Motion after effect is perception-based, not only signal-based

International audienceWe consider motion after effect (MAE) in the context of multistable visual dynamic stimuli, more specifically whether MAE is uniquely driven by the signal (visual stimulus), or whether MAE can also be induced by the percept (visual interpretation of the stimulus). We use a plaid composed of two gratings as a dynamic visual stimulus. We modulate the transparency of the gratings to create an ambiguous plaid (1 signal with 3 possibles percepts: lines that move upward - called ‘coherent’-, the right or the left), and three unambiguous plaids (1 signal with 1 possible percept - coherent, right or left - for each plaid). 22 subjects are presented in each trial with either ambiguous plaids (12 trials) and non-ambiguous plaids (12 trials, four of each type). A trial consisted of two phases: the plaid is dynamic during the first 20 seconds of the trials (induction) and held fixed during the remainder (test). The subject is asked to report the perceived movement of the stimulus (keyboard, continuous reporting) during the whole trial (induction and test). We investigate whether the illusion of movement during the test is opposite to the perceptual history obtained during the induction. To do this, we model the perceptual sequence as a Markov Renewal Process (MRP) with percepts as states. In the non-ambiguous trials, the signal induces the desired MAE. An effect based on percept is measured in the ambiguous trials. This shows that MAE can be induced by percept and are not limited to mere signal adaptation. [JL's thesis funded by Univ. Grenoble Alpes, IDEX IRS (Initiatives de Recherches Stratégiques), Grenoble, France