Neural Correlates of Consciousness Explained
The term “consciousness” often carries different definitions. Phenomenal consciousness is our personal, subjective experience of the present moment. Level of consciousness is the state of wakefulness assessed by various scales described in terms such as comatose, stuporous, obtunded, delirious, confused, and alert. In this review, we summarize the current understanding of consciousness including its neuroanatomic basis.
Tracking the correlations between brain processes and states of phenomenal consciousness, such as feelings of pain, seeings of blue, hearings of trumpet sounds, is the basic method of scientific consciousness research. Searching for such correlations with the help of modern brain imaging techniques has produced, since its inception in the 1990s, a body of remarkable results and a number of competing hypotheses regarding the neural correlates of conscious experience within different sensory modalities.
Each of these theories works towards understanding the neural correlates of consciousness-the neurobiological mechanisms necessary and sufficient to generate a conscious experience. Recent studies have revealed many of the requisite EEG, ERP, and fMRI signals to predict aspects of the conscious experience. We discuss major theories of consciousness, physical exam-based and electroencephalographic metrics used to stratify levels of consciousness, and tools used to shed light on the neural correlates of the conscious experience.
Major Theories of Consciousness
There are at least twenty-two supported neurobiological explanations for the basis of consciousness.
- Higher Order Theory (HOT): Claims that consciousness depends on meta-representations, representations that occur higher in a processing hierarchy.
- Global Workspace Theory (GWT): Postulates a centralized resource shared by multiple independent threads of processing.
- Memory Theory of Consciousness (MToC): Proposes that the phenomenonal consciousness evolved from, and functions as a part of, episodic memory and other explicit memory systems (sensory, working, and semantic memory).
This theory can explain long-standing puzzles in consciousness such as postdictive effects-the finding that conscious perception of a stimulus can be affected not only by conditions preceding a stimulus, but also by the conditions following a stimulus. The entire cerebral cortex and hippocampus function together to serially integrate and time stamp parallel unconscious brain processes to form a linear, coherent stream of conscious experiences. This theory proposes that the experience of consciousness is actually a memory, preceded temporally by unconscious processes of sensation, decisions, and/or actions.
Global Workspace Theory Diagram
Assessing Level of Consciousness
Level of consciousness, that is, the state of wakefulness, is commonly defined clinically through descriptors of psychomotor activity in response to stimulus. The Glasgow coma scale (GCS) is perhaps the most widely used clinical scale of consciousness, favored for its three metric simplicity (eye opening, verbal response, motor response). However, in recent years, it has drawn criticism for lack of content validity, standardization, and inter-examiner reliability.
A simpler measure favored in the intensive care unit (ICU) setting is the Richmond Agitation-Sedation Scale, which plots wakefulness on a single dimension from combative behavior to complete unresponsiveness. Another scale with high inter-examiner reliability, content validity, standardization, and prognostic capability is the Coma Recovery Scale Revised (CRS-S).
Patients may, however, be clinically unresponsive and maintain electroencephalographic (EEG) or functional MRI (fMRI) activity that suggests underlying consciousness. In a large, ICU-based study of brain activity, Claasen et al. used EEG patterns to detect occult brain activity in response to spoken motor commands in patients with acute brain injury, observing that about 15% of clinically unresponsive patients have a “conscious” EEG pattern in response to spoken commands.
EEG and Consciousness
EEG is the gold standard in determining the sleep-wake phase and is a useful tool in assessing the level of consciousness. The Bispectral Index (BIS) is used by anesthesiologists to assist in the titration of anesthetics. EEG frequency ranges and measures of connectivity have been considered as potential correlates of phenomenal consciousness (or at least aspects of it). The alpha frequency (8-12 hz), canonically considered to be an idling rhythm, has recently been thought to have a functional role in the inhibition of task-irrelevant information.
For example, Krasich et al. showed that about halfway between the maximum positive and negative peaks of occipital alpha activity, a stimulus is mostly likely to be consciously perceived. In the post-stimulus period, a drop in parietooccipital alpha power has been suggested as a marker of conscious perception. Event-related potentials (ERPs) are EEG changes time locked to sensory, motor, or cognitive events that provide a means to study electrophysiologic correlates of mental processes.
The conscious perception of a stimulus depends on a variety of factors including internal features of the stimulus itself (such as its salience). External factors that affect stimulus perception include observer state (such as attentional) and context (such as masking) which may affect perception preceding, simultaneously, or following the stimulus. Early ERP components, peaking within 100 ms after stimulus, are known as “sensory” or “exogenous,” as they depend mostly on the physical parameters of the stimulus itself.
In their historic work, Libet et al. reported in 1967 that early ERP sensory evoked potentials over the primary somatosensory cortex are not always sufficient to cause perceptual experience. The P100, a positive occipital ERP component occurring 100 ms after stimulus presentation, has been implicated as the earliest ERP involved in conscious perception and associated tasks. The N140 is a negative ERP peaking around 140 ms.
Multiple studies highlight the N140 as important for perception but there is debate as to whether the signal should be interpreted as awareness, perception, detection, or attention. Our review of early and middle ERP components suggests that they are mainly correlated with attention or unconscious sensory processes. The P300 is a posterior ERP observed 290-450 ms after a stimulus.
Historically, the P300, which includes components P3a (central maximum) and P3b (parietal maximum), has been thought to represent a gold-standard marker of conscious perception. The visual awareness negativity (VAN) is a negative difference wave, observed 200 ms after stimulus presentation, most prominently over occipitotemporal sites, which differentiates aware from unaware states. The VAN may be the earliest reliable electrophysiologic marker that correlates with aware perception after the presentation of a visual stimulus.
In the conscious patient, a TMS pulse sets off a complex chain of events as information reverberates through cortical brain regions, but in the unconscious patient, a similar pulse may induce only a short-lived local effect. Studying EEG recordings in patients with diminished levels of consciousness as determined by GCS, Toker et al. proposed that state of consciousness correlates with the proximity of slow cortical oscillatory potentials to a critical pattern described as the edge-of-chaos.
Similarly, using fMRI, Demetrezi et al. found that compared to healthy controls, those with decreased level of consciousness showed impaired baseline connectivity in bilateral executive control, default mode, and auditory networks. If we assume that consciousness is unitary, prompting us to look for a single neural correlate of consciousness, then it is difficult to make sense of these divergent results from various studies using different measures of brain physiology.
If, however, we take the approach suggested by the MToC-that there are various aspects of consciousness such as visual, auditory, emotional, phonological, motor, and more-then each study may reveal the neural correlate of consciousness for one aspect of consciousness but not others.
Neurological Disorders and Consciousness
Neurological disorders that disrupt the reticular activating system can affect the level of consciousness, whereas cortical disorders from seizures and migraines to strokes and dementia may disrupt phenomenal consciousness. Classically, this group included only disorders that directly impact the level of consciousness, such as TBI and cardiac arrest.
Epileptic seizures offer a window into the requisite anatomy and function for maintaining consciousness as different seizure subtypes affect consciousness differently. A generalized onset seizure includes bilateral motor involvement, loss of consciousness, and involvement of bilaterally distributed networks. The fact that complete loss of consciousness in seizure requires generalized involvement of both hemispheres argues strongly that level of consciousness is a diffuse cortical phenomenon, consistent with MToC but not HOT, GWT, IIT, or LRT. Focal seizures are subclassified as impaired awareness or unimpaired awareness.
Although migraine does not generally affect the level of consciousness, Budson et al. argued that migraine aura is, in fact, a focal disruption of phenomenal consciousness. Rarely, in the case of the basilar sub-type, migraine can affect the level of consciousness, causing confusion. More commonly, migraine aura results in the impairment of various aspects of phenomenal consciousness, such as visual associated with posterior onset of cortical spreading depression.
Stroke, discrete in border and stable over time, is the prototype brain lesion and offers a means to evaluate the neuroanatomic basis of consciousness. At onset, level of consciousness is impaired in about 36% of patients with large hemispheric infarcts and 14% of stroke patients as a whole. Clinical attention to level of consciousness is critical as early decreased consciousness may portend a 2.2-fold risk of in-hospital mortality.
Outside of infarcts that affect wide distributions of cortex (directly or indirectly through edema and hydrocephalus), the best described lesions affecting the level of consciousness are those within the reticular activating system (bi-thalamic and brainstem). Cortical strokes, especially when occurring bilaterally, may cause impairments to various aspects of phenomenal consciousness. For example, unilateral occipital infarcts impair hemifield visual consciousness and bilateral occipital infarcts may cause cortical blindness and a complete loss of conscious visual perception (despite some intact unconscious function).
However, no single or bilateral cortical infarct, unless involving most of the cortex, leads to permanent unconsciousness. If there were a “hot zone” of consciousness (as suggested by HOT, GWT, IIT, and LRT), there should be a cortical stroke syndrome that devastates consciousness; no such cortical lesion has been described.
Delirium is a syndrome characterized by an alteration in attention, awareness, and cognition. Budson et al. suggested that delirium is a state of individuals being awake but unconscious, noting that they do not appear to have conscious control of their own actions. The prevalence of delirium in hospitalized patients is high: up to 23% of hospitalized patients and 88% of palliative care inpatients.
Despite the high prevalence of delirium, the underlying pathophysiology of delirium is unclear. Recent studies have approached the neural substrate of delirium from a connectivity and functional state perspective. Fleishmann et al. suggested that theta band hyperconnectivity and alpha band dysconnectivity may underly delirium.
Although not generally considered a disorder of level of consciousness, dementia almost always affects aspects of the phenomenal conscious experience. Patients with Alzheimer’s disease demonstrated poor ratings in visual imagery, auditory imagery, and spatiotemporal specificity-all core components of the phenomenal conscious experience.
Corticobasal syndrome and alien-limb phenomenon present a disconnection between action and volitional, conscious control. In 2019, the first MRI-based group analysis of alien limb phenomenon in corticobasal degeneration revealed that CBS with alien-limb phenomenon can be differentiated from CBS patients without alien-limb syndrome based on structural differences in the cingulate gyrus as well as the frontomedian cortex, post central gyrus, and temporoparietooccipital regions.
Not all effects of dementia dampen consciousness. In some patients with FTD, a domain of consciousness can be augmented. The fascinating emergence of artistic creativity in some FTD patients has been described since 1998.
Lastly, the absence of level of conscious deficits in neurodegenerative diseases raises questions about the validity of some theories of consciousness. If meta-cognition in the frontal lobe is essential for consciousness, as stated in HOT, one would expect that patients with moderate FTD to become unconscious and lack any conscious awareness of their surroundings. Similarly, if the posterior brain and its connectivity is the center of consciousness as maintained by proponents of IIT and RPT, then patients with posterior cortical atrophy would be expected to have impairments beyond the visuospatial domain and become unconscious.
Brain Changes in Alzheimer's Disease
Causal vs. Non-Causal Accounts of Neural Correlates
Most contemporary attempts at accounting for brain-consciousness correlations are materialist. States of consciousness are not states of some mysterious immaterial substance; they are in some way anchored in detectable and measurable brain states. However, even within the materialist framework the correlation can be explained in various ways. Regular mind-brain correlations form an evidentiary basis which can be used in support of different metaphysical accounts.
The most straightforward approach is to try to explain the correlations in causal terms: NCCs are the causes of the states of consciousness. To “go from correlation to causation” is a move typical of the sciences and it might seem intuitively appealing to treat brain states as the causal sources of states of consciousness. No wonder, then, that we find advocates of the causal interpretation of NCCs. Still, we believe that this explanatory strategy is deeply problematic. It is far less plausible than the identity account of mind-brain correlation.
To see why, consider the picture put forward by the causal accounts. A neurophysiological process causes a phenomenal state of consciousness; therefore, it is different from that state, because causes and effects are always distinct. However, materialist principles dictate that every conscious state must be implemented materially, i.e., by some brain state(s). Since the conscious state is different from the neurophysiological processes that are causing it, it must, on pain of psycho-physical dualism, be implemented by a material process distinct from its neural cause. Thus we end up with two material processes involved in the production of the conscious mental state, not one.
The first material brain process would be, according to the causal approach, the cause of a conscious state. The second neural process then would be the implementation of the phenomenal conscious state P, though it would not be its cause. Which of these two neural processes is to be considered the real NCC? Clearly it is the second kind. The first kind of process undoubtedly regularly correlates with occurrences of the state of consciousness at hand, because it is its regular cause and regular causes do correlate with their regular effects. Let us call this kind of brain process the Causal Neural Correlate of Consciousness (CNCC).
Yet the CNCC is not what the cognitive neuroscientists empirically investigating consciousness are primarily searching for. They are searching for the brain processes of the second kind. Take as an example the firings of C-fibers, beloved by philosophers of mind. These firings are, in fact, the CNCCs of some kinds of pain sensations, because they are the causal antecedents of these sensations. Being a subset of sensory neurons projecting to the spinal cord, C-fibers are not even a part of the brain areas where these kinds of pain sensations arise. They are not its mechanism - that is, the activation of what is called by Swanson.
Causal neural correlates of consciousness precede in time the real or “true” NCCs. They are sometimes labeled the “upstream” correlates of NCC, or NCCpre. Apart from them, one can also distinguish the NCCco or “downstream” correlates of consciousness. Both the neural antecedents and consequents are confounders of the true NCCs. Thus the true NCC is simply what remains after we cut off the various confounders; and, we suggest, the true NCC is what is captured by the mind-brain theory of identity.
The preceding train of thought is, we believe, sufficient to demonstrate that the relation between states of phenomenal consciousness and brain states correlating with them cannot be causal, provided that a materialistic account of consciousness is correct. Empirical studies of consciousness, however, do not always distinguish between the causal and non-causal relations of phenomenal and neural states.
To avoid this confusion, it is profitable to distinguish between horizontal and vertical NCCs. Horizontal causal relations obtain between the proper NCC and its confounds, i.e., its causal antecedents and consequents. The notion of a vertical NCC concerns the very relation between a neural event and a phenomenal conscious state.
Horizontally, neural events N1, N2, and N3 are related causally. The true NCC (N2) is thus causally involved in a horizontal structure of neural events. This paper is about the vertical relations between NCCs and their corresponding states of consciousness. These relations cannot be causal because attempts to treat vertical relations in a causal manner lead to the confusions described above. To accept only non-causal vertical relations means that not only the simple and straightforward causality, but also the more nuanced causal relations such as emergence cannot be admitted as vertical brain-mind relations.
Identity, on the other hand, is not a causal relation, and thus can, in principle, serve as an appropriate candidate for the relation between conscious states and their NCCs. In the following section we shall consider how a theory of identity for the mind-brain relation should look.
Types of Neural Correlates
Scientists searching for NCCs are measuring particular instances of mind-brain correlations, but this is hardly the end-point of their inquiry. They aim to generalize their findings. They want to know not only the neural correlate of a particular phenomenal token, say, the sensation of a sour taste. To put forward an infinite disjunction of particular phenomenal-neural correlations does not sound an attractive research project!
The hope is that by accumulating a sufficient number of particular NCCs, the scientists will be in a position to generalize the data in some meaningful way. They will proceed by typing the phenomenal states into basic kinds and will try to assign to them broadly typed neurophysiological processes, carefully isolated during the experiments. This process needs to be repeatable both intrasubjectively and intersubjectively; hence it cannot stop at the level of individual tokens but must involve types. As Fink put it, it is because science aims at generality that “we aim at types-NCCs.”
This process can begin very crudely, but a rough and ready form of phenomenal and neural type-taxonomy is at least a prerequisite of any empirical science of consciousness. Even if the consciousness scientist sets herself relatively unambitious goals - say, she just wants to compile a list of some law-like bi-directional correlations between phenomenal and neural states - she will need to rely on some notion of phenomenal and neural type, for it will be types of states and processes that eventually have to appear in this list of systematic correlations.
A solution to the vertical mind-body-relation problem, though, need not inevitably appeal to types. The so-called weak (or “token-token”) theory of identity postulates identity only at the level of individual tokens of mental and neural events. However, as the advocates of the weak theory of identity admit, the token-token theory of identity does not permit systematization into law-like psycho-physical generalizations: individual instances of mental and neural events cannot be regimented into correlated type-sets.
The stronger, type-type theory of identity identifies types of conscious mental states with types of brain states and thus allows for systematic psycho-physical type-generalizations. Various forms of this stronger identity theory have been offered over the years. Debates within the philosophy of mind are quite extensive as regards the facets of the identity theory, its pros and cons, but very little attention has been given to the pivotal question of what the neurophysiological and phenomenal kinds are.
Most theorists participating in these debates use the notions of phenomenal and neurophysiological types only intuitively, without giving any explicit principles of individuation. This absence of a common understanding of what constitutes a type is striking, given the centrality of the notion of type within both philosophy and neuroscience. Is pain a type of phenomenal state ...