Consciousness

Abstract

Consciousness is the property of a mental state that there is something it is like to be in it, the subjective experience that accompanies some but not all of the brain's information processing. This article separates the senses the word carries: the difference between being awake and being aware, the access consciousness that makes information available for reasoning and report, and the phenomenal consciousness of experience itself. It sets out the hard problem of why physical processing is accompanied by experience at all, surveys the global workspace and its ignition, and reviews the neural correlates of consciousness and the report paradigms that isolate them. It compares the leading theories, global workspace, integrated information, higher-order, and recurrent processing, on what each predicts and where each strains. Three interactive demonstrations probe bistable perception, the ignition threshold, and the attentional blink.

Keywords: consciousness, global workspace, neural correlates

Consciousness is at once the most familiar fact of a mental life and the hardest to explain. Every waking moment is given as experience, yet the concept the word picks out is not one thing. It runs together a state, being awake and responsive rather than in dreamless sleep or coma; a relation, being aware of some particular content; and a quality, the felt character of that content, the redness of red or the sting of a burn. Cognitive science has made progress by prising these apart and asking, of each, what in the brain it depends on and how it can be measured (Dehaene & Changeux, 2011). The sections below distinguish the varieties of consciousness, state the hard problem that sets the field's terms, separate access from phenomenal experience, develop the global workspace account and its signature of ignition, review the neural correlates and the methods that isolate them, and weigh the rival theories against one another.

What Consciousness Is

A first distinction organizes everything that follows: the level of consciousness and its contents are not the same thing. Level is a matter of state, the global condition of the brain that ranges from alert wakefulness down through drowsiness, sleep, and anesthesia to the minimally conscious state and coma. Content is what one is conscious of, a face, a tune, a decision, and it can be manipulated experimentally while the level is held fixed. The clinical dissociations make the separation vivid. A patient in the vegetative state cycles through sleep and waking with eyes open, yet shows no behavioral sign of awareness; Adrian Owen and colleagues found that some such patients, asked to imagine playing tennis, generated the same organized supplementary-motor activity in a scanner as healthy volunteers, evidence of preserved awareness that behavior alone had missed (Owen et al., 2006). Wakefulness without awareness, in the vegetative state, and awareness without full wakefulness, in a vivid dream, together show that the level and the contents of consciousness are doubly dissociable, and that neither can stand in for the other.

That the contents of consciousness are constructed rather than simply received is shown most economically by bistable figures. A Necker cube is a single unchanging line drawing, yet it is seen first with one face toward the viewer and then, after a few seconds, with the other; the sensory input is constant while the conscious percept flips. Because the stimulus does not change, whatever changes in the brain when the percept reverses is a correlate of the conscious experience itself rather than of the input, which is why bistable perception has become a workhorse for isolating the neural basis of consciousness (Koch et al., 2016). The demonstration below lets the reader drive the reversal and see the two readings of one figure.

The Hard Problem

The problems of consciousness are not all of one kind, and much confusion comes from treating them as though they were. David Chalmers drew the line that now frames the field, between the easy problems and the hard problem (Chalmers, 1995). The easy problems, easy only by comparison, are to explain the functions associated with consciousness: how the brain discriminates stimuli, integrates information, reports on its own states, focuses attention, and controls behavior. These are hard in practice but tractable in principle, because they ask for a mechanism, and a mechanism is the kind of thing cognitive science knows how to find. The hard problem is different in kind. It is to explain why the performance of any of these functions is accompanied by subjective experience at all, why there is something it is like to see red rather than merely to sort wavelengths and respond. One can imagine, without evident contradiction, a system that did everything a person does and yet had no inner life, and the persistence of that thought is what makes the hard problem hard.

Thomas Nagel had sharpened the target two decades earlier with a question that became a slogan for the whole difficulty: what is it like to be a bat (Nagel, 1974)? A bat navigates by echolocation, a sensory mode humans do not have, and Nagel's point was that no amount of objective, third-person knowledge of the bat's brain and behavior would tell one what the bat's experience is like from the inside. Experience has an essentially subjective character, tied to a point of view, and the objective methods of natural science are built precisely to abstract away from any particular point of view. That, Nagel argued, is the source of the explanatory gap: not that the facts of experience are supernatural, but that our present concepts give no grip on how to derive the subjective from the objective. Whether the gap is permanent or merely marks the limits of current theory remains the deepest open question in the study of consciousness, and it is the backdrop against which the empirical programs below define their more modest aims.

Access and Phenomenal Consciousness

If the hard problem sets a limit on explanation, the empirical study of consciousness proceeds by dividing the phenomenon into parts that can be handled separately. The most influential such division is Ned Block's, between access consciousness and phenomenal consciousness (Block, 1995). A state is access-conscious when its content is poised for use in reasoning, the rational control of action, and verbal report, that is, when the information is globally available to the rest of the mind. A state is phenomenally conscious when there is something it is like to be in it, when it has experiential character. In the ordinary case the two coincide: one sees the coffee cup, the seeing has a felt quality, and the content is available to guide the reach and to be reported. Block's claim is that they are nonetheless distinct concepts, and that they may come apart.

The proposed dissociation runs in a particular direction: phenomenal consciousness may overflow access. The evidence Block drew on is the partial-report paradigm, in which observers shown a brief grid of letters report that they saw all of them clearly but can name only the few that a post-stimulus cue directs them to, as though the experience was richer than the handful of items that made it into a reportable, access-conscious form. If that reading is correct, then report underestimates experience, and a purely report-based science of consciousness would systematically miss the phenomenal states that never reach access. The interpretation is contested, and some theorists hold that the apparent overflow is an illusion of unattended detail, but the distinction has proved indispensable for stating what is at issue: whether consciousness is to be identified with the global availability of information, as the workspace theories propose, or with something that can be present even when availability fails (Lamme, 2006).

Global Workspace Theory

The most developed cognitive account of conscious access is global workspace theory, proposed by Bernard Baars and given a neuronal form by Stanislas Dehaene and Jean-Pierre Changeux. Baars began from a functional metaphor: the mind is a vast collection of specialized, unconscious processors working in parallel, and consciousness is the momentary contents of a limited-capacity workspace that broadcasts information from any one processor to all the others (Baars, 2002). To be conscious of something, on this view, is for its representation to be made globally available, written to a shared blackboard that perception, memory, evaluation, and action-planning can all read. The theory explains at a stroke why conscious contents are unified, serial, and reportable, and why the capacity of consciousness is so narrow: a broadcast medium can carry only one message at a time (Baars, 2005).

The neuronal version locates the workspace in a network of long-range cortical neurons, dense in prefrontal and parietal cortex, whose connections can link distant processors into a temporary coalition (Dehaene et al., 1998). Its signature prediction is ignition. As the strength of a stimulus is raised, the workspace does not light up gradually; below a threshold the stimulus produces only local, feedforward sensory activity and is not reported, while above the threshold it triggers a sudden, self-amplifying wave of recurrent activity that engulfs the frontoparietal network and is reported as seen. The transition is sharp and nonlinear, an almost all-or-none jump rather than a smooth ramp, and it marks the boundary between what Dehaene and colleagues call subliminal, preconscious, and conscious processing (Dehaene et al., 2006). Figure 1 contrasts the two regimes, and the demonstration that follows lets the reader push a stimulus across the threshold and watch the reported rate climb.

Neural Correlates of Consciousness

Rather than confront the hard problem head-on, the dominant empirical program asks a narrower and more answerable question: what are the neural correlates of consciousness, the minimal neural mechanisms jointly sufficient for a specific conscious experience? Francis Crick and Christof Koch made the search for these correlates a respectable enterprise by framing it as a matter of correlation, deliberately setting the hard problem aside and looking for the neural events that track whether, and what, a person consciously perceives (Crick & Koch, 2003). The experimental logic is contrastive. Hold the physical stimulus constant and find a condition in which it is sometimes seen and sometimes not, as in bistable perception, masking, or the attentional blink; then whatever neural activity distinguishes the seen from the unseen trials is a candidate correlate of the experience itself, purged of the mere processing of the input (Koch et al., 2016).

A methodological worry haunts this design. If the seen and unseen trials are separated by asking for a report, the neural difference may reflect not consciousness but its downstream consequences, the attention, working-memory, and decision processes that generate the report. This is the motivation for no-report paradigms, which infer what the observer experiences from physiological markers such as eye movements or pupil size rather than a deliberate response, and which suggest that some of the frontal activity once attributed to consciousness belongs instead to reporting it. The debate has crystallized into a dispute over anatomy: whether the true correlates lie in a posterior hot zone of temporoparietal-occipital cortex, as the no-report evidence is taken to favor, or require the prefrontal involvement that the workspace theories predict. Christophe Sergent and colleagues sharpened the target on the temporal side, showing that during the attentional blink a second target is either fully reported or missed entirely, with the brain events underlying access to consciousness appearing as a sudden all-or-none transition rather than a graded fade (Sergent et al., 2005). The demonstration below reconstructs the blink and its telltale sparing at the shortest lag.

Theories Compared

Several theories now compete to say not just where consciousness correlates with brain activity but what consciousness is, and they make genuinely different predictions (Seth & Bayne, 2022). Global workspace theory, already described, identifies consciousness with the global broadcast of information across a frontoparietal network. Integrated information theory, developed by Giulio Tononi, starts instead from the felt properties of experience and asks what a physical substrate must have to realize them; its answer is integrated information, a mathematical measure, written phi, of how much a system's cause-and-effect structure is more than the sum of its parts, so that consciousness is identified with the quantity and shape of that integrated information and can in principle be present wherever the structure is, independent of report or behavior (Tononi et al., 2016). Higher-order theories locate consciousness not in a first-order representation of the world but in a further representation of that state, so that a perception becomes conscious only when the brain represents itself as being in it (Brown et al., 2019). Recurrent processing theory, argued by Victor Lamme, holds that consciousness arises as soon as feedforward sensory sweeps are met by local recurrent feedback within sensory cortex, before and independent of any global broadcast, which places the threshold for experience earlier and lower than the workspace account allows (Lamme, 2006). Table 1 sets the four side by side.

Note. The theories are not merely verbal variants; they disagree about whether prefrontal cortex is necessary, whether report is required, and whether a system with the right internal structure but no behavior could be conscious. Adversarial collaborations now test these predictions directly (Seth & Bayne, 2022).

Worked Example

The global workspace account predicts that conscious access is not graded but sharply nonlinear: as the strength of a stimulus rises, the probability that it is consciously reported should stay near zero, then climb steeply through a threshold, then saturate near one (Dehaene & Changeux, 2011). A convenient way to make that shape quantitative is the logistic function, P(report) equals 1 divided by (1 plus e raised to the power minus k times the quantity x minus x0), where x is stimulus strength, x0 is the threshold at which report is equally likely and unlikely, and k sets how steep the transition is. Take the illustrative values used in the demonstration, a threshold x0 of 50 and a steepness k of 0.18. At a stimulus strength of 40, ten units below threshold, the exponent is 0.18 times 10, or 1.8, so P equals 1 divided by (1 plus e to the 1.8), which is 1 divided by (1 plus 6.05), about 0.14: the stimulus is reported on roughly one trial in seven. At strength 50, exactly at threshold, the exponent is zero, e to the zero is 1, and P equals 1 divided by 2, or 0.50. At strength 60, ten units above threshold, the exponent is minus 1.8, so P equals 1 divided by (1 plus 0.17), about 0.86.

The lesson is in the steepness. A change of twenty units in the stimulus, from 40 to 60, drives the reported rate from 0.14 to 0.86, and most of that swing is packed into the narrow band around the threshold; the same twenty-unit change out in the tails, say from 10 to 30 or from 70 to 90, barely moves the probability at all. A nearly linear encoding at the sensory level is thereby transformed into an almost all-or-none report, which is exactly what the ignition of the workspace is meant to describe and what the trial-by-trial timing of the attentional blink reveals, a second target that is either seen whole or missed entirely rather than dimly half-seen (Sergent et al., 2005). The numbers here are illustrative, but the shape, a steep sigmoid whose midpoint marks a threshold and whose sharpness marks a near-discontinuity in awareness, is the quantitative fingerprint the theory predicts and the data repeatedly show.

Discussion

Consciousness matters to cognitive psychology first because it forces the clearest statement of what a psychological explanation can and cannot deliver: the easy problems ask for mechanisms and are being solved, while the hard problem asks why any mechanism is accompanied by experience and may lie beyond the reach of the contrastive method entirely (Chalmers, 1995). It matters second because the field has nonetheless become genuinely experimental, converting an old philosophical impasse into measurable questions by holding the stimulus fixed and asking what distinguishes a seen trial from an unseen one, and by taking seriously the difference between consciousness and the report that usually accompanies it (Crick & Koch, 2003; Koch et al., 2016). It matters third because the competing theories now make contact with data: global workspace, integrated information, higher-order, and recurrent processing disagree about whether prefrontal cortex is necessary and about whether report is required, and adversarial collaborations are beginning to adjudicate between them (Seth & Bayne, 2022). The open questions are correspondingly deep. Whether the neural correlates sit in a posterior hot zone or demand frontal involvement is unresolved; whether phenomenal experience overflows what can be accessed and reported is contested (Block, 1995); and whether integrated information, present in principle without behavior, could ever be confirmed as consciousness is unclear. What is settled is the reframing itself: that the varieties of consciousness can be separated, operationalized, and studied, even if the reason experience exists at all remains open.

The same contrastive logic reaches into the psychology of voluntary action. Recording the readiness potential, a slow negative shift in cortical activity that precedes movement, Benjamin Libet and colleagues found that this preparation begins several hundred milliseconds before the moment at which a person reports the conscious intention to act, a result that recast the old question of conscious will as an experimental one about the relative timing of brain activity and awareness (Libet et al., 1983). The readiness potential and the timing of conscious intention are examined in the companion article on intention.

Common Misconceptions

Being conscious is the same as being awake.

Level and content dissociate in both directions. A person in the vegetative state is awake, with open eyes and sleep-wake cycles, yet may show no awareness, while covert scanner responses reveal that some such patients are aware despite the absence of any behavioral sign (Owen et al., 2006). Conversely, a vivid dream is rich in conscious content during sleep. Wakefulness is a state of the whole brain; awareness is a relation to particular contents, and neither guarantees the other.

If a stimulus cannot be reported, it was not consciously experienced.

Report is a convenient index of consciousness but may not be identical to it. The distinction between phenomenal and access consciousness allows that experience can overflow what is globally available for report, and no-report paradigms were developed precisely because asking for a response contaminates the measure with attention and decision processes (Block, 1995; Lamme, 2006). Equating consciousness with reportability risks defining away exactly the states most in dispute.

Finding the neural correlates would solve the problem of consciousness.

The correlates are, by design, correlations. Crick and Koch framed the search that way on purpose, bracketing the question of why the correlated activity is accompanied by experience rather than answering it (Crick & Koch, 2003). Even a complete catalog of the neural events sufficient for each experience would leave the hard problem, the explanatory gap between physical process and felt quality, exactly where it was (Chalmers, 1995).

Glossary

Access consciousness.

A state whose content is globally available, poised for use in reasoning, memory, the control of action, and verbal report.

Attentional blink.

The brief failure to report a second target when it follows a first within about half a second, used to catch conscious access failing in time.

Bistable perception.

The spontaneous alternation between two interpretations of one unchanging stimulus, dissociating the conscious percept from the sensory input.

Broadcast.

The making of information from one processor available to many others; in workspace theory, the operation that constitutes conscious access.

Easy problems.

The tractable questions of how the brain performs the functions linked to consciousness, such as discrimination, integration, attention, and report.

Global workspace.

A limited-capacity system that integrates and broadcasts information from specialized processors, proposed as the functional basis of conscious access.

Hard problem.

The question of why physical information processing is accompanied by subjective experience at all, as distinct from explaining its functions.

Higher-order theory.

The view that a mental state is conscious when the brain forms a further representation of being in that state.

Ignition.

The sudden, self-amplifying, nonlinear spread of activity across the frontoparietal network that marks a stimulus crossing into conscious report.

Integrated information theory.

The theory that consciousness is identical to the quantity and structure of a system's integrated cause-and-effect power, measured as phi.

Neural correlates of consciousness.

The minimal neural events jointly sufficient for a specific conscious experience, isolated by contrasting seen with unseen trials.

No-report paradigm.

A method that infers conscious experience from physiological markers rather than a deliberate response, to separate consciousness from reporting it.

Phenomenal consciousness.

The felt, experiential character of a mental state, what it is like to be in it, independent of whether its content is available for report.

Recurrent processing.

Feedback activity that follows the initial feedforward sweep; on Lamme's theory, its local presence in sensory cortex is sufficient for experience.

Report paradigm.

The standard design that indexes consciousness by an observer's overt response, which may conflate experience with its cognitive consequences.

Vegetative state.

A condition of wakefulness without behavioral awareness following severe brain injury, in which covert awareness is sometimes preserved.

Key Researchers

Bernard J. Baars (b. 1946). Cognitive neuroscientist who originated global workspace theory, casting consciousness as the momentary broadcast of information across a limited-capacity workspace to a host of unconscious specialized processors.
ORCID - Google Scholar - Wikipedia

Ned Block (b. 1942). Philosopher at New York University who drew the distinction between access and phenomenal consciousness, arguing that experience may overflow what is globally available for report.
New York University - ORCID - Google Scholar - Wikipedia

David J. Chalmers (b. 1966). Philosopher at New York University who named the hard problem of consciousness and the explanatory gap, framing the distinction between the tractable functions and the residual question of why they are felt.
New York University - Google Scholar - Wikipedia

Stanislas Dehaene (b. 1965). Professor at the College de France whose global neuronal workspace model gave the theory a mechanism, identifying conscious access with a nonlinear ignition of frontoparietal cortex and a taxonomy of subliminal, preconscious, and conscious processing.
College de France - ORCID - Google Scholar - Wikipedia

Christof Koch (b. 1956). Neuroscientist at the Allen Institute who, with Francis Crick, made the neural correlates of consciousness a tractable experimental target and has since pressed the search toward a posterior cortical hot zone.
Allen Institute - ORCID - Google Scholar - Wikipedia

Giulio Tononi (University of Wisconsin-Madison). Psychiatrist and neuroscientist who developed integrated information theory, which starts from the properties of experience and identifies consciousness with a system's integrated cause-and-effect structure, phi.
University of Wisconsin-Madison - ORCID - Google Scholar - Wikipedia

Frequently Asked Questions

What is consciousness?
Consciousness is the subjective experience that accompanies some of the brain's information processing, the fact that there is something it is like to see, hear, or feel. The word covers several things at once: the level of wakefulness, the awareness of a particular content, and the felt quality of that content. Cognitive science studies these by keeping them apart and asking, of each, what neural activity it depends on and how it can be measured (Dehaene et al., 2006).

What is the hard problem of consciousness?
The hard problem, named by David Chalmers, is to explain why physical processing in the brain is accompanied by subjective experience at all. It is distinguished from the easy problems, which ask how the brain performs functions such as discrimination, attention, and report, and which are hard only in practice. One can conceive of a system that carried out all those functions with no inner life, and the difficulty of ruling that out is what makes the problem hard (Chalmers, 1995).

What is the difference between access and phenomenal consciousness?
Ned Block distinguished access consciousness, in which a state's content is globally available for reasoning and report, from phenomenal consciousness, the felt character of experience itself. In ordinary perception the two coincide, but Block argued they are separable and that phenomenal experience may overflow what can be accessed and reported. The distinction frames a central dispute over whether consciousness just is the global availability of information (Block, 1995).

What is global workspace theory?
Global workspace theory, proposed by Bernard Baars and given a neuronal form by Stanislas Dehaene, holds that consciousness is the broadcast of information across a limited-capacity workspace to the brain's many specialized processors. A stimulus becomes conscious when it ignites a self-amplifying wave of activity across a frontoparietal network, making its content widely available. The theory explains why conscious contents are unified, serial, reportable, and narrow in capacity (Baars, 2002; Dehaene et al., 1998).

What are the neural correlates of consciousness?
The neural correlates of consciousness are the minimal neural events sufficient for a specific conscious experience. Francis Crick and Christof Koch made the search tractable by looking for the activity that distinguishes trials on which a fixed stimulus is seen from trials on which it is missed. Whether those correlates lie in a posterior cortical zone or require prefrontal cortex is an active dispute, sharpened by paradigms that measure consciousness without asking for a report (Crick & Koch, 2003; Koch et al., 2016).

What is integrated information theory?
Integrated information theory, developed by Giulio Tononi, begins from the properties of experience and asks what a physical system must have to realize them. Its answer is integrated information, a measure written phi of how much a system's cause-and-effect structure exceeds the sum of its parts. Consciousness is identified with that integrated structure, so it can in principle be present in any system with the right organization, independent of behavior or report (Tononi et al., 2016).

Can consciousness be measured?
Consciousness can be indexed, if not measured directly, by contrasting conditions in which a fixed stimulus is consciously perceived with conditions in which it is not. Bistable figures, masking, and the attentional blink all create such contrasts, and the attentional blink in particular shows access to consciousness arriving in an all-or-none step rather than a gradual fade (Sergent et al., 2005). Covert scanner measures can even detect awareness in patients who cannot respond behaviorally (Owen et al., 2006).

Is consciousness the same as attention?
No. Attention selects information for deeper processing, whereas consciousness is the experience that may or may not follow. The two usually travel together, but they can dissociate: attended stimuli can go unreported and some processing of unattended stimuli proceeds without awareness. Recurrent processing theory even argues that experience can arise in sensory cortex before global attention engages it, placing the threshold for consciousness earlier than an attentional account would (Lamme, 2006).

References