Tracking the contribution of inductive bias to individualised internal models
Fig 4
Contrasting the ideal observer and CT performance in predicting trial-by-trial response times.
A, Performance of the two models in predicting response times on the eight days of exposure to the stimulus sequences governed by the same statistics. Performance is measured as the amount of variance in response times (R2) explained by the particular model. Dots represent mean performance, boxes represent the 25 and 75 percentile of the performances across the population of 25 participants. B, Violin plot of the distribution of mode l performances across the participants on the eighth day of exposure. Grey dots indicate individual participants, lines connect model performances for the same participant. All data on the figure are cross-validated by fitting the model on a set of blocks late in the session and tested on non-overlapping earlier blocks.