A model for the role of feedback in ocular dominance column development.

Abstract

Feedback is ubiquitous in the mammalian brain, and has been shown to play a very important role in perception and cognition. However, feedback is also present in the developing brain, and as such may play an important role in brain development. We present a model to investigate the role of feedback during brain development, specifically in formation of ocular dominance columns (ODC's). Feedback is a unique feature of our model. Results of anatomical and physiological experiments show that feedback loops exist between cortical layer-4 and other neuron populations (the subplate and cortical layer-6) during ODC development. Our model puts forth the hypothesis that these neuron populations locally integrate the activity of layer-4 neurons and feed it back to same, specifically to the post-synaptic sites of geniculocortical NMDA synapses. In this way, the feedback activity directly controls the synaptic strength changes, causing the development of ODC's. Another unique feature of our model is that the ALOPEX algorithm is used change the geniculocortical synaptic strengths. Unlike the Hebb rule, ALOPEX is naturally suited to systems with feedback. Through detailed neuronal simulations, we demonstrate the biophysical feasibility of ALOPEX in the current system. Computer simulations demonstrate that our model is extremely robust. The average column width remains constant against variations in thalamic and cortical parameters. When biologically realistic parameter values are used, the model renders ODC's of an appropriate width. Further, the model successfully replicates the effects of monocular deprivation, reversed monocular deprivation, and strabismus. It also renders a consistent column width under different scales of cortical representation. From these results, we conclude that the role of feedback is to make development robust against biological variabilities.