THE ROLE OF INDIVIDUAL STIMULUS NAMES IN THE EMERGENCE OF EQUIVALENCE-RELATIONS .2. THE EFFECTS OF INTERFERING TASKS PRIOR TO AND AFTER TESTS FOR EMERGENT RELATIONS

In two experiments on equivalence class formation subjects were taught arbitrary trained relations between icons of familiar objects by matc hing-to-sample on a computer screen, before receiving four blocks of t ests for trained and emergent relations. There were six AB/BC training sets so that potentially six ABC equivalence classes would be formed. In Experiment 1, an early discordant group of subjects was taught, pr ior to testing, oral paired associations between the names which the i ndividual subjects gave to the A and the B visual stimuli in combinati ons which were systematically discordant with the AB matching-to-sampl e-based trained relations. In a second late discordant group, discorda nt paired associates were taught after two blocks of tests and in a th ird control group no discordant associations were taught. In the early discordant group it was found that paired associate-based links subst ituted for matching-to-sample-based links in many choices in all types of tests in which they could be implicated so that there was a signif icantly lower proportion than in the other two groups of choices consi stent with equivalence class formation based upon the matching-to-samp le training. It was found that for the late discordant group paired as sociate-based relations did not displace matching-to sample-based rela tions. A high proportion of choices continued to be consistent with eq uivalence classes based upon the preliminary matching-to-sample traini ng, as was found in the control group. In a second experiment with int erpolated paired associate-like training using the visual stimuli them selves (rather than their names) in a similar early versus late design it was found that choices based on discordant visual relations displa ced a considerable proportion of choices based on matching to sample t raining, whether interpolated before or after some initial matching to sample testing. These results suggest that subjects' names for indivi dual stimuli may play a role in the formation of equivalence classes b ut play little part in the maintenance of such classes once formed.