Elsevier

Consciousness and Cognition

Volume 20, Issue 3, September 2011, Pages 866-881
Consciousness and Cognition

PRP-paradigm provides evidence for a perceptual origin of the negative compatibility effect

https://doi.org/10.1016/j.concog.2010.09.014Get rights and content

Abstract

Visual stimuli (primes) that are made invisible by masking can affect motor responses to a subsequent target stimulus. When a prime is followed by a mask which is followed by a target stimulus, an inverse priming effect (or negative compatibility effect) has been found: Responses are slow and frequently incorrect when prime and target stimuli are congruent, but fast and accurate when prime and target stimuli are incongruent. To functionally localize the origins of inverse priming effects, we applied the psychological refractory period (PRP-) paradigm which distinguishes a perceptual level, a central bottleneck, and a level of motor execution. Two dual-task experiments were run with the PRP-paradigm to localize the inverse priming effect relative to the central bottleneck. Together, results of the Effect-Absorption and the Effect-Propagation Procedure suggest that inverse priming effects are generated by perceptual mechanisms. We suggest two perceptual mechanisms as the source of inverse priming effects.

Introduction

One approach to improve our understanding of consciousness focuses on those processes that do not depend on awareness. Masked priming is a crucial experimental paradigm in this research tradition which attracted much interest in recent years. In masked priming, the processing of a visual stimulus is influenced by a prime stimulus encountered before, even if this prime stimulus is rendered invisible. Typically, primes that share task-relevant features with the following target stimulus (congruent condition) lead to faster and more accurate responses while primes sharing features with the alternative target produce slower and less accurate responses (e.g., Mattler, 2003, Vorberg et al., 2003). In the literature, this effect has been called (positive) “response priming”, “motor priming” or “target-priming” effect.

However, Eimer and Schlaghecken (1998) demonstrated a reversal of priming effects with fast and accurate responses on incongruent trials and slow, erroneous responses on congruent trials when a separate mask followed the prime which in turn was followed by the target stimulus. This phenomenon is known as the “negative compatibility” effect or “inverse priming” effect. Since the pioneering work of Eimer and Schlaghecken, the inverse priming effect has been replicated in numerous studies (e.g., Boy et al., 2008, Eimer, 1999, Eimer and Schlaghecken, 2001, Eimer and Schlaghecken, 2002, Eimer and Schlaghecken, 2003, Jaśkowski and Przekoracka-Krawczyk, 2005, Klapp and Haas, 2005, Klapp and Hinkley, 2002, Lleras and Enns, 2004, Lleras and Enns, 2005, Lleras and Enns, 2006, Mattler, 2005, Mattler, 2006, Mattler, 2007, Schlaghecken and Eimer, 2000, Schlaghecken and Eimer, 2002, Schlaghecken and Eimer, 2004, Sumner, 2008, Verleger et al., 2004, Sumner, 2008). In one of the early studies, Schlaghecken and Eimer (2000) found that inverse priming depended on the stimulus onset asynchrony (SOA) between the mask and the target. The effect peaked at a mask-target SOA of about 100–150 ms (cf. Eimer and Schlaghecken, 2003, Mattler, 2007). The literature suggests that two conditions are necessary for the emergence of an inverse priming effect: The prime has to be followed by a separate (masking) stimulus (Boy et al., 2008, Lleras and Enns, 2006) and the following target stimulus has to be delayed by a sufficiently long SOA.

The origin of the inverse priming effect is a matter of debate since its original report. In a first account, the inverse priming effect was located in the motor system (Eimer and Schlaghecken, 2002, Klapp and Hinkley, 2002). According to the so-called “self-inhibition” account proposed by Schlaghecken and Eimer, 2002, Schlaghecken et al., 2006, Schlaghecken and Eimer, 2006), the mask removes the prime-induced perceptual evidence which in turn leads to self-inhibition of the primed motor response. Then, if a congruent target appears, the required response is still inhibited and therefore produces the observed performance costs. On the other hand, self-inhibition of the primed response gives the correct response a relative activation advantage if an incongruent target stimulus follows which results in the observed performance benefits. In the original version of the self-inhibition account the (complete) removal of perceptual evidence for the prime by the mask has been a necessary prerequisite to trigger the automatic self-inhibition mechanism. Due to evidence from a perceptual learning experiment which demonstrated unchanged priming effects despite a gradual increase of prime visibility, Schlaghecken, Blagrove, and Maylor (2008) recently retreated from the claim of a strict causal relationship between prime visibility and inverse priming effects.

The literature provides a number of findings that are consistent with a motor locus of inverse priming effects (e.g., Aron et al., 2003, Klapp and Hinkley, 2002, Schlaghecken et al., 2009, Schlaghecken et al., 2006, Seiss and Praamstra, 2004). For instance, Boy and Sumner (2010) examined the effect of learning a stimulus–response association (S–R association) and found a comparable time course with positive and inverse priming. Moreover, after a switch of S–R associations initially both effects switched too, consistent with a motor locus of inverse priming.

An alternative explanation of inverse priming was stimulated by the fact that early studies used masking stimuli that were either constructed by the superimposition of both prime alternatives or they included stimulus features that were part of both prime alternatives (so called relevant masks). The Object Updating account which has been proposed by Lleras and Enns (2004) provides an alternative explanation for inverse priming with such relevant masks. It is based on two major assumptions. First, it assumes that successively presented stimuli can be integrated into a common object representation which is iteratively updated. Second, this updating process emphasizes the representation of new stimulus features which gives prime-incongruent elements in the mask a relative saliency advantage and therefore facilitates the processing of incongruent target stimuli. In this way inverse priming effects grow out of the perceptual interaction between prime, mask, and target stimuli. In addition, however, Lleras and Enns emphasize that the results of perceptual processing are continuously linked to the motor system where they activate motor responses. Whereas these authors emphasize response related consequences of the perceptual interaction as the source of the effect, we note that the assumed perceptual interaction alone could also contribute to the effect on performance measures. According to this view, the duration of perceptual processing of the target stimulus is reduced on incongruent trials as compared to congruent trials because the perceptual interaction of prime and masking stimuli leads to a salience advantage of prime-incongruent stimulus elements in the mask.

The Object Updating account has been supported by the fact that inverse priming effects are typically much larger when relevant masks rather than irrelevant masks are used (e.g., a grid of vertical and horizontal lines for the case of arrow-shaped prime and target stimuli; Jaśkowski, 2007, Jaśkowski, 2008, Jaśkowski et al., 2008, Jaśkowski and Przekoracka-Krawczyk, 2005, Kiesel et al., 2008, Lleras and Enns, 2004, Lleras and Enns, 2005, Lleras and Enns, 2006, Schlaghecken and Eimer, 2006, Verleger et al., 2005, Verleger et al., 2004). Currently, it seems widely accepted that perceptual interactions between prime and masking stimuli contribute to the inverse priming effect in experiments with relevant masks (Jaśkowski and Przekoracka-Krawczyk, 2005, Klapp, 2005, Lleras and Enns, 2004, Lleras and Enns, 2005, Lleras and Enns, 2006, Mattler, 2005, Mattler, 2006, Mattler, 2007, Schlaghecken and Eimer, 2006, Schlaghecken et al., 2007, Sumner, 2007, Verleger et al., 2004). However, the Object Updating Theory does not cover all instances of inverse priming effects because the effect has also been found when the prime was followed by an irrelevant mask (Eimer and Schlaghecken, 2002, Klapp, 2005, Klapp and Haas, 2005, Lleras and Enns, 2005, Schlaghecken and Eimer, 2006, Schlaghecken et al., 2007, Sumner, 2008). Therefore, it is currently assumed by most researchers that some kind of suppressive mechanism contributes to inverse priming (Jaśkowski, 2007, Jaśkowski and Przekoracka-Krawczyk, 2005, Klapp, 2005, Lleras and Enns, 2006, Mattler, 2005, Schlaghecken and Eimer, 2006, Schlaghecken et al., 2007, Sumner, 2007). However, the exact nature of the suppressive mechanism and the conditions that trigger this mechanism are still unclear. Three different accounts have been proposed. All of them assume that the presentation of the stimulus which follows the prime triggers processes that lead to inverse priming. Beyond this commonality, each account makes different assumptions regarding the role of the inhibitory mechanism. Whereas Mattler (2005) suggested a mask triggered inhibitory mechanism without specifying its locus in the information processing system, there are two more specific proposals which locate the effect in the motor system: The “Onset Triggered Suppression-hypothesis” of Lleras and Enns (2006) and the “Mask Triggered Inhibition-hypothesis” of Jaśkowski and colleagues (Jaśkowski, 2007, Jaśkowski, 2008, Jaśkowski and Przekoracka-Krawczyk, 2005, Jaśkowski and Verleger, 2007).

To integrate the explanations of inverse priming with relevant and irrelevant masks three types of approaches can be distinguished in the literature: On the one hand, two completely separate mechanisms have been proposed to account for inverse priming in the case of relevant (e.g., Object Updating) and irrelevant masks (some suppressive mechanism, Klapp, 2005). For convenience we term this approach in the following ‘Separate-Mechanisms approach’. On the other hand, it has been proposed that the effect results from a single mechanism, which produces larger inhibitory effects in the case of relevant masks and smaller inhibitory effects in the case of irrelevant masks (Jaśkowski, 2008, Jaśkowski and Verleger, 2007). We term this the ‘Single-Mechanism approach’. Beyond this, a third approach can be distinguished which assumes two mechanisms that produce the effect together (e.g., Lleras and Enns, 2006, Schlaghecken et al., 2007). For instance, Object Updating might contribute to the inverse priming effect in the case of relevant masks, and some kind of suppressive mechanism might operate independently from the type of mask by generating a basic part of the inverse priming effect. We term this the ‘Coactive-Mechanisms approach’.

Motor self-inhibition (Eimer & Schlaghecken, 2002) as well as the mask triggered inhibition accounts (Jaśkowski and Przekoracka-Krawczyk, 2005, Lleras and Enns, 2006) of inverse priming ascribed the effect to the response system. Evidence in conflict with such motor system accounts comes from inverse cue-priming effects which seem to be generated by the same mechanism as inverse target priming (Mattler, 2006, Mattler, 2007). In these experiments the prime is followed by a mask and a task cue which indicates, for instance, whether participants have to classify the pitch or the instrument of an auditory sound target. Inverse cue-priming effects cannot be generated in the motor system because the prime is not associated with any specific motor response. Therefore, inverse cue-priming effects conflict with motor accounts of inverse priming, but they are consistent with perceptual accounts.

One previous attempt to localize inverse priming effects was made by Mattler (2006). Applying the inverse target-priming paradigm, this author examined the role of visual similarity between prime and target stimuli for the generation of inverse priming effects. Congruent primes and targets could be either perceptually similar or dissimilar and their effect on performance was compared with that of incongruent primes. If inverse priming effects consist of both, a perceptual and a motor-component, responses should be facilitated on incongruent as compared to congruent trials with perceptually dissimilar primes and targets due to the motor-component, and facilitated on perceptually similar trials as compared to congruent but perceptually dissimilar trials due to the perceptual component. In contrast, however, no difference has been found between congruent and incongruent trials, and only trials with perceptually similar primes and targets produced prolonged and erroneous responses. Comparable findings were made in corresponding cue-priming experiments (Mattler, 2006). These findings cast doubt on the assumption of a motor-component that contributes to inverse priming effects and suggest that the entire effect is generated at non-motor levels of processing.

In the current study we made a new attempt to localize inverse priming effects. We employed a mask that shared stimulus features with all stimulus alternatives. Therefore, our mask can be considered a relevant mask which could allow both mechanisms to contribute to inverse priming effects: A perceptual mechanism which is triggered by the interaction between the prime–mask–target sequence with relevant masks and a suppressive mechanism which is triggered by both relevant and irrelevant masks. We employed the PRP-paradigm which allows localizing inverse priming effects relative to a central processing bottleneck in order to distinguish the contributions of perceptual, central, and response-related components.

Originally the PRP-paradigm has been developed to study the temporal microstructure of processing in dual-task situations (e.g., Vince, 1948, Welford, 1952). However, in recent years the paradigm has been applied to localize the effects of different experimental variables (McCann and Johnston, 1992, McCann et al., 2000, Pashler, 1984, Pashler, 1989, Pashler and Johnston, 1998, Ruthruff et al., 1995, Van Selst and Jolicoeur, 1994). For this purpose, participants perform two choice-RT tasks in rapid succession. Here, we combined a visual priming task (to which we refer as the “shape task” in the following) with an auditory discrimination task, in which participants classified the pitch of a tone (the “tone task”).

The PRP-paradigm is based on two major assumptions. First, for each task the translation of sensory input to motor output is realized by information processing on three consecutive levels (compare McClelland, 1979, Sanders, 1980, Smith, 1968, Sternberg, 1969): The sensory stimuli are detected and identified on the perceptual level of processing. On a subsequent central level of processing, the sensory input is mapped onto the assigned motor response. When the appropriate response is selected, initiation and execution of this response is realized on a subsequent motor level. From this perspective, priming effects could arise from processing at any one (or more) of these processing levels. Second, it is assumed that processing of different information runs in parallel on perceptual and motor levels while the central processing level is conceptualized as a processing bottleneck in which information is processed successively. In consequence, information processing on a subsequent task has to wait if processing at the central level is blocked by a previous task. The assumption of a central bottleneck is derived from the PRP-effect that is typically found in dual-task experiments: When two stimuli, that are associated with two tasks, are presented with a short SOA, the response to the second stimulus is substantially prolonged compared to a condition in which the two stimuli follow each other with long SOA.

It is important to distinguish the present approach, which uses the PRP-paradigm to localize the effect of an experimental variable, from other research which examines the nature of the PRP-effect and the processing bottleneck. Although the later research produced a number of different accounts for the PRP-effect, most of the findings are consistent with the assumption of a central bottleneck that processes stimulus–response mappings successively one after another (for reviews see Pashler, 1998, Pashler and Johnston, 1998, Ulrich and Miller, 2008). For the present approach, it is crucial that a certain stage does not start processing the secondary task before it has finished processing of the primary task. Thus, as emphasized by Miller and Reynolds (2003) the approach depends on the relatively weak assumption that participants generally do process only one task at a time in the bottleneck instead of the strong assumption that participants must process only one task at a time. Supporting evidence for this weaker assumption has been provided for relatively unpracticed tasks with arbitrary stimulus–response mapping, which applies to the tasks in the experiments of our study (Pashler, 1998, Pashler and Johnston, 1998). Therefore, the present approach does not depend on the outcome of the current debate about whether sequential response selection occurs due to an insurmountable structural bottleneck (e.g., Pashler, 1984), or for other reasons like for instance specific instructions (e.g., Meyer and Kieras, 1997a, Meyer and Kieras, 1997b, Ulrich and Miller, 2008). It is sufficient that there is fairly strong evidence that people do process one task at a time when the instructions stress rapid production of the first response (Pashler, 1994, Pashler, 1998). In sum, it seems to be a reasonable working hypothesis that the PRP-paradigm provides good experimental methods to localize the effects of experimental variables (Miller & Reynolds, 2003).

Here we applied two complementary variants of the PRP-paradigm to localize priming effects relative to the central bottleneck. The two variants only differ with respect to the temporal order of the tasks: In the Effect-Absorption Procedure the shape task is performed after the tone task, in the Effect-Propagation Procedure this order is reversed.

The “Effect-Absorption” or “Locus-of-Slack” procedure serves to distinguish between priming effects that are due to (a) pre-bottleneck as opposed to (b) within- or post-bottleneck processing. As illustrated in Fig. 1, the critical manipulation is the SOA between the two tasks. With long SOA, both tasks are processed independently and a potential inverse priming effect should become manifest in response times (RTs) irrespective of its source (as in a single-task paradigm). With short SOA, however, processing at central levels is blocked by the preceding task (central bottleneck assumption). Therefore, the processing on the shape task has to wait after perceptual processing until the central level has finished processing of the first task. Consequently, if the priming effect results completely from different processing times on congruent and incongruent trials at perceptual levels prior to the bottleneck, this difference ought to be absorbed into the “cognitive slack” (Schweickert, 1978) and the effect should disappear it RTs with short SOA. In contrast, if the priming effect results only because congruency modulates processing time at the central bottleneck or at later levels of processing, priming effects in RTs should remain the same with long and short SOA because processing after the bottleneck is assumed to be independent of the secondary task. In sum, equal inverse priming effects at short and long SOA point to a post-perceptual locus of the effect whereas reduced inverse priming effects at short SOA point to a perceptual component of the effect. The contribution of a perceptual component would be reflected in how much of the inverse priming effect is abolished in the condition with short SOA.

The Effect-Propagation Procedure is less prominent but constitutes a natural complement to the Effect-Absorption Procedure because the temporal order of the tasks is simply reversed. Thus, in the present study, the shape task preceded the tone task. Based on the assumptions of the PRP-paradigm, different predictions follow for (a) the case that the inverse priming effect results from pre- or within-bottleneck processing as opposed to (b) the case that it results from post-bottleneck processing. Again, the SOA between the two tasks is varied, yielding the control condition with long SOA and the indicative condition with short SOA. With short SOA, processing of the secondary tone task has to wait until processing of the shape task at the central bottleneck is completed (central bottleneck assumption). If the processing time on congruent and incongruent trials in the shape task differs at perceptual levels or at the level of the bottleneck, the delay of further tone task processing will be directly affected by congruency. This case is depicted in Fig. 2. Because the congruency effect directly determines how long the processing of the tone is blocked at the bottleneck, the response to the tone will be faster on incongruent than on congruent trials. Thus, the entire priming effect should propagate to the RTs of the secondary task if the priming effect is completely generated before the bottleneck or at the level of the bottleneck. In contrast, if part of the inverse priming effect results from processing levels that follow the bottleneck level, this part should not be propagated to the tone task. Thus, if the entire inverse priming effect results from motor levels of processing, there should be no priming effect in RTs of the secondary tone task.

In the present study, we conducted two experiments to functionally localize the origins of the inverse target-priming effect and to gain new insights into the mechanisms underlying this effect and their relative contributions. For this purpose, we employed both variants of the PRP-paradigm. The inverse priming effect was examined with the Effect-Absorption Procedure in Experiment 1 to distinguish the contribution of perceptual and post-perceptual mechanisms. In Experiment 2 the inverse priming effect was examined with the Effect-Propagation Procedure do distinguish the possible contributions of motor and pre-motor components. In each experiment, prime recognition performance was measured in addition to dual-task performance. To anticipate our results, our findings point to a perceptual locus of the entire inverse target-priming effect.

Section snippets

Experiment 1: Effect-Absorption Procedure

The Effect-Absorption Procedure was used to localize the inverse target-priming effect prior to the central bottleneck as opposed to within- or post-bottleneck processing levels. If inverse target-priming is generated at least partially at perceptual levels, priming effects should be reduced in trials with short SOA. Otherwise, inverse priming effects are generated within or after the central bottleneck.

Experiment 2: Effect-Propagation Procedure

To verify the results from the Effect-Absorption Procedure, we employed the complementary variant of the PRP-paradigm, the Effect-Propagation Procedure in Experiment 2. In the Effect-Propagation Procedure the visual shape task had to be executed before the auditory task. This procedure allows to distinguish contributing mechanisms of the inverse priming effect at processing levels before or at the bottleneck as opposed to mechanisms at processing levels after the bottleneck. Again, the

General discussion

In an attempt to determine the origins of inverse target-priming effects, we presumed the assumptions of the PRP-paradigm and utilized two complementary variants of this paradigm. Results of the Effect-Absorption Procedure point to a perceptual locus of inverse priming effects, and results of the Effect-Propagation Procedure provide independent evidence for a locus at perceptual or central levels of processing. Together, both approaches suggest a perceptual origin of the inverse priming

Conclusion

In the present study, we examined inverse priming as an instance of the processing of visual stimuli which remain unconscious due to masking. With the PRP-paradigm we applied a new tool to locate the underlying mechanisms of the effect in an information processing framework. Locating the effect of stimuli which are not consciously perceived by participants contributes to our understanding of the source and function of conscious perception. The current results point to a perceptual source of the

Acknowledgments

This research was funded by the Deutsche Forschungsgemeinschaft Grants MA 2276/3-1 and MA 2276/3-2 awarded to Uwe Mattler.

References (67)

  • R. Ulrich et al.

    Response grouping in the psychological refractory period (PRP) paradigm: Models and contamination effects

    Cognitive Psychology

    (2008)
  • A. Aron et al.

    Inhibition of subliminally primed responses is mediated by the caudate and thalamus: Evidence from fMRI and Huntington’s disease

    Brain

    (2003)
  • F. Boy et al.

    Mask stimulus triggers inhibition in subliminal visuomotor priming

    Experimental Brain Research

    (2008)
  • F. Boy et al.

    Tight coupling between positive and reversed priming in the masked prime paradigm

    Journal of Experimental Psychology: Human Perception and Performance

    (2010)
  • M. Eimer et al.

    Effects of masked stimuli on motor activation: Behavioral and electrophysiological evidence

    Journal of Experimental Psychology: Human Perception and Performance

    (1998)
  • M. Eimer et al.

    Response facilitation and inhibition in manual, vocal, and oculomotor performance: Evidence for a modality-unspecific mechanism

    Journal of Motor Behavior

    (2001)
  • M. Eimer et al.

    Links between conscious awareness and response inhibition: Evidence from masked priming

    Psychonomic Bulletin and Review

    (2002)
  • M. Eimer et al.

    Locus of inhibition in the masked priming of response alternatives

    Journal of Motor Behaviour

    (2002)
  • D.E. Huber

    Immediate priming and cognitive aftereffects

    Journal of Experimental Psychology: General

    (2008)
  • P. Jaśkowski

    The effect of nonmasking distractors on the priming of motor responses

    Journal of Experimental Psychology: Human Perception and Performance

    (2007)
  • P. Jaśkowski et al.

    Mask- and distractor-triggered inhibitory processes in the priming of motor responses: An EEG study

    Psychophysiology

    (2008)
  • P. Jaśkowski et al.

    On the role of mask structure in subliminal priming

    Acta Neurobiologiae Experimentalis

    (2005)
  • P. Jaśkowski et al.

    What determines the direction of subliminal priming

    Advances in Cognitive Psychology

    (2007)
  • S.T. Klapp

    Two versions of the negative compatibility effect: Comment on Lleras and Enns (2004)

    Journal of Experimental Psychology: General

    (2005)
  • S.T. Klapp et al.

    Nonconscious influence of masked stimuli on response selection is limited to concrete stimulus–response associations

    Journal of Experimental Psychology: Human Perception and Performance

    (2005)
  • S.T. Klapp et al.

    The negative compatibility effect: Unconscious inhibition influences reaction time and response selection

    Journal of Experimental Psychology: General

    (2002)
  • A. Lingnau et al.

    The time course of response inhibition in masked priming

    Perception and Psychophysics

    (2005)
  • A. Lleras et al.

    Negative compatibility or object updating? A cautionary tale of mask-dependent priming

    Journal of Experimental Psychology: General

    (2004)
  • A. Lleras et al.

    Updating a cautionary tale of masked priming: Reply to Klapp (2005)

    Journal of Experimental Psychology: General

    (2005)
  • A. Lleras et al.

    How much like a target can a mask be? Geometric, spatial, and temporal similarity in priming: A reply to Schlaghecken and Eimer (2006)

    Journal of Experimental Psychology: General

    (2006)
  • G.R. Loftus et al.

    Using confidence intervals in within-subject designs

    Psychonomic Bulletin and Review

    (1994)
  • U. Mattler

    Priming of mental operations by masked stimuli

    Perception and Psychophysics

    (2003)
  • U. Mattler

    Inhibition and decay of motor and non-motor priming

    Perception and Psychophysics

    (2005)
  • Cited by (0)

    View full text