Can DMTS paradigm be learned only by operant conditioning mechanisms ?

S.E. LEW, B.S. ZANUTTO

Florida, USA

Congreso; 32th Annual Meeting of the Society for Neuroscience; 2002

Society for Neuroscience

 We present a neural network model that learns delay matching to sample (DMTS). Our model of operant conditioning is able to predict the most relevant experimental features for appetitive and for the aversive stimulus (e.g. transfer of stimulus, escape and avoid, response selection and matching law among other experiments), but when it was tested in DMTS its accuracy was close to chance levels. Many authors (e.g. Miller & Cohen, 2001) proposed that the prefrontal cortex (PFC) plays an important role in the learning of rules. On these bases the model has three blocks, one for the region of the PFC involved in the learning of rules (e.g. lateral PFC) and, as in our previous model, another representing the ventro tegmental area and sustantia nigra (VTA-SN) and a last one for the premotor and the motor cortex (PMC-MC). Inputs to the model are traces of CS’s and of US. Neurons in the VTA-SN learn to predict US updating their synapses by the Rescorla Wagner rule. This prediction activates neurons in the PFC increasing their probability of firing. Neurons in the PFC and the PMC-MC update their synapses by hebbian or anti hebbian rule depending on US. After learning, a map of trained rules that emerges from the PFC plays a key role in the execution of responses according to these rules. The model learns DMTS performing close to 90% of accuracy. During learning, the model shows a spread massive activation of neurons in the PFC as can be seen in neuroimaging studies in monkeys and, after criterion is reached only neurons associated with learned rules are firing. Also, the model explains why rats perform close to chance of accuracy when the sample is presented in a novel location.