How similar are recognition memory and inductive reasoning?

Abstract

Conventionally, memory and reasoning are seen as different types of cognitive activities driven by different processes. In two experiments, we challenged this view by examining the relationship between recognition memory and inductive reasoning involving multiple forms of similarity. A common study set (members of a conjunctive category) was followed by a test set containing old and new category members, as well as items that matched the study set on only one dimension. The study and test sets were presented under recognition or induction instructions. In Experiments 1 and 2, the inductive property being generalized was varied in order to direct attention to different dimensions of similarity. When there was no time pressure on decisions, patterns of positive responding were strongly affected by property type, indicating that different types of similarity were driving recognition and induction. By comparison, speeded judgments showed weaker property effects and could be explained by generalization based on overall similarity. An exemplar model, GEN-EX (GENeralization from EXamples), could account for both the induction and recognition data. These findings show that induction and recognition share core component processes, even when the tasks involve flexible forms of similarity.

Event-related potential parameters of category and property violations during semantic category-based induction

“…Behavioral studies showed that induction can be varied with different properties (Hayes and Heit, 2013;Heit and Rubinstein, 1994). For example, Hayes and Heit (2013) compared the inductions based on aquatic mammal premises with three kinds of properties: a blank property (a newly discovered property X), a habitat property (a newly discovered enzyme to assist in the development of a water-resistant body covering) and a mammal property (a newly discovered enzyme in breast milk that is passed to the young during feeding); they found that participants made more positive responses to aquatic non-mammals in the habitat property than in the blank property, and made more positive responses to land mammals in the mammal property than in the blank property. Based on these findings, it is reasonable to infer that meaningful property violations can increase the amount of semantic content violations, which in turn can influence the N2-P3b-N400 effects.…”

Missing the Party: Political Categorization and Reasoning in the Absence of Party Label Cues

“…Rather, we examined whether cues are used differently for party identification and for voting, and differently for high‐ versus low‐knowledge voters, and differently for Democrats versus Republicans. It is known from laboratory experiments that cue weights are influenced by the task (e.g., Hayes & Heit, ; Nosofsky, ); here the novel question is how do cue weights vary for different political judgments. Likewise it is known that cue weights for biological categorization and reasoning depend on expertise (e.g., Medin & Atran, ; Medin, Lynch, Coley, & Atran, ).…”

A dynamic model of reasoning and memory.

“…Decisions in each task (whether the test item has been seen before; whether the test item shares a property with the study set) could be seen as driven by an assessment of the similarity between the test probe and the retrieved sample. These hypothesized parallels have been supported in studies comparing induction and recognition responses to a common set of study and test items (Hayes & Heit, 2013a;Hayes, Fritz, & Heit, 2013;Heit & Hayes, 2011). Notably, these same studies also revealed a reliable empirical difference between the two tasksnamely, that induction leads to higher rates of positive responding to novel test items than recognition.…”