A computational approach for modeling the infant vision system in object and face recognition