The Entwined Mysteries of Anesthesia and Consciousness

THE mechanism by which general anesthetics prevent consciousness remains unknown largely because the mechanism by which brain physiology produces consciousness is unexplained. But the two mysteries seem to share a critical feature—both consciousness and actions of anesthetic gases are mediated through extremely weak London forces (a type of van der Waals force) acting in hydrophobic pockets within dendritic proteins arrayed in synchronized brain systems. Unraveling this common thread may reveal not only how anesthetics act, but also why we are conscious in the first place.What is anesthesia? Anesthesia provides immobility, amnesia, and loss of conscious awareness, although the latter—loss of consciousness—is often omitted from operational definitions.1In recent years, putative sites of anesthetic action for immobility (spinal cord2), amnesia (dorsolateral prefrontal cortex,3amygdala4), and loss of consciousness (networks involving thalamocortical and intracortical—corticocortical—loops, prefrontal cortex, and other areas5,6) have been discriminated both anatomically and in terms of sensitivity to anesthetics. Immobility is least sensitive to anesthetics, followed by loss of consciousness and then amnesia, which is most anesthetic sensitive.7(Implicit memory may occur without consciousness or movement, but at light levels of anesthetic.8) Therefore, lack of movement—even though mediated by spinal cord rather than brain—in the absence of muscle relaxants is a good indicator of both loss of consciousness and amnesia. Autonomic responses are even less anesthetic sensitive than immobility9and, in the absence of autonomic-blocking drugs, are thus useful (although not perfectly reliable) early warning indicators of changes in anesthetic depth.What is consciousness? Unlike other receptor-mediated pharmacologic targets, consciousness is ill-defined, cannot be measured, and generates heated debate about its very nature. Indeed, except for the "dark age" of behaviorism in psychology during most of the 20th century (in which consciousness was, almost literally, a dirty word), conscious awareness has been a prominent mystery in science and philosophy.10However, many articles promising to discuss consciousness avoid the issue, e.g. , using bait-and-switch techniques to describe memory, learning, sleep, or other related activities. Others deconstruct consciousness into a group of cognitive functions so that the essential feature—conscious awareness—gets lost in the shuffle.11In this article, consciousness will be considered equivalent to even minimal awareness, the ineffable phenomenon of pure subjective experience—our "inner life." Thus, conscious awareness can exist irrespective of memory, cognition, or organizational sophistication (e.g. , reflective self-consciousness, higher-order thought, human—as opposed to animal—consciousness12). These more complex levels, although difficult to explain, are relatively straightforward compared with the issue of why or how even a slight glimmer of any form of conscious experience occurs at all.Anesthesia offers a unique and profound opportunity to understand consciousness because it is relatively selective—many brain activities (e.g. , evoked potentials, slower electroencephalography, and autonomic drives) continue during anesthesia while conscious awareness disappears. Thus, details of anesthetic mechanism may illuminate how the brain specifically produces consciousness and vice versa . This article reviews what is known about mechanisms of consciousness and anesthesia, finding that the "fine grain" of neuronal activities supporting consciousness and the molecular actions of anesthetic gases are one and the same—van der Waals London forces acting in hydrophobic pockets of coherently synchronized dendritic brain proteins. London forces are not chemical bonds but weak quantum interactions (in this regard, anesthetic gases differ in their actions from all other pharmacologic agents). Thus, the relative selectivity of anesthetic gases implies that the quantum nature of London forces may play an essential role in brain function leading to consciousness.Because consciousness is not directly measurable or observable, we begin with brain functional organization, systems, and activities known to correlate with consciousness.The particular brain systems and their functional activities related to consciousness are known as the neural correlate of consciousness (NCC). Depending on the level of detailed description, the NCC can be identified without necessarily addressing how consciousness is produced within or by the NCC.Functional frameworks for consciousness stem metaphorically from 17th century French philosopher Rene Descartes'"Theater of Consciousness" (hence "Cartesian theater"). Cognitive scientist Bernard Baars described this idea: "… consciousness acts as a 'bright spot' on the stage, directed there by the selective 'spotlight' of attention …"13Outside the spotlight are vast unconscious contents. But who or what is the audience, and who or what directs the spotlight? Despite these obvious problems, the theater metaphor has proven useful.In the 1970s, artificial intelligence computer models of brain function used a virtual "blackboard" on which specialized processors and knowledge sources could post their hypotheses about particular stimuli and then "vote" on which one was best. In the early 1980s, brain theorists combined this notion with the Cartesian theater metaphor and anatomical evidence about consciousness, resulting in "global workspace" theory.14The stage, blackboard, or workspace is associated with widely distributed ("global") corticocortical neural networks and (in some versions) thalamocortical networks representing perceptual systems and memory. Particular content "on the stage" or workspace is spotlighted or chosen by attentional focusing via "bottom-up" thalamic and limbic inputs and "top-down" executive action from prefrontal cortex. Spotlighted networks become the NCC, which continually changes with dynamically shifting, temporary alliances/networks of neurons. Thus, global workspace models demonstrate a dynamical, functional architecture for the NCC.On the other hand, consciousness may apparently occur in neural networks within localized, selected brain regions. Excessive activity in any feature-selective region may be sufficient on its own for that feature to enter consciousness. Thus, activity in cortical visual "color" area V4 alone can result in the visual experience of color.15Other brain regions have been suggested as NCC candidates, e.g. , the brainstem and limbic system in Antonio Damasio's and Jaak Panksepp's (separate) views of emotional "core consciousness."16So theoretically, consciousness can occur in what may be termed a global workspace (e.g. , for general surroundings, planning and processing—corticocortical and thalamocortical networks) but can also arise in more localized and perhaps separate regions, e.g. , overwhelming colors in a sunset (area V4), profound emotional feelings (brainstem, limbic system). The best scientific evidence for the NCC comes from brain imaging and electrophysiologic monitoring with loss of consciousness due to induction of general anesthesia.Functional brain imaging techniques (positron emission tomography and functional magnetic resonance imaging) show that anesthetic induction/loss of consciousness correlates with reduced metabolic and blood flow activity in brainstem, thalamus, and various regions of cortex, including thalamocortical and corticocortical networks.5However, the metabolic and hemodynamic decreases are delayed secondary effects of loss of consciousness rather than its cause. Electrophysiology provides a better correlate.Electrophysiologic brain monitors used in anesthesiology (e.g. , BIS Monitor®, Aspect Medical Systems, Inc., Newton, MA; Patient State Analyzer, Physiometrix, Inc., N. Billerica, MA; Narcotrend, MonitorTechnik, Bad Bramstedt, Germany) provide reasonably accurate correlates of anesthetic depth and presence or absence of consciousness.17They rely on spectral analyses and measures of synchrony in the electroencephalogram, particularly γ synchrony: approximately 30–70 Hz or higher, in various brain regions. Similar devices measure entropy in the electroencephalogram, or auditory-evoked γ synchrony.18,19A comprehensive electroencephalographic analysis of anesthetic-induced loss of consciousness was conducted by John and Prichep.6Using various anesthetic drugs and techniques, they found that loss of consciousness is a fairly abrupt transition (less than 20 ms) involving interruption of γ synchrony between frontal and posterior cortical regions. Similarly, Imas et al. 20showed that volatile anesthetics disrupt frontal–posterior cortical γ synchrony.Gamma electroencephalographic synchrony reflecting coherence among different brain regions is the best measurable correlate of consciousness, but is difficult to explain physiologically. Experiments show that γ synchrony is marked by "zero-phase-lag coherence,"21,22precisely synchronized voltage fluctuations occurring among varying regions of cortex and thalamus (and spinal cord23). Such precise coherence cannot be easily explained by neural networks involving thalamocortical pacing, recurrent feedback, reciprocal connections, propagating action potentials, and/or synaptic transmissions, which all convey significant delay or dephasing.21,24As will be discussed below, voltage potentials in cortical dendrites connected by electrotonic gap junctions apparently mediate γ synchrony, but even dendritic potentials introduce significant delay. Some collective field effect must be at play, and electromagnetic field–mediated synchrony is untenable.25,26Several experts conclude that a type of quantum field mediates γ synchrony and consciousness.21,27Gamma synchrony is also implicated in "binding" of conscious content. Various aspects of sensory percepts and volitional actions are processed in different cortical regions within single sensory modalities (e.g. , visual shape, motion, color), among different sensory modalities (e.g. , sight, touch and sound), and at different times, separated from each other by 80 ms or so.14But our conscious percepts are somehow "bound" into unified objects and simultaneous events. At any one time, there is only one NCC, and γ-synchronized activities in different brain regions are thought to tie together components of consciousness into unified entities.Anesthesiologist George Mashour28proposed that anesthesia prevents consciousness by unbinding neural activities, primarily by disrupting γ synchrony. This proposal equates binding, consciousness, and the transition from unconscious processes to consciousness through γ synchrony.The transition from unconscious processes to consciousness is a key question. Most authorities agree that only a small fraction of the brain's 100 billion or so neurons manifests the NCC at any one time, although many more are active.13But the same neurons and networks are not always "conscious"—signals and information do not travel to a particular part of the brain where consciousness happens. In the theater metaphor, the spotlight is constantly shifting, with represented content of particular γ-synchronized neural groups becoming conscious sequentially, selected by attentional processes and emotional saliency. (But why γ-synchronized neural activity has the subjective character of experiential awareness remains unexplained.)Therefore, consciousness seems to be a process, a sequence of transitions from unconscious activity to experienced content, e.g. , frames or scenes shifting up to approximately 40 times per second in γ synchrony.29But apparently not all brain activity can become conscious. For example, activity regulating autonomic functions almost never becomes conscious, and during anesthesia, nonconscious sensory evoked potentials and some electroencephalographic activity continue in the absence of consciousness. Because dreams (which can potentially become conscious, i.e. , when we awaken and remember them) probably do not occur during deep general anesthesia, it is possible that unconscious processes capable of becoming conscious are prevented by anesthetics. Therefore, anesthesia may simply inhibit the necessary antecedent to conscious awareness. But then the specific nature of the necessary antecedent—unconscious processes capable of becoming conscious—must be identified and distinguished from those nonconscious processes which lack the potential to become conscious and are resistant to anesthesia. One line of speculation holds that potentially conscious (unconscious, preconscious, dreaming) brain activities manifest as quantum information originating in intraprotein hydrophobic pockets in the NCC, also the precise site of anesthetic action.Brain activities of all types are considered to use networks of neurons as functional units. Two types of neural networks operate in the brain, one of which accounts for γ synchrony.Individual neurons are usually composed of multiple dendrites, one cell body (soma), and one axon (figs. 1A and B). The multiple dendrites receive and integrate many synaptic inputs from axons of other neurons in the form of chemical neurotransmitters, which bind on "postsynaptic" dendritic membrane receptors. Depending on the neurotransmitter, depolarizations (excitatory postsynaptic potentials EPSPs) and hyperpolarizations (inhibitory postsynaptic potentials IPSPs) are integrated, reach depolarization threshold, and trigger action potentials ("firings" or "spikes") through the neuron's single axon.In 1949, Canadian neuroscientist Donald Hebb30suggested that repeated activity of a given synapse decreased its threshold for subsequent firings, that synaptic "plasticity"—dynamical changes in synaptic strength—sculpted and reinforced specific paths through a network of neurons that would then be easily triggered by a partial stimulus. Evidence verified Hebb's idea that neural activity in the form of EPSPs leading to axonal spikes can follow paths of least resistance through low-threshold synapses, like water flowing downhill (fig. 1C). Such neural "assemblies," as Hebb termed them, could be ignited by particular inputs and remain active for hundreds of milliseconds, after which another related assembly would ignite, then another, and so on in a phase sequence. A commonly held view is that, at any one time, a single particular "Hebbian" neural assembly corresponds with the NCC.But axonal–dendritic Hebbian networks/assemblies are incompatible with γ synchrony electroencephalography, which is mediated by local field potentials or surface potentials generated by dendritic EPSP/IPSP activity (i.e. , dendrites may be synchronized, axonal spikes are not).31Although precise zero-phase-lag coherence remains unexplained, coordination of dendritic EPSPs/IPSPs leading to γ synchrony derives from a second type of neural network (fig. 1D)32: neurons connected by dendritic–dendritic gap junctions in conjunction with inhibitory synapses mediated by receptors for γ-aminobutyric acid (GABA).Gap junctions, or electrical synapses, are direct open windows between adjacent cells formed by paired collars consisting of proteins called connexin (fig. 2).33Membrane depolarizations travel bidirectionally across gap junctions so that neuronal processes connected by gap junctions are electrically coupled and depolarize synchronously.34In adult mammalian brain, gap junctions connect dendrites to other dendrites, as well as to axons and glia (and in some cases axons to axons).33Many cortical interneurons have dual synapses—their axons form inhibitory GABAergic synapses on a dendrite of another interneuron or pyramidal cell, while the same two cells share dendritic–dendritic gap junctions. Gap junctions open, close, and change location, controlled by the cytoskeleton within neuronal dendrites. Thus, dendritic–dendritic gap junction networks can adapt like Hebbian assemblies, extend widely through cortex, and account for γ synchrony.The blood oxygen level–dependent signal used in functional magnetic resonance imaging—assumed to represent neural metabolic activity related to cognition and consciousness—corresponds more closely with dendritic potentials than with axonal spikes.35Evidence seems to confirm that γ synchronized dendritic networks represent the NCC.Although axonal spikes are usually considered the primary currency of brain information, dendrites more actively process information and can, for example, change axonal spike threshold in a given neuron.36Some cortical neurons have no axons, and extensive dendritic activity may occur without causing spikes.37EPSPs below spike threshold (historically considered noise by many neuroscientists) oscillate coherently in the γ range across wide regions of brain.38Although it is widely assumed to be so, initiation of axonal spikes is not necessarily the raison d'être of dendrites. Neuroscientists Sir John Eccles,39Karl Pribram,40and others suggested that activities within dendritic–dendritic networks host consciousness.Nor is dendritic processing limited to membrane potentials. Many postsynaptic receptors (including excitatory glutamate and inhibitory GABABreceptors) are metabotropic, sending signals internally into the dendritic cytoskeleton, activating enzymes, and causing conformational signaling and ionic fluxes along actin filaments and microtubules. Accordingly, brain function leading to consciousness might be "… more refined on a higher temporal and smaller spatial scale."41The best electrophysiologic correlate of consciousness—γ synchrony—derives from coherent activities of dendritic postsynaptic receptors, with inhibitory GABAAreceptors in dual synapse interneurons (i.e. , with gap junctions) playing key roles. Although the collective mechanism leading to zero-phase-lag coherence remains unknown, γ-synchronized activities of dendritic proteins regulating EPSPs/IPSPs are apparently essential molecular-level correlates of consciousness: Coherent dynamics of dendritic proteins accounts for γ synchrony. (Anesthetic gases also act on γ-synchronized activities of dendritic proteins.)Proteins perform their functions by changing shape, or conformation, switching between energy minima. For example, ion channels open and close, receptors and enzymes grab ligands and substrates, etc. Many such changes, or prevention of change, occur in response to binding of a ligand at a site on the protein structurally removed (by up to 3 to 4 nm) from the conformational effect. To account for such indirect actions, Monot, Wyman, and Changeux proposed "allosteric" mechanisms in the early 1960s, superseding steric hindrance models, which required more direct contact between ligand and affected sites and activity.42In the absence of ligand, proteins undergo spontaneous dynamical transitions between two or more distinct conformations. According to allosteric theory, binding of a ligand in one particular conformational state stabilizes/activates that state and inhibits the dynamical switching, thereby preventing, depopulating, or deactivating alternative states. Allosteric theory also predicted (correctly) that many proteins such as receptors/ion channels are oligomeric complexes comprised of multiple subunits with rotational symmetry whose function depends on cooperative transitions among the subunits. Thus, ligand binding in one region of one subunit could affect the function of the entire complex.However, allosteric theory does not account for spontaneous conformational dynamics in the absence of ligand. Most functional protein transitions occur in the range from 10−6to 10−11s,43but their regulation remains unclear. Proteins have large energies with thousands of kilojoules per mole available from amino acid side group but proteins are only by approximately 40 protein is a among higher chemical and ionic bonds weak but forces (e.g. , van der Waals acting can the der Waals forces are among or and there are The first occurs between in like two each The second type of van der Waals is between a and a or with a The a temporary in the the and temporary then each The type of van der Waals is the London which occurs between two or (figs. and B). each temporary which then each other like on precise between and are extremely the close, forces groups of weak London forces acting are to protein of London forces which occur within some proteins in regions called hydrophobic pockets can such collective effects (fig. proteins are of amino which into by of amino acid groups by water in these groups (e.g. , of and as well as side groups of and each other by van der Waals water and within protein hydrophobic many two or more form groups or within which and protein pockets can be on the of approximately one the of single hydrophobic weak London to the in protein conformational dynamics by acting and of London interactions exist among the many and groups in the amino that a but only in proteins significant hydrophobic pockets are London forces and apparently to act to other London forces the to the London to protein (and thus via a of the extremely small of relative to that of and the conformational due to is A of a single a by only the of its the electrical on each is equivalent in to that on each acting London forces are thus to and protein by to a the and 1970s, proposed that of proteins in in a voltage (e.g. , membrane or would by available metabolic suggested such proteins would oscillate a quantum coherent state a Some evidence has also been implicated in binding and London forces are in protein proteins with significant hydrophobic dynamics and collective quantum coherence among distributed proteins. we in the brain proteins by hydrophobic London , γ synchronized dendritic the by which anesthetic gases act with relative selectivity to prevent consciousness. London forces in these proteins are the "fine is by anesthesia is often a with various and causing or to loss of consciousness, muscle amnesia, and drugs such as and by loss of consciousness, apparently through actions primarily on well as and electrical through the brain from can provide loss of consciousness: will such but first on anesthetic the of the 20th that anesthetic of a wide of with their in a that water and are also known as This was refined by and anesthetic by was found to correlate with in a particular range which some of in an et al. of anesthetic gases in various with in causing immobility in and and also found a slight of in an site of anesthetic the role of bonds in anesthetic effects in but and et al. that induction of a in an anesthetic by an electrical at or a hydrophobic site Accordingly, sites of anesthetic action are often to as , both and But as measure of anesthetic mediated in the spinal precise of sites loss of consciousness in the brain is unknown, and for example, correlate with to in effects among anesthetics and from occur due to in (e.g. , the a critical lack anesthetic effect and of anesthetic and binding the common and of anesthetic effect remains due to the largely nature of anesthetic of anesthetic gases is for by hydrophobic interactions and van der Waals London neurons in the NCC, where do anesthetics bind and act by hydrophobic interactions and van der Waals London in the Bernard to the anesthetic and found that the of within the cell was on this and other Bernard proposed that anesthesia from of is known that depends on of the protein actin and that anesthetic gases actin in dendritic in after and the that neuronal composed largely of it was assumed that anesthetics bind and act in regions of the that membrane proteins perform essential functions related to membrane attention to anesthetic effects on proteins. In the 1980s, and the apparently of and protein binding by a for anesthetic action in hydrophobic pockets of proteins of light emission from Although anesthetics do in membrane regions at and some for sites of anesthetic action are of evidence to hydrophobic pockets within various brain proteins as primary of anesthetic which hydrophobic pockets (i.e. , which do anesthetics bind and action is relatively selective—many nonconscious brain activities continue during relatively proteins have hydrophobic pockets large for anesthetics of neural Therefore, hydrophobic pockets in which anesthetics act may be to least to some the "fine grain" of the NCC, i.e. , within dendritic proteins γ of synaptic have that anesthetics act in dendrites (and inhibit γ with minimal effects on axonal action potentials and some effects continue to be membrane proteins are the