A neurocomputational model of automaticity and maintenance of abstract rules

Rule-guided behavior is essential in quickly adapting to one´s ever-changing environment. In particular, learned rules can quickly be used in new contexts or applied to new stimuli (which confers an advantage over restricting learning to perceptual - motor associations). Here, we propose a new neurocomputational model of automaticity in rule-guided behavior. The proposed model assumes two parallel neural pathways corresponding to ldquonaiverdquo and ldquoexpertrdquo rule use. The development of automaticity is characterized by a transfer of control of rule-guided behavior from a pathway mediated by the prefrontal cortex to a direct parietal-premotor pathway. The model includes differential equations that describe voltage changes in the relevant brain areas and difference equations that describe the Hebbian learning. A simulation shows that the model accounts for some critical single-cell recording data from several key brain areas as well as some important behavioral results.