Table 1.

Quantification of functional/anatomical overlap

	MI between anatomical (unipolar/multipolar) and functional mapping		MI significance thresholds
	Neuropil	Soma	Neuropil	Soma
Ipsi-/contraversive (Fig. 4F)	0.28 (*)	0.30 (*)	0.17	0.20
Rotation/backward (Fig. 4-1A)	−0.36 (n. s.)	0.21 (*)	0.20	0.17
Forward/backward (Fig. 4-1B)	0.05 (n. s.)	0.04 (n. s.)	0.09	0.04

Functional mapping based on responses to ipsi- VS contraversive motion, but not on rotational VS backward and forward VS backward motion, significantly matches the morpho-anatomical mapping.

 Matching index indicates how well a given functional mapping overlaps with morpho-anatomical mapping of unipolar/multipolar neurons, independently for the IO neuron’s soma and their projectiles in the cerebellum (see Materials and Methods for details). Its absolute value can range from 0 (no overlap) to 1 (perfect overlap), and its sign indicates which pairs are matched. For example, the negative MI for Rotation / backward mapping in the neuropil area indicates that, in contrast to our expectation, the rotation-selective voxels in the cerebellum overlap better with the projections of multipolar neurons. To calculate the statistical significance of the MIs, we formulated a null hypothesis that the measured MI values are observed by chance. If so, random shuffling of the data should not change the MI in a consistent way. We therefore generated the corresponding null distributions of shuffled data and computed their 99.17^th percentiles as MI significance thresholds (one-tailed alternative, 5% significance level, Bonferroni-corrected for six comparisons: 100—5/6 = 99.17). If an actual MI is higher than the threshold, it is less than 5% that such high MI is observed by mere chance (indicated by asterisks). Only the ipsi- VS contraversive functional mapping significantly matched the anatomical mapping in both the IO and the cerebellum.