Registration of neural maps through value-dependent learning: modeling the alignment of auditory and visual maps in...

In the optic tectum (OT) of the barn owl, visual and auditory maps of space are found in close alignment with each other. Experiments in which such alignment has been disrupted have shown a considerable degree of plasticity in the auditory map. The external nucleus of the inferior colliculus (ICx), an auditory center that projects massively to the tectum, is the main site of plasticity; however, it is unclear by what mechanisms the alignment between the auditory map in the ICx and the visual map in the tectum is established and maintained. In this paper, we propose that such map alignment occurs through a process of value-dependent learning. According to this paradigm, value systems, identifiable with neuromodulatory systems having diffuse projections, respond to innate or acquired salient cues and modulate changes in synaptic efficacy in many brain regions. To test the self-consistency of this proposal, we have developed a computer model of the principal neural structures involved in the process of auditory localization in the barn owl. This is complemented by simulations of aspects of the barn owl phenotype and of the experimental environment. In the model, a value system is activated whenever the owl carries out a foveation toward an auditory stimulus. A term representing the diffuse release of a neuromodulator interacts with local pre- and postsynaptic events to determine synaptic changes in the ICx. Through large-scale simulations, we have replicated a number of experimental observations on the development of spatial alignment between the auditory and visual maps during normal visual experience, after the retinal image is shifted through prismatic goggles, and after the reestablishment of normal visual input. The results suggest that value-dependent learning is sufficient to account for the registration of auditory and visual maps of space in the OT of the barn owl, and they lead to a number of experimental predictions.