STUDENT PAPER WINNER
DERIVED STIMULUS RELATIONS UNDER CONTEXTUAL CONTROL PRODUCE
ASSOCIATION AND MEDIATED PRIMING
Robert A. Whelan
NATIONAL UNIVERSITY OF IRELAND - MAYNOOTH
A common method of assessing the nature of semantic relations in cognitive psychology is by means of a lexical decision task, in which two stimuli are presented, and the participant must decide whether both are “valid” stimuli or not. The first stimulus is called the “prime” and the second stimulus is called the “target”. Association priming refers to the tendency of a response time (RT) to be faster if the target is related to, or associated with, the prime. Mediated priming occurs where recognition of the target is facilitated through a mediating stimulus; for example, LION-STRIPES, where lion is assumed to prime the unrelated word stripes through the word tiger.
The concepts of Relational Frame Theory (RFT; see Hayes, Barnes-Holmes, & Roche, 2001) may provide an explanatory framework for priming phenomena. It is suggested that RFT can describe not only a subject's performance in a lexical decision task, but also the history of relational responding that made the performance possible. The present study examined whether stimuli in relational frames would prime each other, relative to previously-seen, but unrelated, stimuli.
Phase one of the preparation consisted of relational pretraining that brought subjects’ responses to the non-arbitrary stimulus relations of sameness and opposition under contextual control. Phase two involved training on a series of conditional discriminations with arbitrary stimuli, with each discrimination being made in the presence of one of the two contextual cues. This relational training established responding in accordance with relations of sameness and opposition between certain arbitrary stimuli (i.e., A1 same as B1 and C1, and A1 opposite to B2 and C2). Other, unrelated, stimuli were employed in the relational training phase as foils (N1, N2 etc.), which were then used during a subsequent lexical decision task. The RFT model predicts that the relational training described above would yield derived same relations between B1-C1 and B2-C2, and derived opposite relations between B1-C2 and B2-C1. Upon reaching criterion for the relational training phase, Subjects 1 and 2 were exposed to relational testing (i.e., conditional discriminations with no feedback). Subjects 3-7 were not exposed to a relational testing phase, to control for the possibility that priming might occur because certain stimuli were paired together in testing.
The lexical decision tasks commenced immediately after the completion of the relational training (and testing where applicable). The subjects’ task was to evaluate whether they had seen both stimuli previously (i.e., press “Y” if both stimuli were seen previously, or press “N” if one or both were novel stimuli). The crucial comparisons were between RTs to pairs of stimuli comprising members of the relational network (e.g., B2-C1), and RTs to “control pairs” composed of one stimulus from the relational network and one stimulus that the subject had been exposed to earlier in the relational training phase, but that did not enter into the relational network (e.g., A1-N1).
In the lexical decision task, all subjects who had reached criterion on the relational training phase responded more quickly to related pairs, relative to control pairs – the characteristic association priming pattern. The mean across-subjects RT for related pairs was 792 ms (SE = 31 ms), in contrast to 1162 ms (SE = 101 ms) for control pairs. Results also suggest that mediated priming occurred: the mean RT for combinatorially entailed pairs was 808 ms (SE = 66 ms). The error rate for related pairs was 2.4%, whereas the error rate for the control pairs was 40%. The higher error rate for unrelated pairs, in comparison to related pairs, is indicative of priming.
In conclusion, it seems that association and mediated priming can be demonstrated through both directly trained and derived multiple stimulus relations, thereby making more plausible the idea that relational frames provide a functional analysis of semantic relations.
REFERENCES
Hayes, S. C., Barnes-Holmes, D., & Roche, B. (2001). Relational frame theory: A post-Skinnerian account of human language and cognition. New York: Plenum Press.