Beyond evaluative conditioning? Searching for associative transfer of nonevaluative stimulus properties

Abstract

Evaluative conditioning refers to the changes in liking of an evaluatively neutral stimulus (the conditional stimulus or CS) as a result of merely pairing it with another, already liked or disliked stimulus (the unconditional stimulus or US). We examined whether other, non‐evaluative stimulus properties of a US can also be associatively transferred to a CS. In a series of experiments, we tried to transfer perceptions of the gender of children and the gender of first names. We found evidence for the associative transfer of these properties but only when participants were aware of the contingencies.

Notes

Correspondence should be addressed to Jan De Houwer, Department of Psychology, Ghent University, Henri Dunantlaan 2, B‐9000 Ghent, Belgium; e‐mail: [email protected]

This research was supported by University of Leuven grant OT/99/11 to Paul Eelen and ESRC‐ROPA grant R022250197 to Jan De Houwer and Steven Glautier. We thank Hilde Hendrickx, Michael Olson, Deb Vansteenwegen, and Eva Walther, for providing us with information about their unpublished studies concerning non‐evaluative stimulus transfer.

We assumed that each choice during the awareness test was independent of the other choices and that the number of correctly identified pairs was thus binomially distributed when participants guess. Given these assumptions, in Experiment 6, a participant was classified as category aware if at least five of the six selected USs were from the correct category. The probability of achieving this category awareness score by chance is smaller than .05. In the other experiments, the criterium for classifying a participant as category aware was a category awareness score of at least 6 out of 9 (Experiments 1–5) or at least 4 out of 4 (Experiments 7 and 8). Due to space limitations, we only report separate analyses for category aware and unaware participants if prior analyses show that category awareness was related to the transfer effect.

We are aware of the fact that aggregating data of different experiments is not entirely appropriate when the design and procedure of the experiments is not identical. However, the different gender transfer experiments were similar to a large degree. Also, when the transfer scores of all 209 participants were analysed using an ANOVA with experiment (1–6) as between‐subjects variable, the main effect of experiment was not significant, F < 1. Another option would have been to perform a meta‐analysis on the data of the six experiments. However, such a meta‐analysis would have too little power to reveal possible moderating variables and could yield an inflated estimate of significance of the overall transfer effect (e.g., Field, 2003).