The relation between consciousness and attention: An empirical study using the priming paradigm

https://doi.org/10.1016/j.concog.2009.12.019Get rights and content

Abstract

Dehaene et al., 2006, Koch and Tsuchiya, 2007 recently proposed taxonomies that distinguish between four processing states, based on bottom-up stimulus strength and top-down attentional amplification. The aim of the present study was to empirically test these processing states using the priming paradigm. Our results showed that attention (prime attended or not) and stimulus strength (prime presented subliminally or not) significantly modulated priming effects: either receiving top-down attention or possessing sufficient bottom-up strength was a prerequisite for a stimulus to elicit priming. When both top-down attention and sufficient bottom-up strength were present, the priming effect was boosted. The origins of the observed priming effects also varied between different processing states. We can conclude that our empirical study using the priming paradigm confirmed the presence of four processing states, which displayed a differential pattern of response priming effects and differential origins of the response priming effects.

Introduction

One way to enhance our understanding of the mechanisms underlying consciousness is to contrast conscious and non-conscious visual processing. Recently, theoretical taxonomies have been proposed to distinguish conscious and non-conscious processing states. Dehaene et al. (2006) for example proposed a taxonomy based on the global neuronal workspace hypothesis (Baars, 1989) that distinguishes between subliminal, preconscious and conscious processing. According to this global neuronal workspace theory (Dehaene & Naccache, 2001), bottom-up activation based on sufficient stimulus strength is necessary, but not sufficient for conscious access. More specifically, two thresholds in human information processing are assumed: a minimal duration to reach neural activity, and the significantly longer duration to reach the consciousness threshold. Subliminally presented or masked stimuli, which reach the first threshold but not the second one, can cause a burst of weak and short-lived activity, but will not reach a conscious state. Thus, insufficient bottom-up activation will prevent conscious access, indicating that bottom-up activation of a stimulus is necessary for conscious access. However, even when sufficient bottom-up strength is present, this does not necessarily mean that the consciousness threshold will be crossed. For example, in a study of the attentional blink, Sergent, Baillet, and Dehaene (2005) showed that stimuli displaying a strong and continuing activation were accompanied by a complete lack of conscious report. Another example is inattentional blindness, where a fully-visible, but unexpected object is not noticed, because attention is engaged on another task, event or object (e.g. Mack & Rock, 1998). Thus, bottom-up activation is not always sufficient for conscious access. What is additionally needed for conscious perception according to the global neuronal workspace theory is top-down attentional amplification: a stimulus must not only reach sufficient bottom-up activation, this activity must also be amplified and maintained over a minimal duration to reach the consciousness threshold. The latter is achieved by top-down attentional amplification. This top-down amplification causes the stimulus-evoked activation to extend to highly interconnected associative areas, allowing information to be held on-line for a long duration and allowing information to spread rapidly to many brain systems. Dehaene and colleagues argue that these two properties are typical of conscious perception.

Thus, according to the global neuronal workspace theory, both sufficient bottom-up strength and top-down attentional amplification are needed for conscious perception. Based on this, Dehaene et al. (2006) proposed a taxonomy that distinguishes three types of processing: (1) subliminal processing, where information inaccessibility is caused by insufficient bottom-up activation. When this subliminal information does not receive top-down attention, the activation will remain very weak. However, when attentional amplification is oriented towards it, the processing of the subliminal stimulus can be strengthened, although conscious access will never be reached. This is in line with the finding that primes made invisible by metacontrast masking remained unseen, even when they received attention (Kentridge, Nijboer, & Heywood, 2008); (2) preconscious processing, where information inaccessibility is caused by a lack of top-down attentional amplification. These stimuli do possess sufficient bottom-up strength, so they can potentially reach conscious access, but they are not consciously accessed yet, because they have not received attention yet, which is also reflected in an incapability to report these stimuli; (3) conscious processing, where the bottom-up stimulus strength is sufficient and attentional amplification is present, leading to conscious access. The former two states are both non-conscious, whereas the latter is a conscious state.

As Dehaene et al., 2006, Koch and Tsuchiya, 2007 proposed a two-by-two taxonomy and argued that top-down attention and consciousness are two distinct phenomena that can be dissociated. More specifically, they distinguish four states. At the one end of the spectrum, there is “attention with consciousness”, where subjects become conscious of objects or events that they attend to. At the other end, there is “no attention no consciousness”, which comprises objects or events that do not benefit from a top-down attentional bias and therefore remain unreported. In between, the two most controversial quadrants are situated. At the one hand, “attention without consciousness” where subjects attend to a location, but still fail to see one or more attributes of an object at that location. Masked priming can for example be situated in this quadrant: attended masked stimuli are able to elicit priming effects. At the other hand, “consciousness in the near absence of attention” where even though no or very little attention is dedicated to a certain object or event, subjects are still able to partially report it. For example, it has been shown that subjects are able to distinguish between male and female faces in the (unattended) periphery (Reddy, Reddy, & Koch, 2006). This taxonomy is in many aspects similar to the one proposed by Dehaene et al. (2006). However, unlike Dehaene et al., who state that in the preconscious condition, conscious reportability is not possible, these authors claim that conscious perception can occur without attention or at least in the near absence of attention.

The first aim of the present study was to empirically test these taxonomies and the distinct processing states using the priming paradigm. In the priming paradigm, the processing of a target stimulus is enhanced when it is preceded by a related prime stimulus, compared to when it is preceded by an unrelated prime stimulus. The priming effect is expressed by a faster and/or more accurate response to related prime–target pairs compared to unrelated pairs. In the present study, we will specifically examine response priming. For example, when subjects are asked to categorize target numbers between 1 and 9 as smaller or larger than 5 by pressing a left-hand button when the target is smaller than 5 and a right-hand button when it is larger than 5, they will respond faster to congruent trials (e.g. 1–4), where prime and target belong to the same semantic category and require the same response, compared to incongruent trials (e.g. 1–7), where prime and target belong to different semantic categories and require different responses (i.e. response congruency effect or RCE, e.g. Dehaene et al., 1998). We note that this kind of response priming has to be distinguished from true semantic priming, where priming effects are not confounded by response effects and stem solely from the semantic relatedness between prime and target. Using this paradigm and presenting primes and targets at two different places on the screen, we were able to manipulate both the bottom-up strength and the top-down attentional amplification of the prime stimuli. In line with the suggestions of Dehaene et al. (2006), the stimulus strength was manipulated by presenting the primes either subliminally (i.e. for a very short duration and masked) indicating that these primes had weak stimulus strength, or clearly visible (i.e. for a longer duration and unmasked) indicating that these primes had enough stimulus strength to reach the consciousness threshold. The top-down attentional amplification was manipulated by directing the subjects’ attention either towards the primes or away from the primes. In our design, a cue was presented that always validly indicated at which of the two places the target would appear on each trial. Thus, attention was always drawn to the place where the target would appear. The primes could either be presented at the same location as the target, indicating that they also received top-down attention, or at a different location, indicating they did not receive attention. We note that manipulating spatial attention and using an exogenous cue is only one possible way to manipulate attention, and different approaches (e.g. an endogenous and/or a temporal manipulation) can be used. Using these two manipulations, we empirically recreated the four processing states mentioned in the taxonomies of Dehaene et al., 2006, Koch and Tsuchiya, 2007, and we can hypothesize in which conditions priming effects should be observed: (1) for subliminal primes that do not receive attention (i.e. weak bottom-up activation and no top-down attentional amplification), no or only a very small amount of priming is expected; (2) for subliminal primes that do receive attention (i.e. weak bottom-up activation but top-down attentional amplification), we also expect, in line with previous demonstrations (e.g. Fabre et al., 2007, Lachter et al., 2004, Van den Bussche and Reynvoet, 2007), to observe limited but significant priming effects; (3) for clearly visible primes that do not receive attention (i.e. strong bottom-up activation but no top-down attentional amplification), we also expect to observe limited priming effects; (4) for clearly visible primes that also receive attention (i.e. strong bottom-up activation and top-down attentional amplification), we should obtain very strong priming effects. Thus, we hypothesize that attention will modulate the observed priming effects (receiving top-down attentional amplification will boost the observed priming effects), and stimulus strength will modulate the observed priming effects (possessing sufficient bottom-up strength will boost the observed priming effects).

The second aim of the present study was to ascertain how priming effects are caused in the four different conditions mentioned. It has, for example, been shown that when primes are made invisible by reducing their stimulus strength (e.g. by masking them), priming effects decayed when the prime–target SOA exceeded 200 ms (Kiefer & Spitzer, 2000), whereas these effects persisted with SOAs above 500 ms when the primes were made invisible by directing attention away from them (e.g. by using an attentional blink paradigm) (Rolke, Heil, Streb, & Henninghausen, 2001). Thus, these findings already indicate that the processing of the primes seems to differ in a condition where subliminal primes receive attention and in a condition where clearly visible primes do not receive attention. Finding qualitative differences between the mechanisms of the priming effects in the four conditions would provide further support for a distinction between the different processing states. More specifically, we will assess whether the observed priming effects are due to the fact that congruent trials are responded to faster (facilitative effect) or to the fact that incongruent trials are responded to slower (interference effect) or both and whether the cause of the priming effect differs between the different conditions.

Section snippets

Participants

Fifty-two students participated in both parts of the experiment as partial fulfillment of a course requirement (45 females; mean age = 18).

Procedure

As a first part of the experiment, all participants received the subliminal condition, where the prime had weak stimulus strength. Fig. 1 shows the sequence of a subliminal trial. First, a 480 ms mask existing of two hash marks (##) was shown at both locations. Then, the cue (++) was presented for 120 ms at one location followed by a 27 ms mask (##). Next, a

Results

Wrong responses (on average 4.9%) were discarded for RT analyses. We first report the results for the novel prime trials, since it has been argued that the influence of automatic stimulus–response (S–R) effects is reduced when using novel primes, and observed priming effects are therefore more likely to stem from semantic analysis of the primes (e.g. Damian, 2001). However, to ensure that the results for the repeated prime trials are nevertheless similar, we also report them in the second part.

Discussion

The aim of the present study was twofold. First, we wanted to empirically test the distinct processing states as defined theoretically in the taxonomies proposed for example by Dehaene et al., 2006, Koch and Tsuchiya, 2007 using the priming paradigm. By manipulating bottom-up stimulus strength and top-down attention, we reconstructed the different processing states and we were able to examine in which conditions priming effects would emerge. Second, our study wanted to unravel the cause of the

References (25)

  • M.F. Damian

    Congruity effects evoked by subliminally presented primes: Automaticity rather than semantic processing

    Journal of Experimental Psychology: Human Perception and Performance

    (2001)
  • S. Dehaene

    Conscious and nonconscious processes. Distinct forms of evidence accumulation?

  • Cited by (0)

    1

    Postdoctoral researcher of the Research Foundation – Flanders, Belgium.

    View full text