1. Certain nonlinear properties of molluscan neurones suggest that network activity could be described in relation to nonlinear relaxation oscillators. 2. A specific example is fitted by the Van der Pol equation with an exponentially decaying damping coefficient, and changes in the associated energy and sensitivity are considered. 3. More precise details of the interaction pattern determined by analysis of a single cycle of burst activity, and it is shown that there are two states of activity, depending on the balance of presumed inhibitory components. 4. Previous results are discussed in relation to an overall decaying oscillation. 5. Changes in subcomponents of a cycle are described for decaying sequences and the constancy of synaptic interaction is demonstrated. The results are briefly discussed in relation to catastrophe theory and learning.

This content is only available via PDF.