Search

In previous studies, we found that bodily symptoms can be learned in a differential conditioning paradigm, using odors as conditioned stimuli (CSs) and CO2-enriched air as unconditioned stimulus (US). However, this only occurred when the odor CS had a negative valence (a selective conditioning effect), and tended to be more pronounced in persons scoring high for Negative Affectivity (NA). This paper considers the necessity and/or sufficiency of awareness of the CS–US contingency in three studies using this paradigm. The relation between contingency awareness and the selective conditioning effect, and between contingency awareness and NA was also considered. Both self reported symptoms and respiratory physiology served as dependent variables. A learning effect on symptoms was found only for participants aware of the CS–US contingency, but not all participants reporting contingency awareness showed a learning effect. No conditioning effects appeared on the physiological measures. Also contingency awareness did not account for the selective conditioning effect, and did not interact with NA. Overall, the necessity but insufficiency hypotheses can only be withhold for group data and not for individual data.

We argue that the effects of evaluative learning may occur (a) without conscious perception of the affective stimuli, (b) without awareness of the stimulus contingencies, and (c) without any awareness that learning has occurred at all. Whether the three experiments reported in our target article provide conclusive evidence for either or any of these assertions is discussed in the commentaries of De Houwer and Field. We respond with the argument that when considered alongside other studies carried out over the past few decades, our experiments provide compelling evidence for a theory that posits a dissociation between evaluative learning and contingency awareness.

The role of contingency awareness in evaluative conditioning has been a contentious issue for quite some time now. Several papers provide a review of the relevant evidence (De Houwer, Baeyens, & Hendrickx, 1997a; De Houwer, Thomas, & Baeyens, in press; Field, 2000; Lovibond & Shanks, in press; Shanks & St. John, 1994), but the conclusions that are reached differ markedly. For instance, whereas De Houwer et al. (in press) concluded that evaluative conditioning is largely independent of contingency awareness, Field (2000, p. 32) argued that there is little unequivocal evidence that evaluative conditioning without contingency awareness is a robust finding. The results of Fulcher and Hammerl provide another interesting contribution to the debate. The aim of this commentary is to illustrate how even a single series of studies such as those of Fulcher and Hammerl can be interpreted in different ways depending on one’s perspective. I will first adopt a sceptical point of view and will try to argue that the evidence presented by Fulcher and Hammerl does not provide conclusive evidence for unconscious evaluative conditioning. Afterward, I will look at the same studies from a more open, broader point of view and argue that the reported results are important and that they are in line with the results of many previous results which also support the hypothesis that, in comparison to other forms of Pavlovian conditioning, evaluative conditioning is largely independent of contingency awareness

Fulcher and Hammerl's (2001) important exploration of the role of contingency awareness in evaluative conditioning (EC) raises a lot of issues for discussion: (1) what boundaries, if any, exist between EC and affective learning paradigms?; (2) if EC does occur without awareness does this mean it is nonpropositional learning?; (3) is EC driven by stimulus–response (S–R), rather than stimulus–stimulus (S–S), associations and if so should it then surprise us that contingency awareness is not important?; and (4) if S–R associations are at the heart of EC, should we automatically assume EC is part of a different learning mechanism to autonomic Pavlovian conditioning (Field, 2000a, 2000b)? This article, after a critical review of Fulcher and Hammerl's work, discusses these issues with reference to what can be realistically inferred about the mechanisms underlying EC.

In three experiments, each of a set colour-unrelated distracting words was presented most often in a particular target print colour (e.g., “month” most often in red). In Experiment 1, half of the participants were told the word-colour contingencies in advance (instructed) and half were not (control). The instructed group showed a larger learning effect. This instruction effect was fully explained by increases in subjective awareness with instruction. In Experiment 2, contingency instructions were again given, but no contingencies were actually present. Although many participants claimed to be aware of these (non-existent) contingencies, they did not produce an instructed contingency effect. In Experiment 3, half of the participants were given contingency instructions that did not correspond to the correct contingencies. Participants with these false instructions learned the actual contingencies worse than controls. Collectively, our results suggest that conscious contingency knowledge might play a moderating role in the strength of implicit learning.

Research has shown repeatedly that attention influences implicit learning effects. In a similar vein, interoceptive awareness might be involved in unaware fear conditioning: The fact that the CS is repeatedly presented in the context of aversive bodily experiences might facilitate the development of conditioned responding. We investigated the role of interoceptive attention in a subliminal conditioning paradigm. Conditioning was embedded in a spatial cueing task with subliminally presented cues that were followed by a masking stimulus. Response times to the targets that were either validly or invalidly predicted by the cues served as index of conditioning. Interoceptive attention was manipulated between-subjects. Half the participants completed a heartbeat detection task before conditioning. This task tunes attention to one’s own bodily signals. We found that conditioned responding was facilitated in this latter group of participants. These results are in line with the hypothesis that a rise interoceptive attention enhances unaware conditioned responding.

In a series of four experiments, we examine the hypothesis that selective attention is crucial for the generation of conscious knowledge of contingency information. We investigated this question using a spatial priming task in which participants were required to localize a target letter in a probe display. In Experiment 1, participants kept track of the frequency with which the predictive letter in the prime appeared in various locations. This manipulation had a negligible impact on contingency awareness. Subsequent experiments requiring participants to attend to features (color, location) of the predictive letter increased contingency awareness somewhat, but there remained a large proportion of individuals who remained unaware of the strong contingency. Together the results of our experiments suggest that the construct of attention does not fully capture the processes that lead to contingency awareness, and suggest a critical role for bottom-up feature integration in explicit contingency learning.

In her commentary of Field (1999), Hammerl (1999) has drawn attention to several interesting points concerning the issue of contingency awareness in evaluative conditioning. First, she comments on several contentious issues arising from Field's review of the evaluative conditioning literature, second she critiques the data from his pilot study and finally she argues the case that EC is a distinct form of conditioning that can occur in the absence of contingency awareness. With reference to these criticisms, this reply attempts to address Hammerl's comments by exploring the issues of how awareness is defined, how it is best measured, and whether it is reasonable to believe that EC uniformly occurs in the absence of contingency awareness. The article concludes that the available evidence supports Field's proposition that EC is, in fact, Pavlovian learning. Copyright 2000 Academic Press.