How difficult is retrieval in analogical thinking?

MAXIMO TRENCH

Buenos Aires

Workshop; 1st International workshop on Analogy, Concepts and Representation; 2009

SADAF, Sociedad Argentina de Analisis Filosófico

The ability to retrieve analogous past episodes from long term memory is central to a flexible use of stored knowledge. This presentation begins by surveying the results yielded by classical experimental studies of analogical retrieval. The experimental procedure followed in these studies, which Blanchette and Dunbar (2000) called "reception paradigm", comprises two distinct phases: an encoding phase, where the base analogs are provided to participants by the experimenters, and a retrieval phase, where experimenters present participants with the target analogs and assess to what extent reading the targets elicits the retrieval of the base analogs. Results from this experimental tradition consistently demonstrate that retrieval of analogs from memory requires that the base and target analogs share a number of identical relations and objets (superficial similarity). Next we survey a small number of naturalistic studies allegedly demonstrating that superficial similarities are not required for analogical retrieval. Instead of assessing the retrieval probabilities of experimentally providad base analogs during a target task, the production paradigm implemented in these studies consisted of presenting a real-world target situation and asking participants to generate their own analogies favoring a concrete line of action for such target situation. The high absolute amounts of analogies lacking superficial similarities proposed by participants led the authors to conclude that naturalistic analogical retrieval is structural in nature, and that prior failures to elicit purely structural remindings originate in a lack of ecololgical validity of the recpetion paradigm used in experimental studies. We conclude with an exhaustive analysis of the procedural mismatches between the paradigms employed in the experimental and naturalistic traditions, and flesh out a number of hypothesis about how to reconcile their aparent constrast of results.