Metacognition is a confound for consciousness–and what to do about that

Seminal work developing metacognitive measures of consciousness gave three reasons for treating task performance, understood in terms of d’, as a confound for consciousness: there is (i) consciousness without performance, and (ii) performance without consciousness; (iii) performance varies depending on specifics of the task without similar variation in consciousness (Lau 2008). This short talk argues that metacognition, understood in terms of meta-d’, is a confound for consciousness for the same reasons: (i) evidence of task-irrelevant consciousness (e.g. Sperling 1960) is equally evidence of metacognition-irrelevant consciousness; (ii) like high d’, high meta-d’ is found in non-conscious, artificial signal detection devices (Küppers et al. 2020); (iii) metacognition varies depending on whether it is probed as confidence or visibility, without similar variation in consciousness (Rausch & Zehetleitner 2016). Furthermore, with metacognition as with performance, addressing these problems by turning to subjective measures comes at the cost of equally pressing problems with bias (Irvine 2013). Conceptual connections between consciousness and self-awareness (Rosenthal 2005) do not make metacognition any less a confound, because the relevant forms of self-awareness are insufficient for metacognition. On the other hand, the talk also makes a positive suggestion about how to measure consciousness metacognitively. (i)-(iii) demonstrate at most that consciousness is non-identical with performance or metacognition, not that consciousness forms no constitutive part of performance or metacognition. As a result, (i)-(iii) do not justify treating consciousness and performance or metacognition as separate causal variables which may controlled independently of one another. Given this constitutive uncertainty, the influential method of matching performance while measuring consciousness metacognitively (Lau & Passingham 2006) should be abandoned, but metacognitive measures may be reliable if they are used as indirect, probabilistic measures of consciousness, along with a cluster of other such measures including performance.

Performance confounds and Nagel’s notion of the subjective

The problem of performance confounds is a central problem in the search for neural correlates of consciousness (NCCs). This short talk argues that current attempts to eliminate performance confounds fail because they ignore the fact that the problem of performance confounds is, fundamentally, a problem about the subjective and objective in Nagel’s sense (1974 / 1986). Current attempts to eliminate performance confounds proceed independently of concerns about the explanatory gap between neural and phenomenal conceptions of the mind (e.g. Lau 2008; Morales et al 2015). The assumption that work on NCCs should set aside such concerns (Crick and Koch 1998) is sustained by a pair of ideas: (1) we can identify NCCs by identifying neural conditions that share conscious experiences’ causal roles; (2) the explanatory gap does not make identifying conscious experiences’ causal roles especially problematic (Chalmers 1996; Shea 2012). Against (2), this talk argues that the explanatory gap concerns conscious experiences’ causal roles as much as their intrinsic character. In particular, the psychological roles distinctive of consciousness are subjective in Nagel’s sense: they are understood only by taking up the subject’s point of view, so we cannot identify them with neural mechanisms’ objective causal roles. Current attempts to eliminate performance confounds do not face up to this problem; they treat consciousness as a causal variable like any other. A more promising, interdisciplinary methodology is to identify causal constraints imposed by the subjective, normative psychological roles distinctive of consciousness, and assess which neurofunctional roles fulfil those constraints.

Perceptual confidence, presence to the mind, and disciplinary autonomy

The view that perceiving involves various degrees of confidence in hypotheses about one’s environment plays a significant role in several recent scientific, philosophical and interdisciplinary theories of perception and perceptual experience. This talk starts by defending a negative claim: such perceptual confidence is not fundamental to a good constitutive account of the subjective character of perceptual experience. The talk then exploits the example of perceptual confidence to develop a positive proposal about the autonomy of philosophical and scientific theories of perception from one another. In short, notions of representation in the philosophy and science of perception are often best understood in terms of different explanatory projects, such that philosophical and scientific commitments about perceptual representation are neither equivalent to, nor directly supported by, one another, even when they are applied to the same episodes of conscious perceptual experience.


Metacognitive Confounds for Conscious Perception, Action and Thought

Consciousness is often supposed to be connected a priori with awareness of one’s own mind: conscious episodes are episodes of which one is, or could be, aware. This is reflected in scientific and philosophical approaches to measuring consciousness which rely, implicitly or explicitly, on subjects’ capacity to assess and distinguish their own experiences. Here I argue that, even granting the relevant a priori connections, such measures suffer from confounds between consciousness and metacognition (cognition about one’s cognition). As a result, these measures underestimate the role of conscious episodes in our psychology, and overestimate the role of non-conscious processes. I make this case with respect to experimental measures of conscious perception and action, and philosophical measures of a conscious thought’s cognitive significance. I also suggest some related respects in which our approach to eliminating confounds in the measurement of consciousness should be more deeply interdisciplinary than it has been to date.

  • 14th June 2022, 2.30pm, at What is it to be Awake?, a multidisciplinary conference at King’s College London in collaboration with the Centre for Philosophy and Art.

Being Awake and Being Aware

I explore the hypothesis that the most basic respect in which being awake is significant is that being awake is being aware; in being awake, you are in touch with reality. I assess how this hypothesis is best understood, in light of some empirical cases which seem to challenge it: cases of the vegetative state in which subjects seem to be awake but not aware, and cases of sleepwalking in which subjects seem to be aware but not awake.

Performance Confounds and Epistemic Concepts in Consciousness Science

I explain and assess the problem of performance confounds, a methodological problem for consciousness science articulated by the neuroscientist Hakwan Lau. I argue that the fundamental problem Lau identifies is broader than he conceives it to be. In particular, it affects the metacognitive measures of consciousness which Lau exploits in order to eliminate performance confounds. The problem is also deep in ways that are neglected by recent philosophical discussions of methodological challenges in consciousness science. I explore its close connection with philosophical problems about the nature of epistemic justification and the ‘explanatory gap’ between neural and phenomenal conceptions of the mind. In light of this connection, I sketch two interdisciplinary approaches to overcoming the problem of performance confounds.

Waking, Knowing and Being Conscious

Being conscious, in the sense in which this state is associated with being awake as opposed to dreaming or sleepwalking, has a distinctive experiential character and epistemic role. The former is reflected in the experience of waking up, the latter in traditional problems about perceptual knowledge. I outline a conception of being wakefully conscious which identifies this state in terms of its role in explaining knowledge about one’s environment and oneself. I suggest that this dual epistemic role may be grounded, in part, in the control of attention. I argue that this conception has some advantages over Matthew Soteriou’s (2019) account of the state in question in terms of a temporal point of view. These advantages are brought out by examining the experience of waking up, a traditional problem about perceptual knowledge, and folk attitudes to sleepwalking and infant consciousness.

Do you need every part of your brain? Would you be the same person if parts were removed? Some patients with epilepsy undergo neurosurgery to remove the specific area of their brain thought to be causing their seizures. In this panel discussion, a patient will share how this operation affected their personality and sense of identity, and leading figures in neurology and philosophy will discuss what these experiences tell us about the relationship between the brain and the mind.

Visual Experience as Mental Action

This talk draws on recent empirical work about attention and visual spatial resolution, together with William James’s discussion of conscious attention, to argue that visual experience is sometimes a form of mental action. I defend this conclusion against some objections to the understanding of attention on which it depends, and I sketch some implications concerning introspective knowledge of one’s visual experiences.

The Limits of Self-Knowledge and the Scope of Rational Agency

The capacity to know one’s own mind is sometimes explained by appeal to the idea that rational propositional attitudes are transparent to higher order reflection on them. Be that as it may, I argue that transparency provides evidence for a significant limit on our capacity to know about our own inarticulate, perception-based rational attitudes: one is often not in a position to know exactly which such attitudes one has. I explore some consequences concerning the scope of bodily and mental agency: details of one’s bodily movement which are often supposed to be non-intentional may in fact be intentional actions; details of one’s propositional attitudes which are often supposed to be intentional actions are in fact non-intentional. The argument about agency takes knowledge first in a familiar way. By contrast, the argument from transparency takes non-epistemic phenomena first, and exploits them to explain and defend the conclusion that our capacity for self-knowledge is limited. I sketch some advantages of this pluralist approach to explanation and argument in knowledge-oriented philosophy of mind.

Brain Scans and Consciousness

Brain-imaging techniques are said to reveal that some so-called ‘vegetative state’ patients, who cannot communicate their experiences through speech or movement, are conscious of their surroundings. If true, this has enormous ethical implications for how we should treat patients who find themselves in this terrifying position, and evidence of this kind has in fact been cited in legal arguments about whether patients should be kept on life support. But is it true? To assess this, we will consider how longstanding philosophical problems about consciousness bear on the claim that consciousness is detectable through brain imaging. I’ll argue that we can achieve a satisfactory answer only by paying closer attention to the connection between consciousness and freedom than most scientists and philosophers in this area usually do.

Attention as Visual Determinacy: Merleau-Ponty, James and the cognitive neuroscience of attention

Rationality Without Self-Attribution

According to a grand philosophical tradition, the distinctive rational role of human consciousness turns on a conscious human’s capacity to attribute the contents of consciousness to herself, as contents of her psychological states. Against this I argue that, in a rational human, higher order reflection is less sensitive than first order belief and action to the contents of consciousness: what we can self-attribute, even under optimal conditions, is less fine-grained than the contents of consciousness which confer epistemic and practical justification. I explore some consequences for broadly internalist conceptions of the rational role of consciousness, concerning non-factive reasons and supposedly non-intentional details of our actions.

Rationality Without Self-Attribution

Partial Report Experiments Couldn’t Possibly Support Overflow

I argue that there are deep problems with appealing to partial report experiments to support claims about the independence of consciousness from cognition. The basic problem is that, due to the ‘explanatory gap’, our knowledge of the functional role of consciousness is limited to its role in thought and report. As a result, where partial report experiments show that more information is processed than is taken up in thought or reported, two interpretations are equally well supported: the contents of consciousness might (i) ‘overflow’ what’s reported, or they might (ii) be just as partial as what’s reported. The experiments provide no support for (i) over (ii). If introspection provided a form of access to consciousness that was independent of thought and report, introspection might—as some theorists claim—cast doubt on (ii). But there is good reason to doubt that introspection has this structure. I argue that this problem is not overcome by recent elaborations of the partial report paradigm, which show for example that the overflowing information exhibits Kanisza grouping (Vandenbroucke, Sligte, Fahrenfort, Ambroziak, & Lamme, PLoS ONE 2012). Such functional roles for the overflowing information do not support the claim that the overflowing information characterizes consciousness, since our knowledge of the functional role of consciousness is limited to its role in thought and report.

Consciousness is Not Awareness of Mental Qualities

  • 14th April 2016, at a British Academy funded workshop, ‘Discriminación perceptiva y consciencia’ (part of my collaboration with Miguel Sebastián), Instituto de Investigaciones Filosóficas, UNAM, Mexico City:

Partial Report, Consciousness and the Introspective Evidence

Does consciousness require cognition? Yes or no? Recent work has shown that either answer is consistent with the results of partial report experiments. Nonetheless, ongoing elaborations of the partial report paradigm are said to provide important defeasible evidence that consciousness is independent of cognition. I argue that this popular claim underestimates the challenge from alternative interpretations of the experiments. The experiments could in fact provide such evidence only given a controversial view about how we know our own conscious states – a view associated with the idea of ‘mental paint’. This view presupposes a contrast between consciousness and cognition, in terms of how we know them through introspection. What’s more, there are good introspective reasons to doubt that consciousness differs from cognition in the way presupposed. Partial report is the wrong place to look for evidence that consciousness is independent of cognition.

Self-Knowledge, Perceptual Evidence and the Significance of Consciousness

What makes consciousness a distinctive source of knowledge? The answer is often thought to lie in the fact that the contents of consciousness are available to higher order reflection. A related, traditional view holds that one’s perceptual evidence sometimes consists in one’s conscious perceptual states. Against these views I argue that, in a rational human, higher order reflection is less sensitive than first order belief to the contents of conscious perception. Therefore the epistemic significance of consciousness lies in its first order cognitive role, and one’s perceptual evidence consists in objects of first order perceptual experience, not in one’s perceptual states. Since this conclusion applies in cases of radical illusion, such as traditional sceptical scenarios, it also suggests that there is nonfactive perceptual evidence.

Self-Knowledge, Perceptual Evidence and the Significance of Consciousness

  • 1st December 2015, at the Mind, Metaphysics, and Psychology seminar, King’s College London (4.30-6.30pm):

Self-Knowledge, Perceptual Evidence and the Significance of Consciousness

  • 20th-21st November 2015, Conference on Consciousness and Accessibility, CNRS, Paris:

Access, Consciousness and Higher-Order Inexactness

The notion of cognitive access to a mental state M is ambiguous between (1) cognition which represents M, and (2) cognition which takes M as computational input. I’ll argue that these forms of cognition are dissociated in experiences of phenomenal continua, in that (1) is less sensitive than (2) to the contents of conscious perceptual states. I’ll then show how this dissociation can be leveraged into arguments to the best explanation, first against ‘higher-order’ theories according to which consciousness consists in (1), and secondly against standard formulations of the ‘global broadcasting’ theory according to which consciousness consists in a species of (2). The result is prima facie reason to accept that consciousness is independent of cognitive access.

Partial Report, Consciousness and Self-Knowledge

  • 12th May 2015, Workshop with Ned Block, Centre for Philosophical Psychology, University of Antwerp:

Determination is all you need

Action, Attention and Consciousness.

  • 26th January 2015, Isaiah Berlin Society at St Paul’s School in London:

Images In and Of the Brain.

Resolution and Experience

Because visual resolution is finite, visual knowledge is inexact. For instance you cannot know, just by looking at a figure, that it’s exactly circular. Williamson and others maintain, on phenomenological grounds, that visual appearances are nonetheless specific: there is a single exact shape which the figure visually appears to have; visual knowledge is inexact insofar as it allows a margin for error in the appearances. I argue that phenomenology does not support this view, given manifest facts about our experiences of phenomenal continua. What’s more, once we give up the view that visual appearances are specific, we can connect recent empirical findings about visual resolution with classic work on visual phenomenology.

  • 30th May 2014, at a Rethinking the Senses workshop on ‘Spatial interactions across vision, audition, and touch: searching for a taxonomy’, Institute of Philosophy, University of London:

Comments on Laurence Harris, ‘Why the vestibular system should be listed in the title of this workshop’.

  • 16th-17th May 2014, at Ned Block and His Critics, a workshop at the Institut Jean-Nicod, Paris:

Introspection, Overflow and Mental Paint

Ned Block champions the use of introspective evidence together with the formal data. Using this method, he defends two different claims about visual consciousness: (i) that visual consciousness is independent of cognitive access; (ii) that the phenomenal character and representational content of visual consciousness are distinct. On the face of it Block’s arguments for (i) and (ii) are independent of one another. I suggest that the argument for (i) in fact depends on (ii), and I explore some consequences for appeals to introspection in this area.

  • 10th May 2014, at the Attention and Perceptual Activity workshop, Consciousness & Self-Consciousness Research Centre, University of Warwick:

Active Attention & Visual Processing

Recent work on voluntary attention and visual spatial resolution supports the claim that visual experience is sometimes a form of basic action (maybe).

  • 27th July 2013, at the Barnard-Columbia Perception Workshop in New York:

The Manifest and the Determinable

  • 13th June 2013, at the University of Durham’s project ‘Philosophy and Psychology: Integrating Research Across Disciplines’ – Workshop on Attention and Consciousness:

Attending, Knowing and Detecting Signals

  • 2nd May 2013, at the Centre for Cognition Research, University of Reading:
    with Andrew Glennerster:

Could the mechanisms that underlie motor control be responsible for perception?

  • 23rd March 2013, at the Mind Network in London:

Visual experience and visible properties

  • 12th February 2013, at Reading Philosophy‘s Work in Progress series (Humss 25, 2.15-4pm):

Visual experience and visible properties

  • 2nd September 2012, at Perceptual Attention, Centre for Philosophical Psychology, University of Antwerp:

Conscious Attention: Focussing the Mind

Recent philosophical work tends to characterize conscious attention as a sui generis kind of conscious experience, over and above conscious perception and conscious cognition. I argue that the nature and functional role of conscious attention are better captured by a more parsimonious analysis, according to which conscious attention consists in variations in each of these more basic forms of consciousness. I defend this analysis by appeal to both conceptual and empirical considerations, and explain how it improves the prospects for intentionalism about conscious experience.

Symposium on ‘Consciousness & Cognitive Access’, with Ned Block, Jérôme Sackur and Ilja Sligte.

  • 2nd March 2012, at Cognitive Science Speaker Series, Philosophy Program, CUNY Graduate Center:

Conscious Attention: Focussing the Mind

Comments on John Schwenkler’s ‘Vision, self-location and the phenomenology of the point of view’. Available here.

  • 20th November 2011, at Attention, Expectation & Awareness in Visual Perception, Columbia University:

Comments on Marisa Carrasco’s ‘Attention alters perception’.

  • 12th September 2011, at New York University Consciousness Project:

Attention, visual knowledge and psychophysics

Attending to the things you see is a reliable way to come to know about their visible properties. Recent experiments generate a challenge to this epistemologically important idea, because they show that attention changes the way properties appear in visual experience. To meet the challenge, I argue that visual experience represents determinable properties. More generally, I argue that we need to recognize the role of determinable properties in visual experience, if we’re to understand what experiments tell us about conscious vision.

  • 17th November 2010, UC Berkeley Philosophy Colloquium:

Attention, visual knowledge and psychophysics

  • 9th June 2010, at Phenomenal Presence Conference, EXRE Project, University of Fribourg:

The visual presence of determinable properties

I explain and defend a theoretically useful way of understanding the idea that properties, such as the shapes and colours of things, are visually present to a subject of experience. I argue that the notion of the visual presence of a property is coherent, scientifically plausible, and appropriate for epistemology, provided that the properties visually present to a subject are determinable, rather than maximally determinate.

  • 14th May 2010, at London-Berkeley Graduate Philosophy Conference:

If “attention alters appearance”, do we know what the world is really like?

  • 24th April 2010, at Berkeley-Stanford-Davis Graduate Philosophy Conference:

If “attention alters appearance”, do we know what the world is really like?