The order of the variables was set by clustering them for endogenous locomotion data (A, lower left part). Only significant correlations are shown. Each matrix is divided into two halves representing the correlation in different groups as indicated below and on their side. Movements smaller than 0.8 mm were discarded in A and C, but not in B, Tm -“Thresholds for movement”. Positive correlations are represented by green dots, negative correlations by violet dots. The size and color of the dots represent the correlation coefficient, as indicated.

A. Correlation matrices of endogenous locomotion (lower left half-matrix) and computer generated correlated walk (upper right half-matrix).
B. Same as in A without discarding movements smaller than 0.8 mm
C. Correlation matrices of fly data in Buridan's paradigm, with narrow (lower left half) and wide (upper right half) stripes.

Highlights in A and B:
Small red squares: The number of pauses is not correlated with the total activity time in the real fly data.
Elongated red rectangles: The median speed is correlated with the total activity time (TT) in the fly data.
Large red squares: The duration of activity bouts (TT but not ST) correlates with the total activity time (ST and TT) in the fly data.

Highlights in A and C:
Small purple squares: The number of walks is not correlated with stripe deviation.
Small red squares in C: The angle deviation and stripe deviation are correlated only in the narrow stripes situation.
Elongated purple rectangle in C: The median speed correlates with the duration of activity periods in the narrow stripes situation. n = 20 flies for each group.