Abstract
This work aims to evaluate the induction of the motion after-effect (MAE) with a multi-stable stimulus ; and the effect of perceptual history during the induction phase on the induced illusion. MAE is a powerful illusion where motion is perceived on a static image. The classical procedure consists of the presentation of a moving field for 20—30 seconds (induction phase), followed by a freezing of this moving field. This procedure induces perception of a movement illusion with a direction opposed to the direction of the moving field. Multi-stable stimuli are stimuli whose properties remain stable across time, but generating perceptual fluctuations between a finite number of interpretations. We used a modified version of the multi-stable dynamic plaid to evaluate the induction of MAE and specifically the weight of signal versus perceptual history in the build-up of MAE. Method: The inducting stimulus is a superposition of tri-stable dynamic plaid and a random dot kinematogram (RDK). The plaid generates three perceptions that alternate and can be controlled through RDK parameters. The stimulus in the induction phase can be ambiguous (multi-stable) or non-ambiguous where desambiguation is done by augmenting the signal level in the RDK. After 20 seconds of induction the stimulus freezes. All along the trial, subjects are asked to hold one of the arrow buttons indicating the direction of perceived movement. Results: All subjects perceive motion both in the induction and freeze phase. Dynamics of perception during the induction phase are similar in ambiguous and non-ambiguous conditions by design. An opposed semi-Markov model predicts direction of perception during the freeze phase better than a continuity semi-Markov model which parameters are trained on the induction phase’s perceptual history. Conclusion: MAE can be induced by multi-stable stimuli. The perceptual history encoded in an opposed semi-Markov model can predict the direction of MAE.