Fig. 7

a Probability distribution of differences in time constants of sensory ROIs (end of adaptation?beginning of the adaptation). Histograms were smoothed using a kernel function with width of 0.1 s. Dotted lines denote mean probability values, shaded areas denote SEM across larvae (normal-reafference control, N = 12, lag-trained non-adapting, N = 13, and lag-trained adapting, N = 6). b Anatomical location of sensory ROIs that decreased their time constants by more than 0.4 s (see Supplementary Fig. ² for anatomical reference). c Schematic diagram of the feedback controller that can implement acute reaction to unexpected perturbations in reafference. Cerebellar internal model monitors the efference copies of motor commands and resulting sensory consequences and learns their transfer function. It calibrates some intrinsic parameters of the controller according to consistent environmental features that can be learned by an internal model. The wavy line denotes the teaching signal used by the internal model to learn the transfer function. The dotted orange arrows denote potential influence of the cerebellum over the feedback controller (modification of time constants of sensory integration, as suggested by a, and/or parameters of the pre-motor circuits). d Mapping of the crucial functional nodes involved in acute reaction and long-term adaptation onto the larval zebrafish brain. See details in the text.