditory modality. The latter are known to be related to space perception but have never been studied in relation to saccadic adaptation.
Once disseminated, the results of the present ANR project will be of high interest for researchers investigating sensorimotor learning in general, and saccadic adaptation in particular, since we are considering this phenomenon from a novel perspective – namely, the conditioning theory. In addition, by assessing the perceptual outcomes of the reinforcement of saccadic motor properties, this proposal will also draw the attention of scientists working within the reinforcement learning framework that most often refuses to make inferences about mental processes. Bringing two different theoretical approaches together - reinforcement learning theories and classical cognitive psychology - will provide innovative and meaningful insights into the understanding of plasticity of human sensorimotor system, thereby opening new lines of research and new perspectives of rehabilitation.
Madame Celine PAEYE (VISION ACTION COGNITION)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
UPDescartes - EA 7326 VISION ACTION COGNITION
Help of the ANR 194,268 euros
Beginning and duration of the scientific project: October 2018 - 36 Months