Speaker: Dr. Alireza Soltani
Postdoctoral Associate
Department of Neurobiology, Yale University, New Haven, USA
Local Organizer: Shahin Nasr and Ali Moeeny
Title: A Cortical Network Model of Probabilistic Decision-making
Time: Tuesday, Dec 19, 2006
Location: School of Cognitive Science, IPM, Tehran
Abstract:
In recent experiments exploring the neural basis of adaptive decision making, neuroscientists have reported single-cell activities in the lateral intraparietal cortex (LIP) correlated with choice options, subjective value of options, prior probability, and evidence that an alternative is rewarded. To investigate mechanisms underlying these processes, we considered a biophysically-plausible spiking neuron model of decision making which is endowed with plastic synapses.
This general model is tested in different probabilistic decision-making tasks. First, it reproduces matching behavior in a task in which rewards on two targets are scheduled randomly with different rates. In this task, plastic synapses can integrate past rewards and estimate local “return” of a choice.
Second, in a perceptual motion discrimination task with unequal prior probability for each outcome, plastic synapses provide an additional signal to shift choice behavior and shorten reaction times toward the more probable alternative, both observed experimentally.
Third, in the weather prediction task, the same plastic synapses can store evidence on how likely each shape is predictive of an outcome.