Central pattern generators in the spinal cord can create a complex motor program even without sensory feedback
A common feature of motor control is the motor program, a set of structured muscle commands that are determined by the nervous system before a movement begins and that can be sent to the muscles with the appropriate timing so that a sequence of movements occurs without any need for sensory feedback. The best evidence for the existence of motor programs is that the brain or spinal cord can command a variety of voluntary and automatic movements, such as walking and breathing (see pp. 706–709), even in the complete absence of sensory feedback from the periphery. The existence of motor programs certainly does not mean that sensory information is unimportant; on the contrary, motor behavior without sensory feedback is always different from that with normal feedback. The neural circuits responsible for various motor programs have been defined in a wide range of species. Although the details vary endlessly, certain broad principles emerge, even when vertebrates and invertebrates are compared. Here we focus on central pattern generators, well-studied circuits that underlie many of the rhythmic motor activities that are central to animal behavior.
Rhythmic behavior includes walking, running, swimming, breathing, chewing, certain eye movements, shivering, and even scratching. The central pattern generators driving each of these activities share certain basic properties. At their core is a set of cyclic, coordinated timing signals that are generated by a cluster of interconnected neurons. These basic signals are used to command as many as several hundred muscles, each precisely contracting or relaxing during a particular phase of the cycle; for example, with each walking step, the knee must first be flexed and then extended. Figure 16-6A shows how the extensor and flexor muscles of the left hind limb of a cat contract rhythmically—and out of phase with one another—while the animal walks. Rhythms must also be coordinated with other rhythms; for humans to walk, one leg must move forward while the other thrusts backward, then vice versa, and the arms must swing in time with the legs, but with the opposite phase. For four-footed animals, the rhythms are even more complicated and must be able to accommodate changes in gait (see Fig. 16-6B). For coordination to be achieved among the various limbs, sets of central pattern generators must be interconnected. The motor patterns must also have great flexibility so that they can be altered on a moment's notice—consider the adjustments necessary when one foot strikes an obstacle while walking or the changing motor patterns necessary to go from walking, to trotting, to running, to jumping. Finally, reliable methods must be available for regulating the speed of the patterns and for turning them on and off.
FIGURE 16-6 Rhythmic patterns during locomotion. A, The experimental tracings are electromyograms (EMGs)—extracellular recordings of the electrical activity of muscles—from the extensor and flexor muscles of the left hind limb of a walking cat. The pink bars indicate that the foot is lifted; the purple bars indicate that the foot is planted. B, The walk, trot, pace, and gallop not only represent different patterns and frequencies of planting and lifting for a single leg but also different patterns of coordination among the legs. LF, left front; LH, left hind; RF, right front; RH, right hind. (Data from Pearson K: The control of walking. Sci Am 2:72–86, 1976.)
The central pattern generators for some rhythmic functions, such as breathing, are in the brainstem (see p. 706). Surprisingly, those responsible for locomotion reside in the spinal cord itself. Even with the spinal cord transected so that the lumbar segments are isolated from all higher centers, cats on a treadmill can generate well-coordinated stepping movements. Furthermore, stimulation of sensory afferents or descending tracts can induce the spinal pattern generators in four-footed animals to switch rapidly from walking, to trotting, to galloping patterns by altering not only the frequency of motor commands but also their pattern and coordination. During walking and trotting and pacing, the hind legs alternate their movements, but during galloping, they both flex and extend simultaneously (compare the different leg patterns in Fig. 16-6B). Grillner and colleagues showed that each limb has at least one central pattern generator. If one leg is prevented from stepping, the other continues stepping normally. Under most circumstances, the various spinal pattern generators are coupled to one another, although the nature of the coupling must change to explain, for example, the switch from trotting to galloping patterns.
Pacemaker cells and synaptic interconnections both contribute to central pattern generation
How do neural circuits generate rhythmic patterns of activity? There is no single answer, and different circuits use different mechanisms. The simplest pattern generators are single neurons whose membrane characteristics endow them with pacemaker properties that are analogous to those of cardiac muscle cells (see p. 489) and smooth muscle cells (see p. 244). Even when experimentally isolated from other neurons, pacemaker neurons may be able to generate rhythmic activity by relying only on their intrinsic membrane conductances (see Fig. 12-4). It is easy to imagine how intrinsic pacemaker neurons might act as the primary rhythmic driving force for sets of motor neurons that in turn command cyclic behavior. Among vertebrates, however, pacemaker neurons may contribute to some central pattern generators, but they do not appear to be solely responsible for generating rhythms. Instead, pacemakers are embedded within interconnected circuits, and it is the combination of intrinsic pacemaker properties and synaptic interconnections that generates rhythms.
Neural circuits without pacemaker neurons can also generate rhythmic output. In 1911, T. Graham Brown proposed a pattern-generating circuit for locomotion. The essence of Brown's half-center model is a set of excitatory and inhibitory interneurons arranged to inhibit one another reciprocally (Fig. 16-7). The half-centers are the two halves of the circuit, each commanding one of a pair of antagonist muscles. For the circuit to work, a tonic (i.e., nonrhythmic) drive must be applied to the excitatory interneurons; this drive could come from axons originating outside the circuit (e.g., from neurons in the brain) or from the intrinsic excitability of the neurons themselves. Furthermore, some built-in mechanism must limit the duration of the inhibitory activity so that excitability can cyclically switch from one half-center to the other. Note that feedback from the muscles is not needed for the rhythms to proceed indefinitely. In fact, studies of >50 vertebrate and invertebrate motor circuits have confirmed that rhythm generation can continue in the absence of sensory information.
FIGURE 16-7 Half-center model for alternating rhythm generation in flexor and extensor motor neurons. Stimulating the upper excitatory interneuron has two effects. First, the stimulated excitatory interneuron excites the motor neuron to the flexor muscle. Second, the stimulated excitatory interneuron excites an inhibitory interneuron, which inhibits the lower pathway. Stimulating the lower excitatory interneuron has the opposite effects. Thus, when one motor neuron is active, the opposite one is inhibited.
Central pattern generators in the spinal cord take advantage of sensory feedback, interconnections among spinal segments, and interactions with brainstem control centers
The half-center model can produce rhythmic, alternating neural activity, but it is clearly too simplistic to account for most features of locomotor pattern generation. Analysis of vertebrate pattern generators is a daunting task, made difficult by the complexity of the circuits and the behaviors they control. In one of the most detailed investigations, Grillner and colleagues studied a simple model of vertebrate locomotion circuits: the spinal cord of the sea lamprey. Lampreys are among the simplest fish, and they swim with undulating motions of their body by using precisely coordinated waves of contractions of body muscles. At each spinal segment, muscle activity alternates—one side contracts as the other relaxes. As in mammals, the rhythmic pattern is generated within the spinal cord, and neurons in the brainstem control the initiation and speed of the patterns. The basic pattern-generating circuit for the lamprey spinal cord is repeated in each of the animal's 100 or so spinal segments.
The lamprey pattern-generating circuit improves on the half-center model in three ways. The first is sensory feedback. The lamprey has two kinds of stretch receptor neurons in the lateral margin of the spinal cord itself. These neurons sense stretching of the cord and body, which occurs as the animal bends during swimming. One type of stretch receptor excites the pattern generator interneurons on that same side and facilitates contraction, whereas the other type inhibits the pattern generator on the contralateral side and suppresses contraction. Because stretching occurs on the side of the cord that is currently relaxed, the effect of both stretch receptors is to terminate activity on the contracted side of the body and to initiate contraction on the relaxed side.
The second improvement of the lamprey circuit over the half-center model is the interconnection of spinal segments, which ensures the smooth progression of contractions down the length of the body, so that swimming can be efficient. Specifically, each segment must command its muscles to contract slightly later than the one anterior to it, with a lag of ~1% of a full activity cycle for normal forward swimming. Under some circumstances, the animal can also reverse the sequence of intersegment coordination to allow it to swim backward!
A third improvement over the half-center model is the reciprocal communication between the lamprey spinal pattern generators and control centers in the brainstem. Not only does the brainstem use numerous pathways and transmitters to modulate the generators, but the spinal generators also inform the brainstem of their activity.
The features outlined for swimming lampreys are relevant to walking cats and humans. All use spinal pattern generators to produce rhythms. All use sensory feedback to modulate locomotor rhythms (in mammals, feedback from muscle, joint, and cutaneous receptors is all-important). All coordinate the spinal pattern generators across segments, and all maintain reciprocal communication between spinal generators and brainstem control centers.