Boundary conditions for the influence of unfamiliar non-target primes in unconscious evaluative priming: The moderating role of attentional task sets

Highlights

• Affective stimulus evaluation can occur for unconsciously perceived masked stimuli.

• Familiar primes that are also presented as targets elicited robust priming effects.

• Unfamiliar primes that were not presented as targets are only effective on low repetition targets.

• Unfamiliar primes do not elicit priming effects on targets repeated at a high frequency.

• Priming by unfamiliar primes requires a task set involving evaluative semantic analysis.

Abstract

Evaluative priming by masked emotional stimuli that are not consciously perceived has been taken as evidence that affective stimulus evaluation can also occur unconsciously. However, as masked priming effects were small and frequently observed only for familiar primes that there also presented as visible targets in an evaluative decision task, priming was thought to reflect primarily response activation based on acquired S–R associations and not evaluative semantic stimulus analysis. The present study therefore assessed across three experiments boundary conditions for the emergence of masked evaluative priming effects with unfamiliar primes in an evaluative decision task and investigated the role of the frequency of target repetition on priming with pictorial and verbal stimuli. While familiar primes elicited robust priming effects in all conditions, priming effects by unfamiliar primes were reliably obtained for low repetition (pictures) or unrepeated targets (words), but not for targets repeated at a high frequency. This suggests that unfamiliar masked stimuli only elicit evaluative priming effects when the task set associated with the visible target involves evaluative semantic analysis and is not based on S–R triggered responding as for high repetition targets. The present results therefore converge with the growing body of evidence demonstrating attentional control influences on unconscious processing.

Introduction

For several decades, automatic emotional evaluations have been investigated with the evaluative priming paradigm. In evaluative priming, primes and targets (words or pictures) either share the same emotional valence (e.g., both positive: baby – rabbit) in the congruent condition or exhibit a different valence (e.g., positive vs. negative: baby – shark) in the incongruent condition (Fazio, 2001, Fazio et al., 1986). In an evaluative judgment task (i.e., a pleasant vs. unpleasant decision) on visible target stimuli, responses are typically faster when primes and targets exhibit the same valence (for reviews, see Fazio, 2001, Klauer and Musch, 2003; for a meta-analysis, see Herring, White, Jabeen, Hinojos, Terrazas, Reyes, Taylor, & Crites, 2013). Evaluative priming is also observed, when the prime stimulus is briefly presented and masked by a visual pattern (e.g., random sequence of letters) so that it cannot be consciously perceived (e.g., Abrams et al., 2002, Draine and Greenwald, 1998, Greenwald et al., 1996, Klauer et al., 2007, Klauer et al., 2003, Wentura and Degner, 2010). The observation of unconscious (subliminal) evaluative priming shows that this effect depends on automatic processes, which are triggered involuntarily.

However, the precise nature of the processes underlying masked evaluative priming is a matter of debate. This is particularly due to the fact that evaluative subliminal priming effects were small and only rarely observed for unfamiliar or novel primes, that is, primes that were never presented supraliminally for an evaluative categorization throughout the experiment as a target (Klauer et al., 2007, Wentura and Degner, 2010).

The dominant account of evaluative priming in the evaluative decision task is the response activation account (Klauer et al., 1997, Klinger et al., 2000, Wentura, 1999). In this task, evaluatively congruent and incongruent prime-target pairings differ not only with regard to evaluative congruency, but also with regard to response congruency: In congruent trials, primes and targets are associated with the same response, whereas in incongruent trials they are associated with different responses. In congruent trials, the response to the target is quickly selected as the response pathway already has received some activation. In incongruent trials, target responding is slowed down by a response conflict (Damian, 2001, Klauer et al., 1997, Klinger et al., 2000, Wentura, 1999). Consequently, the response-related processing account conceptualizes evaluative priming in the evaluative decision task as a variant of response priming.

We are aware that evaluative priming can also arise from automatic activation of emotional semantic prime features, which facilitates subsequent target processing, as proposed in semantic activation accounts (Fazio et al., 1986, Spruyt et al., 2009). However, in behavioral studies, semantic activation processes can only be unequivocally demonstrated in target tasks such as pronunciation, in which evaluative congruency is independent of response congruency (Spruyt et al., 2002, Spruyt et al., 2004). As the present work investigates mechanisms underlying masked priming in the evaluative decision task, we focus on response activation processes and do not discuss semantic activation in detail (for attempts to disentangle processes by using evaluative categorization data, see Eder et al., 2012, Voss et al., 2013).

Within the class of response priming accounts, however, two variants can be distinguished: According to one variant, response priming is based on a direct association between the prime stimulus and the response (Damian, 2001, Klinger et al., 2000) which is acquired during the experiment, perhaps already during a practice phase. This S–R learning can take place when targets are also used as primes.¹ If priming is based on acquired S–R associations, only primes which are also presented as targets should elicit priming effects (Abrams & Greenwald, 2000). It has been claimed that S–R learning explains at least a large part of subliminal priming effects (Damian, 2001, Klinger et al., 2000).

According to the second variant, response priming is based on the implicit application of task-control representations (‘task sets’) (Ansorge et al., 2014, Klauer et al., 2007, Kunde et al., 2003, Neumann, 1990) to the prime, whether consciously presented or subliminally, although it is not required by the task. This account was originally developed for explaining subliminal response priming effects based on non-emotional stimuli such as visual shapes or colors (Ansorge and Neumann, 2005, Neumann, 1990, Neumann and Klotz, 1994), but can be easily applied to the field of masked evaluative priming. According to this view, participants establish a task set on the grounds of the experimental instruction (e.g., “press left key in response to a positive stimulus, press right key in response to a negative stimulus”). If the prime matches this prepared task set, the task set is executed and the corresponding response is activated. This task set execution account of response priming allows for some flexibility compared with an S–R account, because it predicts priming also for unfamiliar, novel primes, which are not presented as targets. Unfamiliar primes can lead to task set execution, as long as they are sufficiently similar to the information specified in the task set (Kunde et al., 2003). This includes an at least coarse semantic analysis of the prime stimulus, especially with regard to its valence, in order to determine whether it is suited to execute the task set (Dehaene et al., 1998, Kiesel et al., 2006). Furthermore, the task set execution account of response priming opens the room for attentional influences such as stimulus expectations, which determine whether a unfamiliar prime is able to elicit priming effects (Kiefer, 2012, Kiefer and Martens, 2010).

Many studies within the field of unconscious priming support this task set execution account (for a recent review, see Ansorge et al., 2014): As already noted above, masked evaluative priming with unfamiliar primes that are never presented supraliminally throughout the experiment has occasionally been observed (Klauer et al., 2007, Wentura and Degner, 2010). Furthermore, subliminal visuo-motor priming effects were influenced by action intentions and stimulus expectations (Ansorge et al., 2002, Ansorge et al., 2010, Ansorge and Neumann, 2005, Tapia et al., 2010, Wokke et al., 2011). There is also evidence that evaluative priming depends on task sets. Evaluative congruency of subliminally presented masked prime words elicited only priming effects on the subsequent target during an evaluative decision, but not during a non-emotional semantic (living/non-living) decision task (Eckstein and Perrig, 2007, Klinger et al., 2000). Based on the assumption that automatic emotional processing requires an attentional sensitization of emotional pathways (Herring et al., 2013, Kiefer and Martens, 2010, Spruyt et al., 2012), evaluative priming was only found if the majority of trials were evaluation trials (Spruyt, De Houwer, Hermans, & Eelen, 2007).

In addition to action intentions or feature-specific attention, unconscious processing has been shown to depend on stimulus expectations. Stimulus expectations establish an attentional set, which indicates what kind of stimuli is likely to occur within a given situation. Expected subliminal stimuli are part or at least similar to representations within the established task set and receive attentional amplification compared with unexpected stimuli (Kiefer, 2012). It should be noted that expectations cannot be established by unconsciously presented stimuli themselves, but must be formed by consciously perceived stimuli presented within a given context, for instance by the visible target stimuli of a priming paradigm. It has been shown that the type of visible target stimuli included in an experiment strongly influences subliminal priming effects: Masked stimuli prime responses only if they are expected and represent possible release conditions for prepared actions to the visible targets (Eckstein and Perrig, 2007, Kiesel et al., 2009, Kunde et al., 2003).

The influence of target set size on unconscious priming by unfamiliar verbal stimuli has been assumed to depend on expectation effects (Kiesel et al., 2006). When target set size was large (40 targets) so that a variety of words from different semantic categories was expected, unfamiliar word primes that were not part of the target set elicited priming effects. However, when target set size was small (four targets) so that attention could be focused on a narrow set of stimuli, unfamiliar word primes did not produce subliminal priming. These findings demonstrate that stimulus expectations establish an attentional set that sensitizes the corresponding processing pathways for unconscious stimuli. Target set size might thus have moderating influences on priming by unfamiliar masked primes also in the domain of evaluative priming. As target set size is frequently small in evaluative priming experiments (10 targets or less), this would explain why masked evaluative priming by unfamiliar primes is not consistently found (Wentura & Degner, 2010). Interestingly, the only study, which observed small, but robust masked evaluative priming effects by unfamiliar primes across experiments used a relatively large target set (48 targets and more, Klauer et al., 2007).

The present study assessed boundary conditions for the emergence of masked evaluative priming effects with unfamiliar primes. As target set size has been identified as an important factor at least for masked non-emotional priming with unfamiliar primes (Kiesel et al., 2006), this factor was systematically varied across three experiments. However, in the study by Kiesel and colleagues, target set size was confounded with target repetition because stimuli from the larger target set were less frequently repeated as visible targets than stimuli from the small target set. This association of target set size and target repetition is typical for many studies because a few targets have to be repeated more often than many different targets to reach the same number of trials per condition (see for instance the meta-analysis by Van den Bussche, Van den Noortgate, & Reynvoet, 2009). However, it is possible that high repetition targets are less susceptible to influences from unfamiliar primes than low repetition targets because reactions are highly automatized (Posner & Snyder, 1975). The activated task associated with the target might mainly include responses based on acquired S–R associations (Damian, 2001) and not an evaluative semantic stimulus analysis, which is the requisite for evaluative priming by unfamiliar primes to occur. The present study therefore sought to vary target set size and frequency of target repetition independently from each other.

Finally, masked priming by unfamiliar primes might be different for pictures and words. However, earlier studies mainly used verbal material (e.g., Abrams et al., 2002, Greenwald et al., 1996, Klauer et al., 2003, Wentura and Degner, 2010). Pictures are assumed to have a privileged access to semantics because they are typically categorized faster than words (Glaser and Düngelhoff, 1984, Kiefer, 2001, Smith and Magee, 1980). Furthermore, evaluative information seems to be more readily available for pictures than for words (Spruyt et al., 2002). As priming with unfamiliar primes depends on at least a coarse evaluative semantic analysis, masked evaluative priming should be more readily generalize to unfamiliar primes for pictures than for words. We therefore systematically compared masked evaluative priming in the pictorial and verbal stimulus format.