Relative blindsight arises from a criterion confound in metacontrast masking: Implications for theories of consciousness

https://doi.org/10.1016/j.concog.2011.10.003Get rights and content

Abstract

Relative blindsight is said to occur when different levels of subjective awareness are obtained at equality of objective performance. Using metacontrast masking, Lau and Passingham (2006) reported relative blindsight in normal observers at the shorter of two stimulus-onset asynchronies (SOAs) between target and mask. Experiment 1 replicated the critical asymmetry in subjective awareness at equality of objective performance. We argue that this asymmetry cannot be regarded as evidence for relative blindsight because the observers’ responses were based on different attributes of the stimuli (criterion contents) at the two SOAs. With an invariant criterion content (Experiment 2), there was no asymmetry in subjective awareness across the two SOAs even though objective performance was the same. Experiment 3 examined the effect of criterion level on estimates of relative blindsight. Collectively, the present results question whether metacontrast masking is a suitable paradigm for establishing relative blindsight. Implications for theories of consciousness are discussed.

Highlights

► Uniformity of criterion content is essential to demonstrate relative blindsight (RBS). ► A recent report of RBS was marred by a confound due to a shift in criterion content. ► With criterion content held constant, there was no evidence of relative blindsight. ► The purported relative blindsight was also affected by criterion level.

Introduction

Studies of the neural correlates of consciousness often rely on two types of responses: objective performance on the task at hand and reports of subjective awareness1 (see Dehaene & Changeux, 2011 for a review). The finding that objective performance in a forced-choice task can be relatively high in the absence of subjective awareness is called blindsight as evidenced in patients with lesions in primary visual cortex (Weiskrantz, 1999; but see Overgaard, 2011, and Overgaard, Fehl, Mouridsen, Bergholt, & Cleeremans, 2008). A related concept called relative blindsight was introduced by Lau and Passingham (2006) to denote the finding that different levels of subjective awareness can be obtained in healthy observers even at equality of objective performance. Notably, relative blindsight is said not to involve any neurological deficit. Here we question the validity of the paradigm used in Lau and Passingham’s study, and hence the inferences drawn from the experimental outcomes.

Lau and Passingham (2006) employed a metacontrast masking paradigm to measure both objective performance and subjective awareness. In this paradigm, a briefly-displayed target is followed by a masking stimulus whose contours are closely adjacent to – but do not overlap with – the contours of the target. The temporal signature of metacontrast masking is a U-shaped function relating the accuracy of target identification to the stimulus-onset asynchrony (SOA) between the target and the mask (Breitmeyer and Öğmen, 2006, Di Lollo et al., 2004). At very short or very long SOAs, the target is identified easily. At intermediate SOAs, however, perception of the target is impaired, leading to the characteristic U-shaped function.

The stimuli used in Lau and Passingham’s (2006) study are illustrated in Fig. 1a, and the essential aspects of the results are illustrated in Fig. 2a, redrawn from Fig. 2 of Lau and Passingham. The observers were required to make two responses. First they identified the target shape (“Square or Diamond?”) and then they reported whether they had consciously perceived the target or had made a guess (“Seen or Guessed?”). The principal objective was to single out two experimental conditions in which the accuracy of target identification was the same but the reports of subjective awareness differed. This was achieved by varying the temporal relationship between the target and the mask over a wide range of SOAs. The critical finding concerned the results at two SOAs: 33 and 100 ms. As indicated by the two segmented ovals in Fig. 2a, the levels of objective performance were approximately the same at the two SOAs, but the level of subjective awareness was significantly lower at the shorter SOA. Lau and Passingham interpreted this asymmetry in subjective awareness as an instance of relative blindsight at the shorter SOA and concluded that different levels of visual awareness can be attained even at equality of objective performance.

This interpretation is questionable, however, on the grounds that the attributes of the display on which the observers based their responses were demonstrably not the same at SOAs of 33 and 100 ms. An example will clarify this issue. Illustrated in Fig. 1b and c is the appearance of the display at short and at long SOAs, respectively. In Fig. 1b the target and the mask are seen as forming a single, temporally-integrated image. Because of the relatively poor temporal resolution of the visual system – relative to, say, the auditory system – two brief stimuli displayed in rapid succession are perceived as a single image. This has been studied using a 5 × 5 dot-matrix with one dot missing at a random location (Di Lollo, 1980, Dixon and Di Lollo, 1994). To study temporal integration, the matrix is presented in two successive sets of 12 randomly-chosen dots each, separated by a variable SOA. At short SOAs (e.g., up to about 40 ms), the two percepts are temporally integrated, and the task of locating the missing dot is compellingly easy, with near-perfect performance. At longer SOAs (beyond about 100 ms), the two sets of dots are increasingly perceived as temporally distinct, with performance approaching chance level. Notably, Dixon and Di Lollo (1994) have shown that at the stimulus durations employed by Lau and Passingham (2006) two sequential stimuli are integrated maximally at the shorter SOA (33 ms), but to a much lesser extent at the longer SOA (100 ms).

A notable aspect in Fig. 1b is that at the shorter SOA the target was defined not by luminance contours but by the presence of four little black triangles that acted as inducers in a Kanizsa figure (Kanizsa, 1955). Thus, the observer’s response was based not on direct perception of the target shape but on an inference stemming from the geometrical arrangement of the four little triangles. To be clear about this, target identification at short SOAs would have been near chance level but for the presence of the four little triangles representing the portions of the background remaining visible through the mask. This was clearly not the case at longer SOAs at which the luminance-defined target shape was perceivable directly (Fig. 1c).

In more abstract terms, it can be said that the observers’ responses at the short and the long SOAs were based on different criterion contents. Kahneman (1968) introduced the term criterion content to denote the attributes of a stimulus on which an observer bases the perceptual decision. It is well-known that in masking experiments the criterion content can vary systematically with SOA (Kahneman, 1968, Sperling, 1965). With reference to Lau and Passingham’s (2006) study, the criterion content at the short SOAs was mainly the target shape inferred from the little triangles provided by the temporal integration of the target and the mask. In contrast, at the long SOAs the criterion content was the luminance-defined target shape itself. Because they were based on different criterion contents, estimates of both objective performance and subjective awareness were not entirely comparable across short and long SOAs.

How the different criterion contents might have influenced the asymmetry in subjective awareness at the two SOAs (given equality of objective performance) is not immediately obvious. A plausible account can be given on the twin assumptions that (a) objective performance depends on the availability of either of the two attributes of the display (little triangles or luminance-defined shape), and (b) subjective awareness (the observer’s willingness to admit to having seen the target’s shape rather than having guessed it) depends on the salience (i.e., prominence or discernibility) of that attribute. Based on the first assumption, the two attributes could mediate equal levels of performance at SOAs of 33 and 100 ms. Based on the second assumption, it is conceivable that the salience of the little triangles at the shorter SOA was less than the salience of the target shape at the longer SOA. This would reduce the observers’ confidence in their acknowledgment of actually having seen the target, i.e., in their level of meta-awareness.

We hasten to add that, while plausible, this account is clearly speculative. The fact remains, however, that different criterion contents mediated the responses at the short and at the long SOAs in Lau and Passingham’s (2006) experiment. What needs to be done is to determine whether the asymmetry in subjective awareness still occurs with a metacontrast paradigm in which performance is based on a single criterion content throughout the SOA domain. This was done in the present work by replicating the study of Lau and Passingham in Experiment 1 and then repeating the experiment with a metacontrast masking paradigm in which a single criterion content mediated the responses at all SOAs.

Section snippets

Participants

Thirty undergraduate students at Simon Fraser University participated for course credit. All reported normal or corrected-to normal vision and were naïve to the purpose of the experiment.

Stimuli and procedures

The stimuli, illustrated in Fig. 1a–c, and the procedures were the same as those used by Lau and Passingham (2006). The experiment was run in a dimly lit room. Observers sat at a distance of approximately 60 cm from a CRT computer screen (NEC AccuSync 120) refreshed at a rate of 140 Hz. At the beginning of each

Participants

Twenty-two observers were drawn from the same population as Experiment 1. None had participated in Experiment 1.

Stimuli and procedures

The stimuli are illustrated in the inset of Fig. 3a. The target, which consisted of two columns of three dots each, subtended 1.2° horizontally and 1° vertically. The mask, which consisted of two sets of two columns of dots that flanked the target columns (Kahneman, 1967; see inset in Fig. 3a), had the same height as the target and subtended 2.4° horizontally. Each dot subtended

Experiment 3

We have noted that the criterion content adopted when carrying out a perceptual task can affect the experimental outcome. The same can be said for the corresponding concept of criterion level. As Kahneman (1968) put it, “Specification of criterion level answers a quantitative question: How reluctant is the subject to give a particular response?” (p. 410). Given a forced-choice question, such as the “Seen or Guessed?” question in the present work, criterion level is said to be conservative when

General discussion

The principal objective of the present work was to examine the phenomenon of relative blindsight using a metacontrast masking paradigm. In Experiment 1 we replicated Lau and Passingham’s (2006) finding of an asymmetry in subjective awareness across two critical target-mask SOAs, given equality of objective performance. If that asymmetry is to be ascribed to relative blindsight, however, a key assumption must be met. Namely, that the criterion content is invariant across the two SOAs. That

Acknowledgments

This work was supported in part by an Alexander Graham Bell Canada Graduate Scholarship from the Natural Sciences and Engineering Research Council of Canada (NSERC) and by a British Columbia Pacific Century Graduate Scholarship to AJ, and by a Discovery Grant from NSERC to VDL.

References (38)

  • D.M. Rosenthal

    Consciousness and its function

    Neuropsychologia

    (2008)
  • K. Sandberg et al.

    Measuring consciousness: Is one measure better than the other?

    Consciousness and Cognition

    (2010)
  • S. Zeki

    The disunity of consciousness

    Trends in Cognitive Sciences

    (2003)
  • D.M. Armstrong

    What is consciousness?

  • B. Baars

    A cognitive theory of consciousness

    (1998)
  • N. Block

    Consciousness, accessibility and the mesh between psychology and neuroscience

    Behavioral & Brain Sciences

    (2007)
  • N. Block

    Comparing the major theories of consciousness

  • B.G. Breitmeyer et al.

    Visual masking: Time slices through conscious and unconscious vision

    (2006)
  • S. Dehaene

    Conscious and nonconscious processes. Distinct forms of evidence accumulation?

  • Cited by (0)

    View full text