Here, we sought to replicate Bargh et al's [1] experiments. This was motivated by three main reasons. The first is simply that the finding, influential as it is, was only replicated twice so far, with neither replication being exact. The first replication [6] required participants to rate the walking speed of a character drawn on a sheet of paper after they had been primed with extreme exemplars associated to the concept of “speed” (e.g.: cheetah vs. turtle). The study found the expected priming effect. However, despite the fact that both this study and Bargh et al's share the same theoretical grounding, the latter only tells us of a bias in judgments of speed, as it did not require participants to actually perform any behavior. The second [7] replication aimed at refining Bargh et al's results by exploring the substantial variability exhibited by participants in their priming effect. The authors managed to replicate the results on walking speed, but the replication, like the original study, can be questioned based on imprecise timing methods (see below for a critical description).

Our second motivation was more conceptual. In social cognition, the assumption that high level semantic priming can occur automatically and outside of conscious awareness is almost taken for granted [8]. Yet, this assumption conflicts with evidences accumulated in cognitive neuroscience. For instance, several authors [9], [10], [11] suggest that two factors are necessary to produce the large patterns of neural activation in higher association cortices that are essential for semantic priming to occur: Top-down attention to the prime and bottom-up stimulus strength (e.g.: its saliency to the participants). Such a pattern of activation can be the signature of semantic processing of the prime [12] and is typically associated with conscious awareness [13]. In Bargh et al.'s experiments 2a and 2b, however, neither of these features were present. Thus, in view of the semantic priming literature, the salience of a concept such as “being old” seems too weak to automatically prime a behavior that is further only indirectly related to old age (i.e., through the concept of slowness), in the absence of any contextual cues relevant to this trait. In Bargh et al.'s account, the association between the categorical prime (old age) and walking speed (a behavior) is mediated by the automatic activation a stereotypical trait (slowness).

The absence of conscious awareness of the relation between the prime and the behavior is considered as evidence that this activation is indeed automatic.

Our third motivation is methodological. Some aspects of Bargh et al.'s [1] experiment's 2a and 2b remain unspecified yet impinge on the interpretation of the results. In particular, we see three potential challenges with the methods of the original study:

First, in the original study, participants' walking speed was measured by a confederate posted in the hallway adjacent to the experiment room. The confederate was unaware of whether a participant had been primed or not. The purpose of this setup was to ensure that the experiment was following a double-blind principle. However, no such precautionary measures are reported concerning the experimenter who administered the task to the participants. Numerous studies, however, have indicated that the experimenter's expectations can influence participants' behavior [14] even in the most controlled experimental environments [15]. In Bargh et al.'s study, the experimenter who administered the task could thus very well have been aware of whether the participant was in the prime condition or not and tune his or her behavior accordingly. This possibility was in fact confirmed informally in our own study, as we found that it was very easy, even unintentionally, to discover the condition in which a particular participant takes part by giving a simple glimpse to the priming material. Experimenters could thus unwittingly have communicated their expectations to participants [16] and influenced their walking speed. Thus, given the fact that subtle cues can influence our behavior [17], controlling for the experimenter's expectations appears to be essential.

Second, walking speed was measured using a manual stopwatch — a method that is prone to error and bias. Manual chronometry requires extra precautions [7] which appear to be absent from the original study.

Third, after the experiments, participants were debriefed using the contingency funnel procedure [18], [19] so as to assess, through increasingly specific questions, (1) whether they were aware of the purpose of the study and (2) whether they were aware of the fact that the words used in the scrambled sentences task were related to the concept of old age. Only 1 out of 19 participants [1] (p.237) were found to be aware of the influence of the primes, — a finding that formed the basis for the claim that the effect of stereotype activation on behavior is unconscious. However, it remains unclear exactly what participants claimed to be unaware of. As Nisbett and Wilson [20] famously pointed out, participants can remain (1) unaware of the stimulus, (2) unaware of their response, or (3) unaware of the fact that the stimulus importantly influenced the response. Thus, we set out to improve on the tests of awareness originally used by Bargh et al. [1] in hopes of better delineating exactly what people were aware of in this situation.

First, in Experiment 1, despite the use of a larger sample and an experimental procedure devoid of the limitations present in the original experiment, we were not able to replicate Bargh et al's [1] automatic effect of priming on walking speed. This led us to assume that crucial factors in this paradigm had remained unidentified. Experiment 2 was aimed at exploring such factors.

Second, in Experiment 2 we were indeed able to obtain the priming effect on walking speed for both subjective and objective timings. Crucially however, this was only possible by manipulating experimenters' expectations in such a way that they would expect primed participants to walk slower. Our results, however, cannot be explained solely in terms of a pure self-fulfilling prophecy effect [14], as the primed participants did not walk faster when tested by an experimenter who believed they would walk faster. Therefore it seems that the primes alone are not sufficient and must be in line with environmental cues such as the experimenters' behavior in order to elicit the effect on walking speed. This is also supported by the fact that contrary factors (i.e.: primes related to the concept of age in conjunction with an experimenter expecting to observer a faster walking speed) did not alter participants' walking speed.

Regarding the subjective timings, we obtained a reverse effect on walking speed (i.e.: participants walking faster). This effect can be explained by the error committed by the experimenters, most likely as a result of their induced expectations. If we had used only human-operated measurement devices, as in Bargh et al.'s [1] experiment 2a and 2b, we would thus have erroneously concluded in a reverse priming effect. This very important point suggests that one must be cautious about (1) the type of measurement used in behavioral priming experiments as well as (2) the experimenters expectations. Subtle differences are particularly prone to external influences and potential biases.

Third, most of the participants primed with the scrambled sentences task were aware of the social category they had been primed with. This result shows that one must be cautious when using the scrambled sentences task as a priming method. One must take into account that our participants were performing the experiment as part of a psychology course, which could have led to higher suspicion towards the scrambled sentences task resulting in a higher degree of awareness of the primes. Additionally, those participants who actually exhibited a slower walking speed reported in good proportion being aware of that particular behavior. Whether automatic behavioral priming can occur without awareness thus remains unclear. As a matter of fact, participants' awareness of the prime and of the primed behavior could have led them to exert better conscious control over the latter and therefore impair its expression.

Finally, priming alone was not sufficient to promote a priming effect on walking speed comparable to Bargh et al's [1]. We also had to manipulate experimenters' beliefs so that they would expect the primed subjects to walk slower. This finding is congruent with recent evidence showing that primed behavior is sensitive to the context in which it takes place [24], [25]. The present finding can be seen in the same light. Experimenters' expectations seem to provide a favorable context to the behavioral expression of the prime. Obviously, this interpretation remains tentative, as we do not know how this process operates. However, it is likely that experimenters who expect their participants to walk slower behave differently than those who expect their participants to walk faster and that such behavioral cues are picked up by participants.

Supporting Material S1