On Lieto Antonio, Cognitive Design for Artificial Minds

The computational cognitive science that uses cognitive modeling in-volves some problems, among which the main one is the problem of model. What makes a computational model a cognitive one? What are the right and relevant constraints to build a model that is not merely a system producing the same performance in specific tasks as the humans do? The focus of Lieto’s proposal is on cognitive architectures, a notion that was introduced by Newell in his attempt to define a unified theory of cognition. They are abstract models between the high-level cognitive capabilities and their neural/bodily implementation, so they are at an intermediate level and their characterization as an integrated mechanism is what allows to build a computational counterpart of them in an artificial system. The more interesting cognitive architectures are, clearly, the more general ones, i.e. the ones modeling the cognitive capabilities at the highest degree of integration among intelligent features. The intermediate nature of cognitive architecture makes the problems of relevant constraints of modeling a crucial one to achieve an actual model of cognitive processes.