Motor patterns for different forms of walking: cues for the locomotor central pattern generator.

F the past decade, we have assessed the posture, hindlimb dynamics, and motor patterns for different forms of cat locomotion, including forward and backward walking 4 and slope walking. Two of our goals for undertaking these studies have been to identify the roles of muscle and inertial forces in the control of limb motion (reviewed in ref. 7) and to determine the mutability of motor patterns associated with various walking forms. With regard to the second goal, we believe that details of the motor patterns, typical of those shown in FIGURE 1, will provide clues about the general configuration of the central pattern generator (CPG) for locomotion. Grillner proposed a model for a multifunctional CPG located at lumbosacral levels that included control units for flexor and extensor muscles of the hindlimb joints. He hypothesized that each unit contained the neural elements required, except tonic facilitation by descending fibers, to generate locomotor-like bursts without sensory feedback or patterned input from supraspinal centers. One advantage of his model is that CPG units can be reconfigured to provide a variety of motor patterns. In FIGURE 2, we expand his model for forward, level walking and illustrate how the unit-burst CPG can be reconfigured to account for the motor patterns that characterize other forms of walking. The excitatory coupling among the four extensor units creates a robust extensor synergy that characterizes forward, backward, and upslope walking (FIG. 2). This synergy is disrupted during downslope walking as hip extensor activity is replaced by hip flexor activity during stance (FIG. 1C). To model this, we replaced the excitatory connection between the hip and knee extensor units with an inhibitory connection, and added a diagonal excitatory connection between the knee extensor and hip flexor units. By doing this, we created a “dual-phase” unit for the hip flexor muscles, one that generates both stanceand swing-related activity during the step cycle (FIG. 2). We modeled another dual-phase unit in the upslope CPG to account for the addition of stance-phase activity that knee flexor muscles exhibit during upslope walking (FIG. 1B, see ST). In the upslope model, we facilitated the knee flexor unit with excitatory connections from all adjacent units; this is consistent with data showing that the ST is active for most of the cycle. Following Grillner’s model, we inhibited the knee flexor unit with inhibitory connections from all adjacent units to account for the ST activity during level walking that is limited to a brief burst at the onset and the end of swing (FIG. 1A). During backward walking, however, ST activity coincides with that of other flexor muscles, and we modeled this by making the connections to the knee flexor unit similar to those of the hip and ankle flexor units.