Fig. 4
From: A cerebellar mechanism for learning prior distributions of time intervals
Comparison of TRACE with human behavior in the RSG task. a Human behavior during the RSG task with three different prior distributions, Short (black), Medium (dark red) and Long (red). Data adapted from previous work 3. Note that bias toward the mean of the prior is larger for longer interval ranges. b The corresponding prediction from TRACE. Dots and circles mark individual responses and averages respectively. To directly compare TRACE and human behavior, scalar noise was added to the output of TRACE to simulate responses (Methods section). Parameters of the model were not tuned to the data and a single-linear scaling factor was used to calibrate model output for all conditions3
