Dopamine and norepinephrine interactions during visuomotor learning of simple and complex tasks: a computational model

S. E. LEW, H. G. REY, B. S. ZANUTTO

EEUU

Congreso; Neuroscience 2007; 2007

Society for Neuroscience

In this work, a computational model with the ability to learn simple (e.g., Visual Discrimination, VD) and complex (e.g., Delayed Matching to Sample, DMTS) tasks is presented. Dopamine and Norepinephrine effluxes at cortical and sub cortical brain areas are in strong relation with responses exploration and behavioral performance. Due to an exploration-exploitation mechanism based on biological hypotheses, the model learns to respond properly in changing environments.Effects of dopamine and norepinephrine over neurons in cortical and subcortical structures are analyzed based on different behavioral paradigms. The model includes the role of dopamine modulation in the excitability and learning on prefrontal cortex and motor-related structures neurons. Neurons at those structures change their synaptic weights following a Hebbian or anti-Hebbian rule depending on the released dopamine. The increase in reward rate allows the dopamine signal to induce exploitative behavior. On the other hand, tonic level of norepinephrine modulates the energy flow from sensory inputs to motor structures and the excitability of dopaminergic neurons, allowing the shifting between exploitation and exploration strategies.Tested on simple and complex paradigms, the computational model predicts behavioral and physiological results, including learning of response selection, VD and DMTS tasks and faster learning after each reversal procedure. The model provides a computational framework to the exploration-exploitation dilemma for adaptive agents.