This is an analysis of whether biomechanical or kinematic variables are controlled by descending reticulospinal commands to the spinal cord during escape responses (C-starts) in the goldfish. We studied how the animal contracted its trunk musculature to orient an escape trajectory. We used trunk EMG recordings as a measure of the reticulospinal output to the musculature and we simultaneously gathered high-speed cinematic records of the resulting movements. We found that the escape trajectory is controlled by (1) the relative size of the agonist versus the antagonist muscle contractions on two sides of the body and (2) the timing between these contractions. We found no separate signal for forward propulsion (or force) apart from the initial stage 1 bending of the body. Rather, the neural specification of force is embedded in the commands to bend the body. Thus, our findings demonstrate the importance of the angular kinematic components, or direction changes, caused by the descending reticulospinal command. This new direction change concept is important for two reasons. First, it unifies the diversity of C-start movement patterns into a single and rather simple quantitative model. Second, the model is analogous to the systematic EMG and kinematic changes observed by others to underlie single joint movements of limbs in other vertebrates such as primates. As in these cases, the fish capitalizes on the mechanical properties of the muscle by setting the extent and timing of agonist and antagonist contractions. This, plus the fact that sensory feedback is likely to be minimal, may enable the animal to reduce the number of computational steps in its motor commands used to produce the escape response. Because horizontal body movements in fish are a fundamental vertebrate movement pattern produced by a highly conserved brainstem movement system, our findings may have general implications for understanding the neural basis of rapid movements of diverse body parts.

As a model for learning how reticulospinal networks coordinate movement, we have analyzed the function of the Mauthner (M-) neurons in the escape response of the goldfish. We used water displacements of 3-6 micron to elicit C-start escape responses. These responses consist of 2 fundamental movements that grade into each other: Stage 1 lasts 15-40 msec and rotates the body 30 degrees-100 degrees about the center of mass; stage 2 is an axial acceleration that moves the center of mass 2-6 cm. Combined, the 2 stages result in trajectory turns ranging from 15 degrees to 135 degrees. Thus, these data show that M-initiated C-starts are not fixed movement patterns. The durations of stage 1 body muscle EMGs were correlated with turn angles achieved during stage 1. Since variable stage 1 EMGs are not seen when the M-cell is triggered by itself, other circuits, independent of the M-cell, must control the extent of the initial turn, and consequently escape trajectory. Furthermore, turning angles of stages 1 and 2 were correlated, allowing escape trajectory to be predicted, on average, 26 msec after movement started. This suggests that the commands for escape trajectory should be organized by the end of stage 1. In concert with this, the time of onset of the stage 2 EMG preceded the stage 2 onset by a range with a mean of 28.4 msec, typically putting the stage 2 command at the beginning of stage 1 movement. Thus, stage 2 initiation does not require motion-dependent feedback. Our findings indicate that the Mauthner cell initiates the first of a series of motor commands that establish the initial left-right decision of the escape sequence from the side of the stimulus, whereas parallel circuits simultaneously organize the command controlling the escape angle.