352-245-6169 mdthomasdc@gmail.com

Note: the following represents my attempts to educate myself regarding central pattern generators and are primarily quotes I have saved that have served that purpose. I am putting this onto my blog because I am working with several other people to understand

how we regulate our posture and our movement and by making it available I can expedite a way to view our complex neurological function and provide language that is shared by the scientific community as a whole. Quotes are rendered in italics. My words are in normal font.  

Central Pattern Generators 

Central pattern generators (CPGs) are neural networks that can produce rhythmic patterned outputs without rhythmic sensory or central input. CPGs underlie the production of most rhythmic motor patterns and have been extensively studied as models of neural network function.

My understanding of automaticity was that only a few cells had this propensity, i.e.: the sinoatrial node in the right atrium, the thalamus, and certain celss in the gut. It seems that automaticity is much more generalized than I had known.

A central goal of neuroscience is to understand how nervous systems produce movement. The simplest movements are reflexes (knee jerk, pupil dilation), which are involuntary, stereotyped and graded responses to sensory input, and have no threshold except that the stimulus must be great enough to activate the relevant sensory input pathway. Fixed action patterns (sneezing, orgasm) are involuntary and stereotyped , but typically have a stimulus threshold that must be reached before they are triggered, and are less graded and more complex than reflexes. Rhythmic motor patterns (walking, scratching, breathing) are stereotyped and complex, but are subject to continuous voluntary control. Directed movements (reaching) are voluntary and complex, but are generally neither stereotyped nor repetitive.

Rhythmic motor patterns comprise a large part of behavior. They are also complex (unlike reflexes) yet stereotyped (unlike directed movements) and, by definition, repetitive (unlike fixed patterns). As a consequence of this combination of behavioral importance and experimental advantage, rhythmic motor pattern generation has been studied extensively. This work has shown that the basic rhythmicity and patterning of rhythmic motor patterns are produced by neural networks termed central pattern generators.

Central pattern generators(CPGs) are neural networks that can endogenously (i.e. without rhythmic sensory or central input) produce rhythmic patterned outputs;¦p.1

All rhythms require: (1) two or more processes that interact such that each process sequentially increases and decreases, and (2) that, as a result of this interaction, the system repeatedly returns to its starting condition. Consider a pendulum at the moment of release. At this time, the pendulum bobs gravitational potential energy is at its maximum and the bobs momentum is zero. If the bob were not attached to the pendulum arm, when the bob was released it would simply fall: its potential energy would decrease monotonically, and its momentum would increase monotonically, until it hit the ground. However, the pendulum arm constrains the bob to move in a semicircular path. This constraint links the bobs potential energy and momentum such that, as the bob descends, its momentum increases and its potential energy decreases until, at the bottom of the swing (when the pendulum arm is vertical), these two qualities reach their respective maximum and minimum. At this time the bobs momentum forces the bob to ascend the other arm of its trajectory, and as its momentum decreases its potential energy increases until again (at the other highest point of its swing) its momentum is zero and its potential energy is maximum. The bob then descends, momentum and potential energy coordinately increase and decrease (on the downswing), and decrease and increase (on the upswing), until the bob reaches its original starting point, at which point the cycle repeats. P.2

Key to understanding rhythm generation in this (and many other network-based CPGs) is the concept of a half-centre oscillator. A half-centre oscillator consists of two neurons that individually have no rhythmogenic ability, but which produce rhythmic outputs when reciprocally coupled. P.3.

Limb and segment movements are coordinated during multilimb-segment behaviors such as terrestrial locomotion and undulatory swimming. Experiments in which central nervous systems are progressively isolated suggest that, although sensory feedback and the mechanical properties of the musculoskeletal system contribute to this coordination, it also arises in part from central mechanisms. Experiments in limbed vertebrates have shown that individual limbs can produce stepping movements, and experiments in lamprey and leech have shown that a few or even individual segments can produce a basic swimming motor pattern. These data have been interpreted as evidence that each limb, and each or at most a few segments, have their own CPG (a unit oscillator), and that central coordination among limbs and segments arises from coordinating connections among these oscillators.

It is important to note that in some systems (e.g. the leech) the strength of the interoscillator coupling is as great as that inside the unit oscillators. The effects of strong coupling on unit oscillator activity, and even whether the unit oscillators would maintain identifiably separate identities when so coupled, is less understood. P.6.

As animals mature, there are changes in the rhythmic motor patterns they express. For instance, tadpoles swim, but frogs hop; chicks hatch, but then walk; humans crawl, then walk, then run. Furthermore, humans can easily learn novel rhythmic motor patterns (e.g. swimming strokes, dances) that, once learned, seem as ˜automatic and ingrained as do clearly CPG-driven motor patterns such as walking.

How are new motor patterns acquired, and what happens to the developmentally discarded motor patterns? All available evidence in vertebrates suggests that: (1) CPG development does not require movement-induced sensory feedback, or even muscle innervation; (2) later rhythmic motor patterns arise by modification of the CPGs that generated earlier patterns; and (3) the ability to produce motor patterns that are expressed at only one developmental stage (hatching) is not lost as the animal matures, but can be re-induced by applying the proper sensory input at more mature stages (Sillar, 1996). These data suggest that fundamental CPG properties are innately established, and the acquisition of new rhythmic motor patterns by the CPG become increasingly multifunctional, presumably as a result of increasing CPG synaptic and cellular complexity, and additional extra-CPG descending input. P.10.

As a result of the work since the 1960s, we have considerable understanding of how CPGs produce their neural outputs, and some understanding of how CPGs interact, how sensory feedback contributes to their activity, and how modulation increases CPG flexibility. Work continues apace to understand more fully the roles played by these aspects of motor pattern generation. Behavior, however, consists not of spike trains, but of movement patterns that are continually adapted to match environmental variation. The role of effector characteristics (e.g. muscle dynamics, limb inertia, joint resistance) in the transformation of neural inputs into motor outputs, and how corrective sensory input is integrated into CPG activity, are relatively little understood. Therefore, a complete understanding of how CPGs generate behavior will not be achieved until all of these issues are resolved. P.11.

Hooper SL. Central Pattern Generators. Embryonic ELS 1999. Macmillan Publishers Ltd. Basingstroke, Hampshire, England. (accepted for publication February 2000). http://www.els.net/elsonline/html/A0000032.html [all quotes above are found here]

My suspicion is that just as there appear to be small collections of cells “ so called central pattern generators “ that generate repetitive behavior, so also there will turn out to be small collections of cells that generate intrinsically random behavior.

Wolfram, Stephen. A New Kind of Science. Wolfram Media, Inc. Champaign, IL 2002 p. 1011.

Wolfram is a very smart guy. He may be on to something.

It has been known for over a century that there is a variation of 3-4% in the stride interval of humans during walking.[5], and only in the last decade did Hausdorff et al. [1] demonstrate that the stride-interval time series exhibits long-time correlation, suggesting that the phenomenon of walking is a self-similar, fractal activity. Subsequent studies by West and Griffin [3,4] supported the conclusion that the human gait time series is fractal. However, more recently it was determined that these time series, rather than being monofractal, are weakly multifractal [6,7].

Human locomotion is known to be a voluntary process, but it is also regulated through a network of neurons called a Central Pattern Generator (CPG) [8], capable of producing a syncopated output.

The model that we present here, the super CPG (SCPG), assumes that the central nervous system is coupled to the motocontrol system, and together they control the locomotion of the gait cycle.

West BJ, Scafetta N. A nonlinear dynamical model of human gait. arXiv:cond-mat/0212203 v1 9 Dec 2002.

In order to generate and control complex behaviors, the brain needs not to explicitly solve a system of coupled equations. A more plausible mechanism is the construction of a vocabulary of fundamental patterns, or primitives that are combined sequentially and in parallel for producing a broad repertoire of coordinated actions. This concept appears to be at the basis of neural control of movement p.640.

Nature achieves something akin to programming through the biological mechanisms of synaptic plasticity [6]-[9], that is by the variation in efficacy of neural transmission brought about by past history of pre- and postsynaptic signals. P.641

Adaptation “ the ability to carry previously learned motor skills into new mechanical contexts “ is the most distinctive feature of biological motor systems and is widely investigated by neurobiologists [10]-[14].

The geometrical nature of operations that the brain must carry out in the recognition of objects and in the execution of movements is a central issue in neural information processing. In particular, some critical operations in the generation and control of movements can be formulated as coordinate transformations. Sensory information about the state of motion of the body comes from a variety of signal sources, each being concerned with a particular mechanical variable. For example, skeletal muscles are endowed with fusiform sensors that measure the muscle strain and its rate of change [15]. Muscles are also endowed, at the junction with tendons, with Golgi tendon organs [16] that sense variations in muscle force. Other receptors are sensitive to the displacement of the joints, to pain, to temperature, etc. Signals from other sensory organs, like the eyes and the vestibular organs, provide information about the position of the body and of its parts with respect to the environment. This variety of sensory channels is matched by a relatively uniform motor structure. The neural signals controlling muscle contractions are generated by the motoneurons, which are located inside the grey matter of the spinal cord. Muscles are partitioned into groups of fibers-called motor units-that receive common innervation from a single motoneuron. The force generated by a muscle is graded by a distribution of neural activities over the motoneurons. Each motor unit generates tension. As motor units are connected both in series and in parallel within a muscle, either tensions (parallel) or strains (series) are combined additively.

While there are several possible coordinate systems to describe different sensory and motor signals, these coordinate systems fall quite naturally into three classes: actuator coordinates, generalized coordinates, and endpoint coordinates [17]. P.641

In a series of experiments [42]-[46], the activity induced by chemical and electrical stimulation of the spinal interneurons of the frog was found to spread to several groups of motoneurons. This distribution of activity was not random but imposed a specific balance of muscle contractions. The mechanical outcome of the evoked synergistic contraction of multiple muscles was captured by a force field (Fig 30. The activation of a group of muscles generated a force that was recorded by a sensor at the endpoint of the limb. This force vector changed in amplitude and direction according to the position of the limb. The resulting force field converged toward a location in the reachable space of the limb- a stable equilibrium point. At this location, the force vanished and a small displacement of the endpoint in any direction induced a restoring force. The analysis of the force field induced by stimulation of the spinal interneurons revealed that such activation leads to the generation of a stable posture. P.645

The control of natural movements, such as underwater maneuvering and manipulation, involves the ability to solve complex mathematical problems. Our current knowledge of biological systems suggests that the living organisms approach this task not by solving explicitly large systems of differential equations but by combining building blocks of primitives that implement elementary motor behaviors. Some of these primitives are embedded in the genetic design of the nervous system. Others arise in each individual from the experience of mechanical interactions with the environment and from a variety of learning mechanisms. The peripheral structures of the vertebrates nervous systems, coupled with the muscle viscoelastic properties, leads to the definition of a family of mechanical waves, or time-dependent force fields. The nervous system modulates these waves and combines them to generate a rich repertoire of adaptable behaviors. P.648.

Mussa-Ivaldi FA, Solla SA. Neural Primitives for Motion Control. IEEE Journal of Oceanic Engineering 29(3):640-650. July 2004.

How do all the discrete quanta from which we are formed all work together, often seamlessly, as an organism nested within its environment? This concept of central pattern generators begins to paint a picture of how the nervous system can create homeostasis and rhythm seems to be a major component. I suspect that information is not just being tramsmitted synaptically through the nervous system and humorally through the immune system but also through the electromagnetic field that is formed by the physiological activity of our living bodies. A field effect begins to explain some seemingly anomalous functions such as sophisticated piano playing in which fingers are moving literally faster than they should be able to if the medium of control is all synaptic transmission. A single radio can receive the entire transmission that is being broadcast throughout the listening area. What is required is that the receiver can tune into the specific frequency. Mechanical tension within the organism may modulate this tuning. It has been observed by Ho (Rainbow and the Worm: the physics of organisms. World Scientific Publishing Co. Singapore 1998.) that the proteins in our bodies tend to line up with the long axis (vertical axis) of the body and that our cells resemble solid state electronics in their function. This opens the door to much speculation including a piezoelectric effect, the possibility of superconductivity at room temperature, and the visioning of organisms as liquid crystal in nature.