Consciousness typically lags behind the brain’s “initial electrical activation” by around 200-300ms. This implies that some higher order computation takes place to enable conscious access and to bridge the gap between brain activation and awareness.
The gap is demonstrated by a range of experiments, my favorite of which is Libet’s hand movement experiment, where participants were told to flick their wrist when they chose to, while watching a fast clock, and to report the timing of their decision with millisecond accuracy. While they reported the conscious urge to move around 200ms before the action, an EEG brain scan showed that the brain activity underlying the decision began half a second earlier. This shows that a decision is made, you experience making the decision 500ms later, and then 200ms after that, you execute on it. This distinction between the mechanisms underlying brain activity, and the mechanisms underlying the conscious perception of it, can be illustrated perfectly in Alzheimer's Disease(AD). In AD, the initial processing in the forgotten brain circuit is often preserved, but the higher order computation required for conscious access is not. Alzheimer’s patients have working memories, they just can’t access them.
The biochemical pathway below is central to bridging the gap between the "initial electrical activation” that creates your thoughts and the computation that allows you to be conscious of them:
BDNF↑→TrkB↑→PI3K/Akt↑→inhibits GSK3β↓→tau binding↑→Microtubule stability↑, assembly↑
It is activated by exercise, novel experiences, reasoning, sun exposure and psychedelics. Strong stimulation of the pathway is known to improve mood, enhance memory, and increase neuroplasticity, allowing the brain to grow, learn and adapt. Weak stimulation of this pathway is linked with AD, Depression, Schizophrenia, and Parkinson's Disease, which I will argue fundamentally, are Disorders of Consciousness. I will present the case that, for higher order conscious computation to occur, microtubules must be abundant and stable. Microtubule growth and stability are outcomes of the BDNF biochemical pathway.
The final targets of the pathway are microtubules. In a brain of 86 billion neurons, there are 5 quadrillion microtubules. Dismissed by most as simple components of the cell skeleton, they emerge as dynamic, nanoscale polymers with unique electrical, mechanical and information processing properties. Arranged in a cylindrical lattice structure, woven together by Microtubule Associated Proteins (MAPs), microtubules form highly ordered structures that participate in synaptic plasticity and learning. These highly ordered networks are built as a consequence of strong MAP binding and lattice maintenance.
Construction of these networks relies on the pathway I highlighted earlier. It is initiated by Brain-Derived Neurotrophic Factor (BDNF), a protein that binds to the TrkB receptor and sets off various metabolic cascades, all of which are important, yet one may unlock the answer to one of science’s most enduring mysteries: consciousness. Many papers in Neurobiology indirectly support microtubule stability as a critical underlying mechanism for their findings, but few point to it as a factor worth considering, as the implications are not yet widely accepted.
Synaptic plasticity, learning and memory formation relies on BDNF. Not only does BDNF inhibit GSK3β, stopping it from breaking down microtubules and removing MAPs, it stimulates mTOR to build more MAPs such as tau, CRMP2 and MAP2; which then actively bind and stabilise the microtubule lattice.
BDNF and MAP levels in the brains of patients suffering from AD, Depression, Schizophrenia, and Parkinson's Disease provide a great foundation to my argument. As established with regards to AD, the “initial electrical activation”, even in the “forgotten” or consciously inaccessible parts of the brain, is present. The memory centre of the brain, and specifically the memory circuits that the individual can’t remember will still successfully fire action potentials (the mechanism by which nerve cells transmit electrical signals), yet they still aren’t conscious of those memories. It is crucial to say now, that microtubules are not involved in the firing of the action potential, they do however, seem to be involved in your conscious access to it. This involvement is illustrated by the fact that BDNF and tau are low specifically in the memory centre (Hippocampus) of AD patients, the pre-frontal cortex of Schizophrenic patients and the Substantia Nigra and Dopaminergic neurons of Parkinson's Disease patients, causing them to suffer from motor rigidity, tremors, and motivational deficits, a list of symptoms aligning precisely with conscious exclusion of those brain regions.
A substance that, backed by a growing body of evidence, is being used as a treatment for the previously named diseases, has recently been found to stabilize microtubules and build their network. Commonly known as a “mind and consciousness expander”, Psilocybin, a psychedelic drug, will provide some clinical, and I hope more intuitive evidence for the relationship between microtubules and consciousness. It is documented through EEG studies on psychedelics that overall signal complexity, diversity and richness increases on a whole brain scale, as described in the Entropic Brain Hypothesis. Based on recent evidence, however, on a molecular and subatomic scale, psilocybin creates low entropy, ordered states in the microtubule lattice. The apparent contradiction dissolves when viewed through a multi-scale lens, where increased brain level entropy comes about due to atom scale order under psilocybin.
The boost in signal richness is thought to be created by psilocybin’s role in stimulating the serotonin (5HT2A) receptor, with both the hallucinogenic and therapeutic effects attributed to it. I believe that, while serotonin stimulation is important, psilocybin’s newly discovered affinity for the TrkB receptor, the same receptor that BDNF activates, leading to microtubule stability, is imperative to its positive health outcomes. Based on these findings, psilocybin therapy operates through two mechanisms:
- Activation of 5HT2A receptor and subsequent serotonin release, which creates a dampening of negative valence, increased activity in the cortex, and broadening of sensory integration.
- Overlapping activation of the BDNF pathway, stabilising the microtubule lattice, enhancing the strength of the activated brain circuits and growth of new connections.
Up to an hour after the administration of psilocybin, BDNF has already begun the process of reinforcing the microtubule lattice and network, and structural consolidation of the brain circuits that were stimulated, particularly in the brain areas most stimulated by serotonin, leading to positive long-term changes. This also aligns with the reported timescale of psilocybin induced increases in perceived consciousness.
Naturally these changes in stimulated circuits happen slowly through Long Term Potentiation (LTP), the main mechanism involved in learning and memory consolidation. It involves Glutamate, AMPA and NMDA receptors, calcium and more to change the structure of the whole cell and synapse network. This mechanism, as you may imagine based on the rest of my piece, culminates in microtubule stabilization via the same BDNF and MAP mechanism as psilocybin. Memory consolidation through LTP can be blocked by Nocadozole, a microtubule polymerization inhibitor, which obstructs the growth of microtubules.
LTP works through a multi-step process.
- Learning triggers neuronal activation in specific brain regions
- Initial disassembling of microtubule ends in the activated neurons, priming them for further construction
- Biochemical tagging of the neurons for later reinforcement
- Microtubules in tagged neurons experience hyper-stabilization, where MAPs reinforce the lattice and new receptors are added to the neuron
- Reactivation of those neurons during sleep, releasing more BDNF and further reinforcing the lattice
- Long-term maintenance and periodic reactivation of the newly consolidated circuit
Injection of Nocadozole before step 2 blocks long term learning but puzzlingly allows for use of the existing network for short term recall of newly learned information. The short term memory trace does not survive the subsequent steps of LTP and is no longer consciously accessible in 24 hours. Injection of Nocodazole at step 4 blocks the hyper-stabilization of the microtubule network, resulting in a memory trace that is not consciously accessible but exists as a circuit in the brain. This finding is reinforced by injection of microtubule stabiliser Paclitaxel which has the exact opposite effect to Nocodazole at step 4, enhancing LTP beyond baseline levels and markedly exceeding the reduced LTP and learning ability observed with nocodazole.
Fig1: Breaking down microtubules reduced memory recall, while stabilizing them enhanced it.
Inducing LTP in the forgotten memory circuits in mice with early Alzheimer’s disease either with optogenetic stimulation (direct stimulation of genetically modified cells with light), or with targeted stabilization of microtubules using Epothilone D, was shown to return conscious access to the memories. Both of these techniques converge on the same mechanism, a stable microtubule scaffold is the key differentiator between stored information in the brain and the brain wide broadcast of information we call consciousness. This targeted stimulation was able to restore conscious access to the brain circuits underlying lost memories.
An even stronger link between consciousness and microtubule stability comes from anaesthesiology. There is strong evidence that volatile anesthetics like Isoflurane bind to microtubules, inducing instability in the lattice, and therefore loss of consciousness. Destabilising the lattice using Nocodazole sensitises the brain to anaesthetics, accelerating loss of consciousness, while stabilising it using Epothilone B delayed anaesthetic induced loss of consciousness, and hastened recovery. This shows that consciousness can be manipulated by modulating microtubules.
Microtubules are dynamic polymers, but BDNF signalling shifts them towards a more ordered and lower entropy state. Under basal conditions, microtubules are constantly shifting, growing and shrinking, a behaviour driven by entropy at an atomic scale. Microtubule Associated Proteins (MAPs) like tau, or chemicals like Epitholone B order the microtubules, connecting them in a network, lowering the configurational entropy of the network by reducing the number of different spatial arrangements those particles can take.
The relationship between conscious computation and low configurational entropy is described by Nick E. Mavromatos, Andreas Mershin, and Dimitri V. Nanopoulos, distinguished physicists at MIT, King’s College London and Cern in their paper: “On the Potential of Microtubules for Scalable Quantum Computation”. While they don't formulate a theory of consciousness, they provide an explanation for how microtubules could theoretically contribute to the 500 ms worth of computation required for conscious access to a neural stimulus.
The researchers propose that quantum mechanical processes can occur within the microtubules. This happens through an interaction between the tubulin proteins that make up microtubules and the ordered water within their 15nm hollow cores. Each tubulin protein consists of two subunits carrying opposite charges, creating what physicists call dipoles. When these dipoles interact with the structured water, they can create discrete regions where quantum coherent coupling becomes possible, allowing the classical rules of physics to give way to quantum mechanics. This produces superposition states where the system exists as a combination of all its possible configurations. During this phase, quantum interference between these configurations can bias the system toward the most stable and energetically favorable outcome.
What makes this theory interesting is how this coupling is only possible given the rigid, MAP reinforced microtubule environment visible in healthy brains. It allows for coherent coupling of dipoles at very small distances between the dipoles and ordered water in the microtubule cavity. In these areas, an external stimulus, such as light or electricity, can excite multiple tubulin dimers, placing them into a superposition state. Tubulin proteins have multiple degrees of freedom; conformational flexibility, vibrational modes, charge distributions, and large dipoles. During superposition, the proteins exist in all these states simultaneously, and can become quantum entangled across the entire microtubule.
The authors of the paper propose that, with their predicted decoherence times of 10⁻⁶ to 10⁻⁷ seconds, a million times longer than previous estimates due to the dipole-dipole interactions temporarily overcoming thermal decoherence, the system has enough time to “explore” all possible pathways for signal transmission in parallel. As the quantum state decoheres via slow water leakage through the microtubule walls, the system naturally settles into a configuration that minimizes energy loss, and provides the most efficient signal transmission path in the form of a classical wave called a soliton. The soliton can take multiple shapes depending on the quantum computation’s chosen configuration. The authors then claim solitons, by interacting with each other through MAPs, can implement logical operations, travelling at up to 155m/s.
The process of conscious calculation therefore operates at a similar speed as myelinated nerve fibers, at around 150m/s. It is initiated by a local electrical event in a neuron, like an action potential, which rapidly creates coherence in the microtubule, which then decoheres into a solitonic wave that carries both energy and information along the MAPs across microtubules. This allows a “timing signal” to be spread across the neuron, biasing it towards firing at a certain phase. Brain circuits generally fire in rhythm, allowing them to propagate the signal without interruption by other stimuli. A stimulus out of phase will often be ignored by the brain’s machinery and will not gain conscious access.
For a given stimulus to achieve conscious access, it must be broadcast across the entire brain, and therefore must be delivered in phase with similar stimuli. This is where the quantum computation comes in, effectively telling the neuron to “fire at this exact moment”. If this phase bias caused by the computation is broadcast far enough, the message will travel in a synchronized manner, and will successfully be broadcast across the brain, leading to a conscious thought.
This sort of synchronized oscillation, and bias towards firing at a certain time can be artificially induced within the brain by Transcranial Functional Ultrasound Stimulation (tFUS) or using Terahertz (THz) radiation. tFUS delivers low intensity, high frequency sound waves through the skull with millimeter precision, producing small oscillations within cellular structures. These induced oscillations, at the right frequencies, are capable of interacting with microtubules, shifting dipole oscillations and vibrational modes. This can induce the same kind of phase bias that I described with regards to quantum computationally generated solitons, biasing neurons toward firing in certain phases. This is consistent with studies showing the tFUS can modulate action potentials, increase the probability that a certain stimulus reaches the conscious threshold and induce meditative states in humans.
THz radiation interacts with microtubules in a different but equally fascinating way, affecting the vibrational modes of tubulin proteins, modulating their dipole oscillations. Experimental evidence suggests that volatile anesthetics shift these oscillations in proportion to their potency. This suggests that THz stimulation can modulate microtubule oscillations directly, stabilizing them or reducing their probability of a successful timing bias.
While I set out with the goal of finding the quantum substrate of consciousness, I have ended up no further than where I began: with the sense that consciousness is merely “what it feels like to perform computation”. Quantum computation may be involved in setting the timing of oscillations through microtubule oscillatory dynamics, but it does not seem to be at the core of a more satisfying explanation.
The aforementioned oscillatory dynamics do, however, open the door to much speculation. If quantum coherence only sets the timing, and consciousness is what it feels like to run a phase-aligned global computation, then we can try to shape oscillations, and modulate which content gets conscious access and for how long. If an experiment using tHZ or tFUS along with pharmacological microtubule stabilization can keep early sensory responses unchanged, but drive target regions to fire in phase in later stages, then I predict that subjective access to the content from those target regions should rise in lockstep.
Though highly speculative, experimental validation of these ideas could unlock novel applications and offer compelling evidence for their role in consciousness:
- Alzheimer’s Disease Treatment:
In AD, memories still fire, they just never cross the threshold into conscious access because the microtubule-based phase timing is degraded, causing the signals to be drowned out by surrounding activity. If we can restore BDNF signalling and microtubule stability (via exercise, epothilones, or psilocybin’s TrkB activation), and resynchronise the loops connecting memories and cognition (hippocampus to pre-frontal cortex) with tFUS, we can reconstruct the phase-aligning machinery, allowing concealed memories to once again attain global neural ignition and conscious recall.
- Learning Acceleration
During sleep, the brain “replays” the firing patterns of the day preceding it, whether they constitute a new swing sequence from golf practice or or insights acquired from an article on consciousness. This happens by sending signals through the hippocampal to prefrontal cortex loop for the article and hippocampal to motor and cerebellar loops for the golf swing. Multiple memories compete for consolidation, and the ones with synchrony tend to win. By using tFUS, we can bias the learning towards specific loops, enhancing replay for selected memories.
- Flow State Induction
In peak performance states, or “flow” states, fronto-parietal control networks and sensory and motor networks operate in high frequency synchrony, allowing rapid and precise action with minimal conscious interference. By using tFUS to align oscillations across these networks, we can artificially recreate the phase coherence that naturally emerges during flow states, reducing noise and enhancing efficiency. THis would result in a network biased toward uninterrupted and efficient computation, with minimal conscious interruption.
- Creativity Boosting
Creative insights often emerge when high-level executive control in the pre-frontal cortex is loosened, allowing more interaction between normally segregated brain networks. Psychedelic drugs are known to amplify creativity by reducing segregation and boosting cross network communication. Using tFUS to reduce alpha band synchrony and boost gamma and beta oscillations across associative networks could increase the probability of novel connections forming.