A manuscript has been submitted and a pre-print now exists for this project!
Figure: Coordination patterns (PCA) generated from one axis of turns (pitch, roll, or yaw) can be used to discriminate (LDA) turns in other directions but only if the precise spike timing (set by the parameter sigma) is on the scale of several milliseconds. If sigma is set to ~1s the classification tasks collapse to using approximate rate codes and decoding of behavior is reduced both within and across turning conditions. While the coordination motifs (PCA spaces) can decode behaviors across different axes. The specific patterns of spikes that discriminate the turns are different. This is determined by the lack of codirectionality of the LDA axes, which in most cases are not significantly above what would be expected by chance (red lines in D, H, and L)
We captured the comprehensive, spike-resolved motor program from 9 moths undergoing six, discrete turning behaviors. We elicited these six behavior states with wide-field drifting visual stimuli about the flight axes — pitch, roll, and yaw while simultaneously recording forces and torques in all six degrees of freedom from the body. We developed a simple linear spike train decoding pipeline based on principle components dimensionality reduction and linear discriminant analysis and showed that this straightforward approach is sufficient to predict behavior nearly perfectly, but only if within-wing stroke spike timing information is included. We also demonstrated that as few as half the muscles can be used to retain this near perfect decoding performance, linking coordination to redundancy in encoding across the entire moth flight motor program. We then used this comprehensive motor representation to test if muscle covariation present in one pair of visual stimulus conditions can be used to decode behavior in a different pair of visual stimulus conditions. We found conserved muscle coordination patterns at the level of motor unit spike timings in these functionally distinct behaviors. If we identified the coordination patterns underlying pitch turns alone, we could still successfully decode roll and yaw turns. So while the moth needs to use different specific patterns of spike timing to achieve the six turns, it does so by using the same patterns of motifs across the motor system.
