Derived relations and generalized alteration of preferences

The present study examined the role of derived relations in the generalizability of the evaluative conditioning effect. Healthy university students participated. Four geometrical shapes were first established as discriminative stimuli for the contingent presentation of pictograms (B1, B2, C1, and C2, respectively). We then assessed the reinforcing properties of B1 versus B2, and C1 versus C2 by using simultaneous discrimination tasks: at baseline (baseline assessment), after pairing B1 with aversive slides plus noise and B2 with pleasant slides (test I), and after employ- ing equivalence training and testing to establish B1 as equivalent to C1 and B2 as equivalent to C2 (test II). Most participants (82 %) in the experimental condition, as compared with the control conditions (17 % and 10 %), selected the discriminative shapes for B2 (test I) and C2 (test II) on most trials, replicating and extending previous findings. Subsequently, the geometrical shapes were established as equivalent to the letters X, Y, W, and Z, respectively, which then served as antecedent stimuli in simultaneous discrimination tasks as before (test III). As was expected, only participants in the experimental condition showed preference for the novel letters that were established as equivalent to B2-producing and C2-producing shapes. These findings suggest that the evaluative conditioning effect may extend far beyond the stimulus being de/valuated and narrow the behavioral repertoire.