Consciousness, Neurobiology and Quantum Mechanics

Hameroff classifies all the mainstream approaches to consciousness as 'classical functionalism'. Functionalism takes no account of what the brain is made of or of anything finer grained than the level of neuron-to-neuron connections. It believes that these connections could be copied in another material such as silicon, and that the resulting construct would be conscious. However, Hameroff argues that although axonal spikes and synaptic connections clearly play a key role in information processing in the brain, they may not be the main currency of consciousness. Hameroff argues that quantum processing in microtubules within the dendrites and gap junctions between dendrites are the main currency of consciousness.

The main case against quantum processing in the brain has always been that any quantum coherence in the brain would decohere faster than the time taken for any useful biological process. Hameroff accepts that this is in principle a valid argument. However, Hameroff claims that the microtubules may be screened from their environment by a gelatinous non-liquid ordered state that arises in the neuronal interior.

A further objection to quantum processing is that even if it arose in one neuron, it would difficult for it to communicate across the brain. This is countered by the suggestion that there could be quantum tunneling at gap junctions between neurons. In recent years, gap junctions have been discovered to be more widespread in the brain than was previously thought. They are also correlated with the 40Hz gamma synchrony. This oscillation was at one time promoted by Crick and Koch as the most promising correlate of consciousness. However, the idea fell from favour with mainstream neuroscience, when it was discovered that the gamma synchrony correlated with dendritic activity rather than axonal spiking.

In general, Hameroff argues that the emerging evidence of neurobiology has moved in favour of the Orch OR model over the last decade, not withstanding the continued unpopularity of the theory. P Hameroff summarises his proposals in the early part of the chapter. He thinks that consciousness arises in the dendrites of neurons that are connected by gap junctions to form 'hyperneurons', and that these are related to the gamma synchrony. Axonal spikes and synapses are seen as making inputs to and receiving outputs from the microtubular process as part of an interactive systems.

Hameroff touches on the famous Libet experiments that demonstrated a 500ms timelag between a stimulus and the perception of it entering consciousness, although the subject is not aware of this time lag, as a result of a so-called backward referral in time. The mainstream has tended to favour an interpretation resembling the Dennett 'multiple drafts' concept, which would involve an after the event reconstruction of what had happened. Hameroff, however, thinks that the backward referral in time should be taken seriously. This was also the view of Roger Penrose, who suggested that backward referral might be indicative of quantum activity.

Hameroff points out that changes in dendrites can lead to increased synaptic activity. This is basic to ideas about learning, memory and neural correlates of consciousness. The changes in dendrites involve the number and arrangement of receptors and the arrangement of dendritic spines and dendrite-to-dendrite connections.  Axon potentials or spikes have been assumed to be the main basis of consciousness, but Hammerof suggests that there could be other candidates. Electrodes implanted into the brain detect mainly the activity of dendritic gap junctions plus inhibitory chemical synapses. Thus the detected synchrony derives from dendrites rather than axonal spikes.

The main function of dendrites is seen to be the handling of input signal into the neuron, which may eventually result in an axon spike. However, this is not the whole story, since many cortical neurons have dendrites but no axons. Here dendrites interact with other dendrites. Also there can be extensive dendritic activity with no spikes. The evidence suggests that there are complex logic functions in the dendrites, and these may oscillate over a wide area, while remaining below the axon spiking threshold. Many post-synaptic receptors send signals into the dendrite cytoskeleton.

Gamma synchronies, in the 30-70Hz range, have aroused interest as possible correlates of consciousness. Gray and Singer (1) found coherent gamma oscillations in the brain that were dependent on visual stimulation. It was suggested that this synchrony could solve the binding problem, which is the problem of how the different inputs into the brain are bound together into a single conscious experience. It was suggested that the synchrony relected the activity of a relevant assembly of neurons. Varela (2) noted that synchrony operated whenever the processing of spatially separated parts of the brain were brought together in consciousness. Gamma synchrony has been demonstrated across cortical areas, hemispheres and the sensory/motor modalities. The synchrony is involved in a range of brain activities including perception of sound, REM dream sleep, attention, working memory, face recognition and somatic perception. Also gamma decreases during general anesthesia and returns on waking from this. Hameroff regards gamma synchrony as the best overall correlate of consciousness.

He further addresses the question of how the gamma synchrony is mediated. There is coherence over large areas of the brain, sometimes including multiple cortical areas and both hemispheres of the brain, with zero or near zero phase lag. If the synchrony was based on the axon/synapse system a considerable lag would be expected. In fact, the lack of coherence between the synchrony and axonal spike activity has led to a reduction in the amount of mainstream attention paid to the gamma synchrony.

Hameroff points to gap junctions as an alternative to synapsses for connections between neurons. Neurons that are connected by gap junctions depolarise synchronously. Gap junctions play a more important role in the adult brain than was previously supposed. Numerous studies (3) show that gap junctions mediate the gamma synchrony. A neuron may have many gap junction connections but not all of them are necessarily open at the same time. The opening and closing of the junctions may be regulated by the microtubules. Hameroff suggests that cells connected by gap junctions may in fact constitute a cell assembly, with the added advantage of snchronous excitation. Cortical inhibitory neurons are heavily studded with gap junctions, possibly connecting each cell to 20 to 50 others (4). The axons of these neurons tend to form inhibitory GABA chemical synapses on the dendrites of other interneurons.

Hameroff moves on to discuss the role of the cytoskeleton, which is seen to determine the structure, growth and function of neurons. Actin is the main constituent of dendritic spines and is present throughout the neuronal interior. Actin can depolymerise into a dense meswork, and when this happens the interior of the cell is converted from an aqueous solution into a gelatinous state. Furthermore, when this happens the whole of the cytoskeleton forms a negatively charged matrix around which water molecules are bound into an ordered state (5). It is noted that the neurotransmitter glutamate binding to NMDA and AMPA receptors cause gel states in actin spines (6).

The cytoskeleton of the dendrites is distinct both from that found in cells outside the brain and from the cytoskeleton found in the axons of neurons. The microtubules in dendrites are shorter than those in axons and have mixed as opposed uniform polarity. This appears a sub-optimal arrangement from a normal structural point of view, and it is suggested that in conjunction with microtubule associated proteins (MAPs), this arrangement may be optimal for information processing rather than supportive structural functions. These microtubule/MAP arrangements are connected to synaptic receptors on the dendrite membrane by a variety of calcium and sodium influxes, actin and other inputs (7). Alterations in the microtubule/MAPs network in the dendrites correlate with the arrangement of dendrite synapatic receptors (8). Studies (9) demonstrate that the cytoskeleton is also involved in signal transmission. It is suggested that the microtubule lattice is well designed to represent and process information.

Tubulin switches between two conformations. It is suggested that tubulin conformational states could interact with with neighbouring tubulin by means of dipole interactions. The dipole-coupled conformation for each tubulin could be determined by the six surrounding tubulins.
Hameroff describes protein conformation as a delicate balance between contervailing forces. Proteins are chains of amino-acids that fold into three dimensional conformations. Folding is driven by van der Waals forces between hydrophobic amino-acid groups. These groups can form hydrophobic pockets in some proteins. These pockets are critcal to the folding and regulation of protein. Amino acid side groups in these pockets interact by van der Waals forces. Nonpolar atoms and molecules can have instantaneous dipoles.

Hameroff discusses the process of anesthesia which erases consciousness, but leaves many non-conscious functions intact. Anesthetic gas molecules are soluble in a lipid-like hydrophobic environment. Such areas are present in the brain in the lipid regions of cell membranes and in hydrophobic pockets within proteins.  It is suggested that anesthetic gas molecules interact with amino-acid groups via London forces, altering the normal action of London forces on the conformation of protein.

Hameroff discusses quantum information processing. Quantum superpositions where the quantum waves represent multiple possibilities for the state of a particle, are known to persist until quanta are either measured or naturally interact with the rest of the environment. Hameroff takes the view that the original mainstream interpretation, Copenhagen Interpretation, puts not only consciousness but the concept of reality itself outside physics. Alternatives interpretations include the 'many worlds' view, where there is no collapse but the superpositions continue in multiple worlds and David Bohm's idea in which the quanta are guided by active information.

Penrose's own take on the wave function collapse suggests that it is a real event. He sees superposition as a separation in the underlying space-time geometry. Each quanta is embedded in a bit of space, and as the superpositions grow further apart, a blister or separation appears in space-time. This can be viewed as the same thing as the beginning of the multiple world view, but instead of going on to generate separate universes, if the separation between superpositions grows to more than the Planck length, the wave collapses and chooses one of the superposed alternatives.

The normal quantum wave collapse is seen as an entirely random choice of the state of a quantum particle, from amongst the various superpositions of states. However, these collapses involve interaction with the environment. Penrose suggests that a quanta, which does not interact with the environment will undergo objective reduction (OR) when the separation between superpositions begins to exceed the Planck length. He also suggests that while the normal collapse is totally random OR is not totally random but involves a non-computable process. This is suggested because Penrose thinks that the brain manifests a non-computational aspect, and that the wave function collapse is the only place in the universe where such a thing can exist. Penrose also proposes that OR based quantum computation occurs in the brain.

It is important to stress that quantum computing as such is not expected to generate consciousness. In quantum computers, which many researchers, are now trying to develop quantum collapse will occur as a result of measurement or interaction with the environment. It is only in the event of OR that non-computability and consciousness could be brought into play.

Hameroff goes on to look at some of the detail of the theory that he and Penrose developed as to how consciousness could be based in microtubules in the brain. It is suggested that quantum compuations take place in microtubules orchestrated by the inputs of synapse via MAPs. Hence the theory is often known as Orch OR for orchestrated objective reduction. The computations are suggested to persist for 25 ms, which would link them to the 40Hz gamma synchrony, viewed as a correlate of consciousness even in more mainstream theories. The computations are terminated by objective reduction. It is proposed that in dendrites, the tubulin sub-units of the microtubules interact by dipole coupling so as process information. The tubulin conformation is governed by quantum London forces, so that the tubulins can exist as quantum superpositions of different conformations. In superposition the tubulins would be qbits in a quantum computer, computing by means of non-local entanglement with other tubulin qbits. This entanglement would not just be with tubulins in the same microtubule, but other microtubules in the same dendrite, and in other dendrites connected by gap junctions. Neurons connected by gap junctions can be viewed as a single hyperneuron, and the hyperneuron can be seen as a conventional neuron assembly.

The dendritic interiors alternate between two states as a result of the polymerisation of actin protein. In the depolymerised form the interior of the neuron is aqueous and microtubules signal and process information classically. There are synaptic inputs to the microtubules during this phase. When actin polymerises the interior of the dendrite becomes quasi-solid of gelatinous, and water near to the proteins becomes ordered as a result of the actin gelation. Debye layers of counterions may also shield the microtubules, due to the charged C-termini tails on the tubulins. This is suggested to make the microtubules sufficiently isolated from the environment for quantum superposition to occur in the tubulins. The geometry of a quantum computer lattice could be formed so as to be resistant to decoherence. Microtubules are suggested to have a structure which is particularly suitable for error correction. Coherent pumping of energy and quantum error correction may thus help to prevent decoherence. Quantum error correction involves a code that can detect and correct decoheence in a quantum system.

Hameroff claims to refute Tegmark's attempt to disprove the Penrose/Hameroff model. This is significant as Tegmark's criticism of Orch OR has been widely accepted as a completely satisfactory dismissal of the theory, and responses to Tegmark are habituaaly ignored.  Tegmark calculated microtubule decoherence time as being 10^-13 seconds, which would certainly be much too short for any neural activity. However, he worked on the basis of his own model for quantum activity in microtubules, which was never proposed by Hameroff or anyone else, basing his calculation on a 24nm separation of solitons from themselves along the microtubules, whereas Orch OR proposes a superposition separation distance six orders of magnitude smaller. For some reason, Tegmark did not choose to address the Penrose/Hameroff model. This invalidates his particular approach, whatever the truth is about decoherence, but it has not prevented his work from being quoted as an absolutely reliable refutation of Orch OR.