Role of Behavior in the Development of Complex Neuronal Properties

It has long been recognized that changes in neural activity within an organism's brain may cause specific changes in its behavior. However, much less is known about the specific role of self-generated movement and of behavioral interactions with an environment in the development of complex neuronal properties. To address this question, we designed Darwin V, which consists of a simulated neuronal model embedded in a real world device capable of autonomous behavior. The model has four major components: a visual system incorporating several key properties of mammalian visual cortex, a taste system based on conductance, sets of motor neurons capable of triggering behavior, and a diffuse ascending (value) system. As Darwin V encounters stimulus objects in its environment, synaptic modifications within its visual maps produce pattern-selective and translation invariant neuronal responses. Due to value-dependent synaptic modifications in sensorimotor connections, visual neurons become linked to motor neurons triggering appropriate behaviors. Closer analysis reveals that the acquisition of translation invariance depends critically upon continuous self-generated movements of Darwin V. In addition, the allocation of map areas selective for particular stimulus classes depends upon frequency and historical sequence of stimulus encounters. In summary, the analysis of Darwin V provides several examples for the role of environment and behavior in the development of complex neuronal responses.