A. Innis et al., STIMULUS EQUIVALENCE - EFFECTS OF A DEFAULT-RESPONSE OPTION ON EMERGENCE OF UNTRAINED STIMULUS RELATIONS, Journal of the experimental analysis of behavior, 70(1), 1998, pp. 87-102
Default-response options, intended to measure uncertainty, sometimes a
re included in discrete-choice measurement in an attempt to clarify st
imulus control of remaining response options. Recent experiments have
used a default-response procedure to investigate emergent stimulus rel
ations, but no study to date has compared effects of different default
-response procedures on emergence of the untrained relations that defi
ne stimulus equivalence. Five groups of college undergraduates (each n
= 16) completed a conditional discrimination training procedure to in
state the stimulus relations prerequisite to three three-member equiva
lence classes; a training review intermingling all of the explicitly t
rained relations; and tests for emergent relations. The groups differe
d in terms of (a) presence versus absence of a ''none'' option during
emergent relations tests and (b) the amount of experience with ''catch
trials'' in which ''none'' was the correct selection. Stimulus equiva
lence was demonstrated in 94% of subjects in a control group who were
trained and tested without the ''none'' response option and without ca
tch trials and in 41% of subjects in the ''none'' groups. Among subjec
ts in the ''none'' groups who failed to demonstrate equivalence initia
lly, 95% did so when retested under control-group conditions. Across '
'none'' groups, probability of equivalence class formation was positiv
ely correlated with amount of experimental experience with catch trial
s in preliminary training and equivalence testing. Among the emergent
relations defining stimulus equivalence, reflexivity was most often pr
ecluded by the ''none'' option, although there was evidence of group d
ifferences in relation specificity. These results suggest that a defau
lt-response option can interfere with the formation of emergent relati
ons, and that the effects are contextually sensitive. Although there m
ay be advantages to employing default-response procedures in studies o
f emergent stimulus relations, the responses they control should be vi
ewed as behavior under specific stimulus control rather than a generic
expression of uncertainty.