Traditional results using a cued-recall paradigm have allegedlydemonstrated that distant analogs tend to be retrieved less oftenthan disanalogous matches maintaining only surface similarity.Recent results, however, suggest that said advantage may be due tothe inadvertent inclusion of structural similarity in surface matches.In two experiments we had distant analogs compete in LTMwith two types of surface matches lacking any degree ofstructural overlap, but equated with the target in terms ofelement similarities. Distant analogs were less retrieved thatstories maintaining similar first-order relations and objects withthe target, but no overlapping structure. This differencedisappeared when surface similarity involved only similar objects.Results show that the surface superiority effect relies on thetype of surface matches that compete with distant analogs, thussuggesting a more complex picture of the forces that governaccess to similar items in memory.

Against the typical results from laboratory studies, it has beensuggested that retrieving distant analogs might be easy in real-life, where we tend to encode familiar situations with expert-like schemas. In each of two experiments, we formed twogroups of participants who, as determined by a questionnairepresented during a first session, had reported that they haveexperienced an event corresponding to a schema-governedcategory (Experiment 1) or to a system of schema-governedcategories (Experiment 2). While the episodes reported by oneof the groups belonged to the same domain as the target analogto be presented during the second session, those of the othergroup belonged to a different thematic domain. During atemporally and contextually separated session, the experimenterspresented both groups with a target analog belonging to theschema-governed category for which participants had reporteda base analog. Participants had to retrieve an autobiographicalepisode that they considered analogous to the situationpresented by the experimenter. In line with traditional studies,we found that retrieving distant instances of relationalcategories is much more difficult than retrieving closeinstances.

The present study was aimed at assessing the effect of object similarities on participants' evaluations of analogical quality. Results from an experimental condition in which the relations involved in the compared situations were explicitly highlighted, showed that general object similarities (membership to same category) positively affected the evaluations of analogical quality. In contrast, no such effect was found under another experimental condition in which the analogical comparisons between the same situations were framed by a schema-governed category. An analysis of participants' justifications revealed that the object similarities that were taken into account under this second condition were related to central dimensions of the schema-governed category that was used to frame the analogies. We explain these findings within the category assignment approach developed by Minervino et al., and discuss the implications of this alternative perspective of analogical reasoning for the role of similarities between entities playing several thematic roles.