Figure 2.

Development during the training cycle of the example simulation. A: limb kinematics during the training cycle, sampled every 500 s. Top and bottom: x-position relative to the center of limb travel (left limb: top; right limb: bottom) as a rolling mean (dark line) and range (lighter shading) over 5,000 s (initial zero-padding indicated by gray background). The rolling mean maximal velocity for each limb is shown in the middle, positive upward for the left limb and positive downward for the right limb. Note the slight increase in velocity and range of motion during the initial phase of the training but with no mean deviation from the neutral position. B: synaptic weight development for each neuron type during the example simulation. Each input source (excitatory interneuron, inhibitory interneuron, and primary afferent) is shown with separate color code. After roughly a days’ worth of experience, the synaptic weights of the network quickly stabilize, with only slow and sparse changes in the later stages. INe, excitatory interneurons; INi, inhibitory interneurons; MN, motoneuron.