Data and code: "Priors for perceived speed combine supramodally for visual-tactile motion but not for audio-visual motion"

Data and analysis code accompanying the manuscript "Perceptual speed priors are selectively shared across sensory modalities" (Tonelli, Phan & Alais). Perception combines incoming sensory evidence with priors learned from the statistics of the environment, producing central tendency — the regression of estimates toward the mean of the stimulus range. This work used central tendency as a behavioral probe of whether such priors are shared across the senses or kept specific to each modality. In three experiments, participants reproduced the speed of moving stimuli, with each modality assigned its own range of speeds, in unimodal blocks and in a block where the two modalities were randomly interleaved. A shared prior predicts that interleaving makes estimates converge on a common mean, whereas modality-specific priors predict that each modality retains its own bias. The deposit contains trial-level and per-participant data for the three experiments — audio–vision (n = 20), audio–vision with reduced visual precision (n = 15), and touch–vision (n = 16) — together with the R script reproducing the full statistical pipeline (linear mixed models, planned contrasts, effect sizes, Bayes factors, and the convergence index) and the MATLAB script generating the main figures. All files and instructions for running the analyses are described in README.md.