Deliberative/Reflexive Learning Dichotomy

Learning is a capability that is incorporated in many cognitive architectures, but it can take a number of different forms. Deliberative learning attempts to compensate for the limited storage inherent in any agent by reflecting on the utility of a given piece of knowledge: if some knowledge is not useful, or at least too expensive to store or utilize relative to its utility, then it is not stored. Reflexive learning allows for completely "automatic" learning, where every new piece of knowledge is stored; knowledge is retrieved using a very fast matching algorithm to compensate for the large knowledge base that accumulates. It is argued that relexive learning is more psychologically valid approach, since humans are generally incapable of NOT learning or "un"-learning something that we decide is not sufficiently useful

Back to List of Dichotomies