Equivalence-equivalence as a model of analogy: Further analysis

Equivalence-equivalence is assumed when training of A-B and A-C matching tasks not only leads to matching same-class B and C stimuli but also to matching BC compounds with same-class elements (e.g., B1 C1-B2C2) and with different-class elements (e.g., B1C2-B2C3). Like classical analogies (a: b:: c: d), equivalence-equivalence requires matching same functional relations. Experiments 1 to 4 examined equivalence-equivalence in 5-year-old children. In each experiment, subjects were tested for equivalence-equivalence before equivalence and, if they did not show equivalence-equivalence, also after the equivalence test. The experiments included various procedural arrangements designed to facilitate equivalence-equivalence, all of which failed. Only 8/18 children showed equivalence-equivalence, 2 before (11%) and 6 after equivalence (33%), irrespective of the facilitative procedures that were used. Adults served in Experiment 5. This experiment was the same as Experiments 1 through 4 but without facilitative arrangements. All adults showed equivalence-equivalence, most of them before equivalence. These and previously collected findings (Carpentier, Smeets, & Barnes-Holmes, 2002) suggest that equivalence-equivalence is an age-related performance similar to that which has been reported in earlier developmental studies on classical analogies. Yet, one should be cautious using equivalence-equivalence as a model for analogical reasoning. The testing procedures in both types of tasks are sufficiently different to permit the performances to be based on different behavioral processes. Recent studies have shown that after being trained on multiple arbitrary match-to-sample tasks, humans match not only functionally same stimuli but also functionally same stimulus relations (Barnes, Hegarty, & Smeets, 1997