Contracting Arm Demonstration

	The following is another demonstration; this one shows the firing patterns
  of several different neuron types in the model.  There are six, interconnected 
  neural networks; one for each of the six simulated muscle fibers in the biceps.
  Every network recieves the same input signal, yet the model has trained itself 
  so that resulting firing patters of each network are not the same.  This reflects 
  a similar division of labor in biological muscles (e.g. between the different
  types of muscle fibers).

	Above is a model of an arm that is contracting to a fully flexed,
  or partly flexed, position based on what fitness is selected for the
  simulation.  It is doing this by contracting a set of 6 muscle bundles
  located in the bicep muscle.  Beside the arm are 6 lights representing
  key neurons in the network that controls the contraction of each muscle
  bundle. Every time one of the neurons fires, its color changes, resulting
  in one of three different outcomes.  When an Alpha-motoneuron fires,
  this causes the muscle bundle to contract, which in turn raises the 
  arm slightly.  When a Renshaw cell fires, it sends an inhibitory signal
  to the Alpha-motoneurons, possibly preventing them from firing.  And when an 
  Afferent cell fires, it sends a signal to the Alpha-motoneuron that 
  promotes additional firings.  Below is a diagram that better illustrates 
  the relationship between the different neurons.

	As you can see, the neurons are far from independent of each other. The
  Alpha-motoneuron connects to the Renshaw cell, the Afferent cell, and
  the muscle bundle.  When the neural network begins firing, each neuron must
  be synchronized to fire at precisely the right moment in order for the arm
  to make a smooth, controlled motion.