Most accounts I read involving action potentials and synapses and the like tend to focus mostly on the action potential as a mere automatic reaction to another similar event happening upstream.
From a more theoretical point of view, it would be nice to know if (and under what circumstances) a neuron would be able to generate an action potential, in the absence of an upstream neuron firing first.
I suspect that the nervous system, at least in higher animals, is not simply a relay system that processes and transmits information from senses to actuators like the limbs. The relay system picture is fine for short-timescale, reactionary behavior but it is hard to imagine any longer-timescale learning processes happening this way. This is what motivates my question.