Quantum Computation in Brain Microtubules? The Penrose ...

Quantum Computation in Brain Microtubules? The Penrose-Hameroff 'Orch OR' Model of Consciousness [and Discussion] Author(s): Stuart Hameroff and P. Marcer Source: Philosophical Transactions: Mathematical, Physical and Engineering Sciences, Vol. 356, No. 1743, Quantum Computation: Theory and Experiment (Aug. 15, 1998), pp. 1869-1896 Published by: The Royal Society Stable URL: http://www.jstor.org/stable/55017 Accessed: 30/09/2009 06:43 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=rsl. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

The Royal Society is collaborating with JSTOR to digitize, preserve and extend access to Philosophical Transactions: Mathematical, Physical and Engineering Sciences.

http://www.jstor.org

(Th&cRoyaISocceti/

(flip

Quantum

computation

The

Penrose?Hameroff

in

brain

microtubules?

'Orch

of

consciousness

By

Stuart

Hameroff

and Psychology, The of Anesthesiology USA Tucson, AZ 85724,

Departments

model

OR5

University

of Arizona,

Potential

features of quantum could explain enigmatic of con? aspects computation The Penrose-Hameroff model reduction: 'Orch objective (orchestrated that and a form of quantum superposition quantum computation suggests OR') occur in microtubules?cylindrical within the protein lattices of the cell cytoskeleton brain's neurons. Microtubules couple to and regulate neural-level functions, synaptic and they may be ideal quantum because of dynamical lattice structure, computers subunit states and intermittent isolation from environmental inter? quantum-level actions. In addition to its biological the Orch OR proposal differs in an setting, essential envisioned in which collapse, quantum computers way from technologically or reduction to classical output states, is caused by environmental decoherence (hence In the Orch OR proposal, of microtubule reduction quan? introducing randomness). tum superposition to classical states occurs by an objective output factor?Roger from instability in Planck-scale Penrose's quantum gravity threshold sep? stemming arations in spacetime states following Penrose's Output geometry. (superpositions) are neither totally reduction deterministic nor random, but influenced objective by a non-computable in fundamental factor ingrained spacetime. Taking a modern panview in which protoconscious and Platonic values are embedded psychist experience in Planck-scale the Orch OR model portrays consciousness as brain spin networks, linked to fundamental activities ripples in spacetime geometry. sciousness.

Keywords; consciousness; orchestrated objective

1.

Quantum

quantum reduction

computation; objective reduction; brain (Orch OR); microtubules;

computation

and

consciousness

for quantum states implementing multi? computation rely on superposed in to linear ple computations parallel, according simultaneously, quantum superposi? tion (see, for example, Benioff 1982; Feynman k Josza 1986; Deutsch 1985; Deutsch In is of quantum principle, computation capable specific applications beyond 1992). the reach of classical Shor 1994). A number of techno? computing (see, for example, aimed at realizing these proposals have been suggested and are being logical systems for quantum evaluated as possible substrates computers ions, electron (e.g. trapped spins, quantum dots, nuclear spins, etc. (see table 1, Bennett (1995) and Barenco The main obstacle to realization of quantum is the problem computation (1996))). of interfacing to the system while also protecting the quantum state (input, output) from environmental If this problem can be overcome, decoherence. then present day classical computers may evolve into quantum computers. Proposals

Phil Trans. R. Soc. Lond. A (1998) 356, 1869-1896 Printed in Great Britain 1869

? 1998 The Royal Society T^ipCPaper

S. Hameroff

1870

The workings of the human mind have been historically described as metaphors of In ancient Greece memory was like a 'seal ring information technology. contemporary in wax' and in the 19th century the mind was seen as a telegraph circuit. In switching has been the dominant this century the classical for the brain's computer metaphor If quantum becomes a technological computation reality, consciousness seen as some form of be Indeed enigmatic quantum computation. inevitably

activities.

may features of consciousness the brain.

have

already

led to proposals

for quantum

computation

in

as an emergent of clas? Conventional explanations portray consciousness property in the brain's neural networks activities sical computer-like functionalism, (e.g. materialism, computationalism 1986; Den? reductionism, physicalism, (Churchland The for a k Sejnowski current candidate nett 1991; Churchland leading 1992)). of correlate' consciousness involves neuronal circuits oscillat? 'neural computer-like in the thalamus and cerebral cortex. Higher-frequency oscillations ing synchronously 'coherent are to mediate as 40 known suggested temporal binding of Hz') (collectively for Koch et al. k Joliot conscious Crick 1990; example, 1990; experience Singer (see, for as to whether coherence et al. 1994; Gray 1998). The proposals example vary, or resonates in cortical in the thalamus but 'thalamo-cortical networks, originates for consciousness. view of the neural-level substrate 40 Hz' stands as a prevalent and feelings? Conventional But how do neural firings lead to thoughts ('function? features. These include (1) the fall short on the mind's enigmatic alist') approaches or qualia, our 'inner life' (see, for example, nature of subjective experience, Nagel into uni? of spatially distributed brain activities Chalmers 1974; 1996); (2) 'binding' from preconscious sense of 'self; (3) transition tary objects in vision and a coherent to 1989, 1994, 1997); and consciousness; processing (Penrose (4) non-computability (5) free will. Functionalist novel property

assume that conscious generally experience appears as a approaches surface this the at a critical level of computational On complexity. threshold has neither would seem to deal with issues (1) and (3); however, a conscious in electrophysioand there are no apparent differences nor predicted, been identified the nature and conscious between non-conscious activity. Regarding logical activities no not functionalism offers testable we are of experience unfeeling 'zombies'), (why in vision and self is often function? of attributed Problem by 'binding' predictions. (2) 40 Hz), but it is unclear why temporal correlation alists to temporal (e.g. coherent As an explanation of experience. without correlation per se should bind experience it is also unable to account for is based on deterministic functionalism computation, or free will Penrose's may be missing. non-computability proposed (4), (5). Something in which macro? have been suggested To address these issues, various proposals For to the brain's known neural activity. are connected scopic quantum phenomena that Marshall coherent the problem of unitary quantum binding, (1989) suggested occurred condensation states known as Bose-Einstein among neural proteins (cf. Penrose 1987; Bohm k Hiley 1993; Jibu k Yasue 1995). Preconscious-to-conscious of a quantum wave func? were identified transitions by Stapp (1993) with collapse In k Eccles another axon terminals Beck tion in presynaptic proposal, (cf. 1992). within the brain's neurons are viewed as selfmicrotubules called assemblies protein reduction?Orch OR': see, objective computers quantum organizing ('orchestrated k Penrose k Hameroff Penrose for example, 1995; Hameroff 1996a, 6; cf. Hameroff 1997,

1998a-d).

Phil Trans. R. Soc. Lond. A (1998)

Quantum

computation

in brain

microtubules?

1871

of macroscopic states in biological At first glance the possibility quantum systems to extreme cold seems unlikely, either avoid thermal require appearing (to noise) or to achieve coherent states. as in laser-like And pumping pro? energetic technological of the quantum state from the environment posals, perfect isolation quantum (and/or error correction com? codes) would be required while the system must also somehow with the external world. Living cells including the brain's neurons municate seem warm and wet for delicate states which would seem suscepti? quantum unsuitably ble to thermal

noise

and environmental decoherence. conditions However, specific states in have microtubules evolved supporting quantum may (see ?3). In addition to its biological the Orch OR proposal differs significantly setting, in another envisioned The latter quantum computers. regard from technologically of states through reduction would arrive at output quantum superpo? ('collapse') sition to classical states by environmental decoherence?the state would quantum the in external The outcome states be interrupted world. by quantum technological would therefore reflect deterministic at reduction influenced computers processing randomness. by some probabilistic that isolated quantum On the other hand, Penrose (1989, 1994, 1996) has proposed avoid environmental decoherence will which reduce nonetheless eventually systems threshold reduction' due to an objective related to an intrinsic ('objective (OR)) of fundamental Unlike the situation fol? spacetime geometry (see below). outcome states which reduce to environmental due Penrose's decoherence, lowing are selected on the deterministic reduction influence by a non-computable objective a non-algorithmic implies pre-reduction quantum computation. Non-computability nor random, a property which Penrose process which is neither deterministic (e.g. to conscious and understanding. This clue suggests thought 1997) also attributes feature

with objective reduction be involved in computation quantum may somehow consciousness. The objective factor in OR is an intrinsic feature of spacetime itself (quantum grav? the Penrose from with notion that mass is equivalent to general relativity begins ity). He curvature. concludes that spacetime quantum superposition?actual separation of mass from itself?is to simultaneous curva? equivalent spacetime (displacement)

that

in opposite or separations in fundamental directions, causing 'bubbles', reality the bubbles as with a Penrose views critical unstable, objective degree (figure 1). in instantaneous to classical of separation reduction states. resulting unseparated are therefore events reductions which the fine scale of Objective reconfigure spaceAs described in ?5, modern pan-psychists time geometry. attribute protoconscious to a fundamental of physical experience reality. If so, consciousness property might tures

an experiential OR events rippling through medium. self-organizing in the brain? If so, they would be expected Could OR events be occurring to coin? with cide with known neurophysiological time-scales. The crit? processes recognized ical degree of spacetime Penrose's reduction is related separation causing objective to quantum gravity by the uncertainty principle involve

E = h/T, where

E

is the gravitational of the superposed mass (displaced self-energy from the for diameter of its atomic h is Planck's constant example, itself by, nuclei), over 2tt and T is the coherence time until OR self-collapse. The size of an isolated related to the length of time until self-collapse. superposed system is thus inversely Phil Trans. R. Soc. Lond. A (1998)

S. Hameroff

1872

Time

Figure 1. (a) Schematized protein (tubulin) capable of switching between two conformational states governed by London force interactions in a hydrophobic pocket. (Tubulin may actually have multiple, smaller, collectively governing hydrophobic pockets and more than two possi? ble states. For simplicity a one-pocket two-state protein is illustrated). Top: protein switching between two conformational states coupled to localization of paired electrons (induced dipoles) within a hydrophobic pocket (see ?2). Bottom: quantum superposition (simultaneous existence in two distinct states) of the electron pair and protein conformation. (6) Four-dimensional spacetime may be schematically represented by one dimension of space and one dimension of time: a two-dimensional 'spacetime sheet'. Mass is curvature in spacetime, and the two spacetime cur? vatures in the top of the figure represent mass (e.g. a tubulin protein) in two different locations or conformations, respectively. Mass in quantum superposition (mass separated from itself) is simultaneous spacetime curvature in opposite directions, a separation, or bubble in spacetime. At a critical degree of separation, the system becomes unstable and must select either one state or the other (from Penrose (1994), p. 338). 1 kg cat) would self-collapse superposed systems mythical (e.g. Schrodinger's atom would undergo OR only after superposed (OR) in only 10~37 s; an isolated then IO6 years! If OR events occur in the brain coupled to known neurophysiology, we can estimate that T for conscious OR events may be in a range from 10 to 500 ms. activities such as 25 ms 'coherent 40 Hz', 100 ms This range covers neurophysiological ms Libet's 500 OR events coupled to and EEG rhythms sensory perceptions. (1979) mass. a of ms would few activities 100 require nanograms superposed roughly for and reduction best suited materials quantum computation objective Biological Large

are proteins,

particularly

assemblies

2.

of proteins

Proteins

and

called

microtubules.

qubits

which perform a variety of functions are versatile macromolecules Proteins by chang? include mem? functions muscle their conformational Such movement, shape. ing ion channels, and closings of protein molecular brane firing via openings binding, movement and phase of cytoplasm. Life is organized metabolism, enzyme catalysis, by changes in protein shape. as linear chains of hundreds of amino acids are synthesized Individual proteins The precise manner of folding for which 'fold' into three-dimensional conformation. and repellent forces among its various amino acid each protein depends on attractive Phil. Trans. R. Soc. Lond. A (1998)

Quantum

computation

in brain microtubules?

1873

and a current view is that many possible intermediate conformations groups, final linear the one of amino complete precede sequences (Baldwin 1994). Although their final acid chains are known for many proteins, three-dimensional predicting folded shape using computer has proven difficult if not impossible. This simulation is known as the 'protein folding problem' and so far appears to be 'NP conundrum in theory, but the space and time required the answer can be calculated complete': is prohibitive. of any classical com? computer protein folding is a quantum Perhaps side

putation (L. Crowell, personal communication)? The main driving force in protein folding occurs as uncharged non-polar groups of particular amino acids join together and avoid water. Repelled by solvent water, each other (by van der Waals forces?see 'hydrophobic' non-polar groups attract and themselves As a result, within the interior. protein bury intraprotein below) of of side occur, composed hydrophobic pockets groups non-polar (but polarizable) amino acids such as leucine, and isoleucine, phenylalanine, tryptophan, tyrosine valine. Volumes of the pockets (ca. 400 A3, or 0.4 nm3) volume of a single protein, and their physical solvent resemble olive oil.

are roughly tne total ^-250 characteristics most closely

For exam? Though small, hydrophobic pockets may be critical to protein function. molecules which anaesthetic ablate consciousness exert their effects ple, gas reversibly in hydrophobic of neural for Franks Lieb k pockets proteins example, 1982, (see, bind in hydrophobic called pockets by weak physical interactions 1985). Anaesthetics London dispersion forces, a type of van der Waals force. Why are these weak localized so important to protein function interactions and consciousness? in a living state are dynamical, Proteins and only marginally A protein stable. of 100 amino acids is stable against ca. denaturation 40 kJ whereas by only mol-1, of kJ mol-1 are available in a protein from side-group thousands interactions includ? van der Waals forces. conformation is a 'delicate balance ing protein Consequently forces' (Voet k Voet 1995). among powerful countervailing in proteins Transitions occur at many time- and size-scales. For example, small amino acid side chains move in the pico-femtosecond time-scale (10_12-10-15 s), and conformational transitions in which proteins move globally and upon which protein function generally depends occur in the nanosecond (10~9 s) to 10 picosecond k McCammon (10-11 s) time-scale (Karplus 1983). These global changes (e.g. as in to involve collective of actions various represented schematically figure la) appear activities bond van intraprotein rearrangements, dipole oscillations, (e.g. hydrogen der Waals forces). The types of forces operating amino acid side groups within a protein among include charged interactions such as ionic forces and hydrogen as well as inter? bonds, actions between in neutral dipoles?separated charges electrically groups. Dipoleare known interactions as van der Waals forces and include three types: dipole (1) (3) the sion

permanent dipole; (2) permanent dipole-permanent dipole-induced dipole; and induced Induced interactions are dipole-induced dipole. dipole-induced dipole weakest but most purely non-polar forces. They are known as London disper? are forces, and although quite delicate (40 times weaker than hydrogen bonds)

numerous and influential. The London force attraction less than a few kilojoules; thousands usually however, other forces cancel out, London forces in hydrophobic conformational states. Phil Trans. R. Soc. Lond. A (1998)

between any two atoms is occur in each protein. As can pockets govern protein

1874

S. Hameroff

London forces ensue from the fact that atoms and molecules which are electri? and spherically have instantaneous nevertheless electric cally neutral symmetrical in their electron distribution. The electric dipoles due to instantaneous asymmetry clouds of adjacent field from each fluctuating to others in electron dipole couples in electron amino acid side groups. Due to inherent localiza? uncertainty non-polar are quantum forces which govern protein conformation effects which tion, London couple apparently Milloni 1994).

to 'zero-point

fluctuations'

of the quantum

vacuum

(London

1937;

within hydrophobic were proposed pockets Quantum-level dipole oscillations by and Conrad (1994) suggested Frohlich (1968) to regulate protein conformation, pro? of various conformations before one is teins use quantum possible superposition which depend selected. et al. (1995) showed functional protein vibrations Roitberg and Tejada on quantum effects centred in two hydrophobic residues, phenylalanine states exist in the protein et al. (1996) have evidence to suggest coherent quantum in determining ferritin. be using quantum Could proteins superposition ('qubits') states (bits)f? their conformational The possible situation as in figure la using the microtubule may be characterized dimers with two con? are peanut-shaped Tubulins tubulin as an example. protein and they undergo several types of conformational nected monomers changes (see, can shift 30? from the one monomer for example, Cianci et al. 1986). For example, tubulin dimer's vertical axis (Melki et al. 1989; cf. Yagi et al. 1994). At the top of fig? between two such states, governed by hydrophobic ure la a tubulin protein switches have several may actually pocket electron pairs coupled by London forces. (Tubulin for simplicity we and occupy more than two states; however, pockets hydrophobic states in the top The two possible consider one pocket and two states per protein.) the one bit of information. of figure la may be viewed as representing If, however, then con? protein la, pocket electron pair is superposed hydrophobic bottom), (figure and exists is also superposed from the external formation environment) (if isolated A properly and isolated in both states simultaneously array of configured ('qubit'). interactive

protein

qubits

could

constitute

3.

a quantum

computer.

Microtubules

of living cells are functionally by webs of protein polymers?the organized are microtubules, selfof the cytoskeleton cytoskeleton (figure 2). Major components of lattices subunit whose walls are hollow hexagonal cylinders assembling crystalline for a variety of bio? are essential known as tubulin proteins (figure 3). Microtubules and establishment cell division cell functions movement, including logical (mitosis) self-assemble In neurons, microtubules of cell form and function. and maintenance then microtubules and form synaptic axons and dendrites to extend connections; for and maintain and cognitive learning responsible strengths help regulate synaptic and other of the role of microtubules functions. description (For a more complete Hameroff et al. in see structures functions, Dayhoff cognitive (1994), cytoskeletal have microtubules Hameroff While and k Penrose traditionally (1994).) (1996a) Interiors

f If so, the mechanism of anaesthetics may involve disruption of electron mobility required for quan? tum superposition in hydrophobic pockets of neural proteins. Experimental evidence has shown that anaesthetics inhibit mobility of free electrons in a non-biological corona discharge (Hameroff Sc Watt 1983). Phil. Trans. R. Soc. Lond. A (1998)

Quantum

computation

in brain microtubules?

1875

Figure 2. Schematic of neural synapse showing cytoskeletal structures within two neurons, (a) vesicles (spheres) into the synaptic cleft. Presynaptic axon terminal releases neurotransmitter Thick rod-like structures within the axon are microtubules; thinner filaments (e.g. synapsin) facilitate vesicle release, (b) Dendrite on post-synaptic neuron with two dendritic spines. Micro? tubules in main dendrite are interconnected by microtubule-associated proteins. Other cytoskele? tal structures (fodrin, actin filaments, etc.) connect membrane receptors to microtubules (based on Hirokawa (1991)).

Figure 3. (a) Microtubule (MT) structure: a hollow tube 25 nm in diameter, consisting of 13 columns of tubulin dimers arranged in a skewed hexagonal lattice (Penrose 1994). (b) Top: each tubulin molecule may switch between two (or more) conformations, coupled to London forces in a hydrophobic pocket; bottom: each tubulin can also exist in quantum superposition of both conformational states (figure la) (cf. Hameroff k Penrose 19966).

been

considered

as purely structural has demonstrated recent evidence components, and communication functions et al. signalling 1997; Maniotis (Glanz Microtubules with interact membrane structures 6; Vernon k Wooley 1995).

mechanical 1997a,

Phil Trans. R. Soc. Lond. A (1998)

1876

S. Hameroff

Figure 4. Microtubule automaton simulation (from Rasmussen et al. 1990). Black and white tubulins correspond to black and white states shown in figures la and 3. Eight nanosecond time-steps of a segment of one microtubule are shown in 'classical computing' mode in which conformational states of tubulins are determined by dipole-dipole coupling between each tubulin where and its six (asymmetrical) lattice neighbours calculated by /net = (e2/47C?) X^=i(Wr?)> yi and ri are inter-tubulin distances, e is the electron charge, and e is the average protein permittivity. Conformational states form patterns which move, evolve, interact and lead to the emergence of new patterns. and activities ical signals. How could

by linking

proteins

(e.g. fodrin,

ankyrin)

and 'second-messenger'

chem?

Theoretical classical information microtubules implement processing? coherent for that subunit tubulins microtubule excitations, undergo propose Frohlich in a mechanism the example, by gigahertz range by suggested ('pumped Frohlich exci? & Onsager 1968, 1970, 1975; cf. Penrose 1956)f. phonons') (Frohlich of tubulin within have been suggested to support tations subunits microtubules Hameroff for k and information Watt 1982; example, computation processing (see, to 'clock' compu? et al. 1990). The coherent excitations are proposed Rasmussen transitions tubulins tational acting as 'cells' as in occurring among neighbouring tubulins 'cellular automata'. molecular-scale Dipole couplings among neighbouring models

in the microtubule

lattice

act as 'transition

rules'

for simulated

microtubule

automata

f Experimental evidence for Frohlich-like coherent excitations in biological systems includes obser? vation of gigahertz-range phonons in proteins (Genberg et al. 1991), sharp-resonant non-thermal effects of microwave irradiation on living cells (Grundler k Keilman 1983), gigahertz-induced activation of microtubule pinocytosis in rat brain (Neubauer et al 1990) and laser Raman spectroscopy detection of Frohlich frequency energy in biomolecular systems (Genzel et al. 1983; Vos et al. 1992). Phil. Trans. R. Soc. Lond. A (1998)

Quantum

computation

1877

in brain microtubules? (b)

(a)

33; Outcome State

Reduction

Superposition Figure 5. Schematic of quantum computation of three tubulins which begin (a) in initial classical in which all possible states coexist. After states, then enter isolated quantum superposition reduction, one particular classical outcome state is chosen (6). transmission and learning information exhibiting processing, (figure 4) (Rasmussen et al. 1990). in the nanosecond automata scale offer a poten? Classical microtubule switching in the brain's increase Conventional computational capacity. tially huge approaches focus on synaptic 1011 brain neurons, 103 synapses per neuron, switching (roughly in the ms range of 103 operations and predict about 1017 bit per second) switching states per second for a human brain (see, for example, Moravec as 1987). However, cells each contain tubulins 107 Bass k typically biological approximately (Yu 1994), in microtubule nanosecond automata predicts roughly 1016 operations per switching second per neuron. This capacity could account for the adaptive behaviours of singlecell organisms for example, who elegantly like Paramecium, swim, avoid obstacles and find food and mates without the benefit of a nervous system or synapses. As the human brain contains about 1011 neurons, nanosecond microtubule automata offer about 1027 brain operations second. per will not by itself address However, even a vast increase in computational complexity the difficult issues related to consciousness. coherent states and quantum Quantum with reduction could do so. Figure 5 computation objective possibly (Orch OR) illustrates the general idea for quantum with tubulins: three tubulins computation are shown in initial states, in isolated of possible states during which superposition occurs in and outcome states. Figure 6 computation quantum single post-reduction Phil Trans. R. Soc. Lond. A (1998)

1878

S. Hameroff

Figure 6. Microtubule automaton sequence simulation in which classical computing (step 1) leads to emergence of quantum coherent superposition (steps 2-6) in certain (grey) tubulins due to pattern resonance. Step 6 (in coherence with other microtubule tubulins) meets critical threshold related to quantum gravity for self-collapse (Orch OR). Consciousness (Orch OR) occurs in the transition from step 6 to 7. Step 7 represents the eigenstate of mass distribution of the collapse which evolves by classical computing automata to regulate neural function. Quantum coherence begins to re-emerge in step 8. the quantum mode (grey) and automata microtubule computation entering to nonreduction threshold meeting objective (between steps 6 and 7) for self-collapse in figure 7, pre-reduction chosen outcome states. As described quantum computably and the objective to correlate with preconscious is suggested processing, computation can give moment. A series of such moments reduction process itself to a conscious rise to a stream of consciousness description). (see ? 5 for a more complete would have to somehow avoid states in brain microtubules Macroscopic quantum Nature may and still communicate with the environment. environmental decoherence and communication. have solved this problem with alternating phases of isolation in cytoplasm which are embedded Microtubules and other cytoskeletal components exists in alternating phases of (1) 'sol' (solution, liquid); and (2) 'gel' (gelatinous, shows

transformations' of biological the most primitive activities, 'sol-gel Among of and disassembly neurons and other living cells are caused by assembly the protein ions through actin (e.g. regulated calmodulin, by calcium cytoskeletal in basic are essential transformations in turn regulated Sol-gel by microtubules). and such as ('amoeboid') cellular activities movement, growth and synaptic formation et al. 1995). Transitions vesicle release (Miyamoto neurotransmitter 1995; Muallem and some actin gels can be can occur rapidly cycles per second), (e.g. 40 sol-gel transmitted deformation without response quite solid, and withstand (Wachsstock of microtubules et al. 1994). Cyclical encasement by actin gels may thus be an ideal mechanism isolation computing quantum (figure 8)f. A biphasic cycle of microtubule solid). within

f An additional mechanism may also shield microtubules. Dan Sacket at the National Institutes of Phil Trans. R. Soc. Lond. A (1998)

Quantum

computation

in brain

microtubules?

1879

(a)

Collapse, 'OrchOR"

Conscious

25 msec Figure 7. An Orch OR event, (a) Microtubule simulation in which classical computing (step 1) leads to emergence of quantum coherent superposition (and quantum computing (steps 2, 3)) in certain (grey) tubulins. Step 3 (in coherence with other microtubule tubulins) meets critical threshold related to quantum gravity for self-collapse (Orch OR). A conscious event (Orch OR) occurs in the step 3 to 4 transition. Tubulin states in step 4 are non-computably chosen in the and evolve classical to neural function, Schematic collapse, by computing regulate graph (b) of proposed quantum coherence (number of tubulins) emerging versus time in microtubules. Area under curve connects superposed mass energy E with collapse time T in accordance with E ? h/T. E may be expressed as Nt, the number of tubulins whose mass separation (and separation of underlying spacetime) for time T will self-collapse. For T = 25 ms (e.g. 40 Hz oscillations), Nt = 2 x 1010 tubulins. is thus

phase of classical suggested: computation; (1) a 'sol' liquid communicative and (2) a 'gel' solid-state isolated quantum computing phase. events may also be shielded either in hollow microtubule cores or Key quantum

Health has shown that microtubules are surrounded by a pH-dependent condensed phase of charge and counter ions which can protect them from thermal noise. The phase is produced by the C-terminal end of the amino acid chain which comprises tubulin and which protrudes externally with a surplus of eight negative charges. Cations (calcium, magnesium, etc.) balance the negative charges producing a plasma? like isolation of sufficient 'Bjerrum length' (ca. 0.7 nm) to defeat kT, the energy of thermal background. The pH dependence may explain the evanescent nature of the 'clear zone' surrounding microtubules (Stebbings & Hunt 1982). Microtubules could also implement quantum error-correction codes. The structure of microtubules involves helical pitches which repeat at Fibonacci series periodicities of 3, 5, 8 and 13 rows. Errorcorrection mechanisms could propagate along these pathways, operating on qubits following longitudinal or other pathways. Phil Trans. R. Soc. Lond. A (1998)

S. Hameroff

1880

Figure 8. Three interconnected microtubules (1) enter phase of (white) act in gelation-quantum communication isolation (2-3) alternating with phase of solution-environmental (1,4). Cycles may occur rapidly, e.g. 25 ms intervals (40 Hz).

pockets intraprotein gases are known to act). Fea? hydrophobic (where anaesthetic in the internal coherence cell environment is supported by the sibility of quantum radical pairs which become sepa? that quantum observation spins from biochemical in cytoplasm rated retain their correlation (Walleczek 1995). states do occur within neuronal But if isolated cells, could quantum cytoplasmic and synapses to spread macroscopically the membranes throughout they traverse involves quantum brain? One possibility tunnelling through gap junctions, primitive windows between neurons and glia (figure 9). Cells interconnected electrotonic by gap fire like which one neuron' form networks 'behaving giant synchronously, junctions for synchronized neural activity such et al. 1991), and possibly accounting (Kandel which separate neural pro? as coherent 40 Hz (Jibu 1990). Unlike chemical synapses are 3.5 nm, within range for quantum cesses by 30-50 nm, gap junction separations but distributed, high levels of gap junctions appear unevenly Widespread, tunnelling. networks k Abelson in the thalamus and cortex (Micevych 1991). Thalamo-cortical neurons with sol-gel phases coupled to synchronized 40 Hz of gap junction-connected brain isolate states across volumes. could quantum large transiently activity

4.

Quantum

computing Penrose?Hameroff

with

objective Orch OR

reduction:

the

model

k Hameroff and details of the Orch OR model are given in Penrose Full rationale are listed here. & Penrose Hameroff and points Key (1996a, b). (1995) tubulin proteins in brain microtubules are states of individual (1) Conformational in events London forces to internal sensitive hydrophobic quantum pockets) (e.g. in both classical and quan? with other tubulins interact and able to cooperatively tum computation automata) (microtubule (figures 3-6). Classical phase computation activities and other neural membrane chemical synapses regulates (e.g. figure 2). coherent computation supporting quantum emerges superposition (2) Quantum subunit tubulins pockets of microtubule among London forces in hydrophobic (e.g. in In this a manner described phase, computation quantum by Frohlich (1968, 1975)). to the Schrodinger evolves equation according linearly among tubulins (quantum Actin and a condensed microtubule charge phase surrounds, gelation automata). microtubules and insulates phase. during the quantum isolates, Phil Trans. R. Soc. Lond. A (1998)

Quantum

computation

in brain microtubules?

1881

Figure 9. Schematic diagram of proposed quantum coherence in microtubules in three dendrites interconnected by tunnelling through gap junctions. Within each neuronal dendrite, microtubule-associated protein (MAP) attachments breach isolation and prevent quantum coherence; MAP attachment sites thus act as 'nodes' which tune and orchestrate quantum oscillations and set possibilities and probabilities for collapse outcomes (orchestrated objective reduction: Orch OR). Gap junctions may enable quantum tunnelling among dendrites in macroscopic quantum states.

quantum phase in neural micro? (3) The proposed superposition/computation tubules to preconscious which until the continues corresponds processing, (implicit) threshold for Penrose's reduction is reached. reduction objective Objective (OR)?a discrete event?then occurs (figures 5-7), and post-OR tubulin states (chosen nonto regulate microtubule automata and proceed by classical synapses computably) other neural membrane activities. The events are suggested to be conscious have (to for reasons that relate to a merger of modern physics and philo? qualia, experience) sophical pan-experientialism (see ? 5). A sequence of such events gives rise to a stream of consciousness. states link to those in other neurons and glia by tun? quantum (4) Microtubule nelling (Jibu

coherent membranes through gap junctions photons traversing (or quantum k Yasue 1995; Jibu et al. 1994, 1996)). This spread enables macroscopic quan-

Phil. Trans. R. Soc. Lond. A (1998)

S. Hameroff

1882

cells turn states in networks of gap junction-connected volumes brain large (figure 9). and possibilities for preconscious (5) Probabilities influenced

(neurons

and glia) throughout

are superpositions quantum of microtubule-associated by biological including which tune and 'orchestrate' oscillations quantum (figure 9). ('MAPs'), term the self-tuning 'orchestrated' OR process in microtubules' objective feedback

attachments

proteins We thus reduction?Orch OR. events of preconscious and duration Orch OR pro? may be of variable intensity (6) ? E time of, for exam? from for a preconscious processing cessing. Calculating h/T, 40 Hz), E is roughly the superposition/separation ple, T = 25 ms (thalamocortical For T = 100 ms (alpha EEG), E would involve 5 x 109 tubulins. of 2 x 1010 tubulins. For T ? 500 ms (e.g. shown by Libet et al. (1979) as a typical preconscious process? E is equivalent to IO9 tubulins. Thus 2 x 1010 ing time for low-intensity stimuli), in isolated quantum for 25 ms (or 5 x IO9 coherent superposition tubulins maintained for 100 ms, or 109 tubulins for 500 ms, etc.) will self-collapse tubulins (Orch OR) and a conscious event. neuron is estimated to contain about Each brain IO7 tubulins (Yu k Bass (7) then Orch OR of of each neuron's tubulins became 10% coherent, 1994). If, say, neurons would be required within roughly 20 000 (gap junction-connected) tubulins for a 25 ms conscious event, 5000 neurons for a 100 ms event, or 1000 neurons for a 500 ms event, etc. encoded in information Orch OR event binds superposed (8) Each instantaneous moment: a reaches threshold at a particular whose net displacement microtubules is thus bound into a 'now' event. As quantum variety of different modes of information are irreversible in time, cascades of Orch OR events present a forward state reductions flow of time and 'stream of consciousness'. issues of of the Orch OR model to enigmatic In the following section, applications elicit

will be examined.

consciousness

5.

Orch

OR

and

enigmatic

features

of consciousness

in ? 1: (1) the nature of of consciousness were described features enigmatic from and sense of in vision 'self; (3) transition experience; subjective (2) 'binding' will. and free to consciousness; processing preconscious (5) (4) non-computability; Can Orch OR address these issues? is the most difficult. How does the brain The problem experience (1) of subjective raw sensations? There have always been two types of and feelings 'qualia', produce whereas was created by the cerebrum, that consciousness answers. Socrates argued as a fundamental saw conscious and other Plotinus experience 'pan-psychists' Thales, Five

feature of reality. consciousness follow Socrates: Modern functionalists/computationalists generally others find this in neural networks. the brain's from However, complexity emerges and are driven to embrace view alone unable to accommodate experience, subjective or pan-experientialism. some form of pan-psychism, con? after ancient pan-psychists, rudimentary Spinoza Following (1677) assigned as an and objects, and Leibniz to all particles sciousness (1766) saw the universe each having a primitive units ('monads'), number of fundamental infinite psycho? as who saw the universe logical being. Whitehead (1929) was a process philosopher Phil Trans. R. Soc. Lond. A (1998)

Quantum

computation

in brain microtubules?

1883

4

Figure 10. A spin network. Introduced by Penrose (1971) as a quantum-mechanical description of the geometry of space, spin networks describe a spectrum of discrete Planck-scale volumes and configurations (with permission from Smolin (1997) and Rovelli k Smolin (1995a, 6)). Average length of each edge is the Planck length (10~~33 cm). Numbers indicate quantum-mechanical spin along each edge. Each quantum state of spacetime is a particular spin network (Smolin 1997). of events. He described monads with spon? fundamentally being comprised dynamic as and mind-like of them entities limited duration taneity creativity, interpreting of Each to bears a occasion, Whitehead, according quality ('occasions experience'). akin to 'feeling' by virtue of occurring in a 'wider field of protoconscious experience'. Could this 'wider field' be the universe itself? Could protoconscious exist experience in empty space? What is empty space? Democritus described empty space as a true void, whereas Aristotle saw a background filled with substance. Maxwell's 'plenum' 19th-century 'luminiferous ether' sided with Aristotle, but attempts to detect the ether failed and Einstein's special relativity agreed with Democritus: empty space was an abso? lute void. However, with Einstein's its curved general relativity space and distorted to reverted a endowed metric. geometry richly plenum?the spacetime At very small scales spacetime is not smooth, but quantized. occurs at Granularity the incredibly small 'Planck scale' (10~~33 cm, 10~43s) which Penrose (1971) portrays as a dynamical of quantum spider-web spin networks (figure 10) (Rovelli k Smolin describe vol? Planck-scale 1995a, b; Smolin 1997). Spin networks spectra of discrete umes and configurations which dynamically evolve and define spacetime geometry. Planck-scale could provide Whitehead's basic field of protoconscious spin networks Whitehead occasions are quantum state Shimony experience. (1993) has suggested As described in ? 1, Penrose's reductions. are reductions objective (quantum state) bubble-like and collapse in fundamental separations spacetime geometry extending Phil. Trans. R. Soc. Lond. A (1998)

1884

Reduction

Superposition Figure 11. Schematic quantum computation in spacetime curvature for three mass distributions (e.g. tubulin conformations in figure 5) which begin (a) in initial classical states, then enter iso? lated quantum superposition in which all possible states coexist. After reduction, one particular classical outcome state is chosen (6).

to the level of spin networks. Figure 11 illustrates quantum superposition of spacetime Orch OR events could be Whitehead and objective reduction geometry. occasions of experience. view consistent with modern physics, Planck-scale In a pan-psychist spin networks values. encode protoconscious experience ('fundamental') (qualia) as well as Platonic of quantum varieties of Particular spin geometry convey particular configurations and aesthetics. The proposed Orch OR events meaning protoconscious experience, in an experiential downward to processes Planck-scale occur in the brain, extending downward

involves brain activities to The basic idea is that consciousness coupled reality. self-organizing ripples in fundamental information link to macroscopic biol? How can near-infinitesimal protoconscious in an the Orch OR be As described 6, emergent phenomenon ? process may ogy? London forces in hydrophobic of in quantum mediated pockets through geometry tubulin and other proteins. is (2) binding, and it is poten? The second difficult issue related to consciousness medium.

Penrose states (see, for example, nature of quantum resolved by the unitary was a of Bose-Einstein con? that feature Marshall binding suggested 1987). (1989) In the Orch OR model, an densates among certain of the brain's neural proteins. dis? whose net mass/spacetime information instantaneous event binds superposed moment: different modes and time-scales at a particular reaches threshold placement

tially

of information are bound into a unitary 'now' event. from preconscious Problem processing (3) is the transition Phil Trans. R. Soc. Lond. A (1998)

to consciousness

itself.

Quantum

computation

in brain microtubules?

1885

In Orch OR, preconscious is equivalent to the quantum processing superposition and then abruptly of interact Potential possibilities phase quantum computation. a slight quake in spacetime As quantum state reductions are self-collapse, geometry. flow of cascades of a forward time Orch OR events irreversible, present subjective and 'stream of consciousness'. is potentially to cog? with objective reduction applicable Quantum computation like face recognition choice may require Functions and volitional nitive activities. a series of conscious consider illustration

events single

at intermediate arriving Orch OR events in these

For the purpose of solutions. two types of cognitive activities

(figure 12). or Carol? woman's face. Is she Amy, Betty Imagine you briefly see a familiar in For Possibilities a example, may superpose quantum computation. during 25 ms with information of preconscious occurs computation processing, quantum (Amy, states of microtubule tubulin subBetty, Carol) in the form of 'qubits'?superposed As threshold reduction is reached, an for objective units within groups of neurons. conscious event occurs. The superposed instantaneous tubulin qubits reduce to defi? nite states, becoming bits. Now, you recognize that she is Carol! (an immense number in a human brain's IO19 tubulins). of possibilities could be superposed for example, In a volitional act possible choices may be superposed. Suppose, you are selecting dinner from a menu. During preconscious shrimp, sushi and processing, in are a As for threshold reduction superposed pasta quantum computation. objective is reached, the quantum state. A choice is made. state reduces to a single classical You'll have sushi! How does the choice actually criteria be described occur? Can the selection by a These deterministic relate to algorithm? questions problems (4), non-computability, and (5), free will. The problem in understanding free will is that our actions seem neither totally nor random (probabilistic). deterministic What else is there in nature? As previously in OR (and Orch OR) the reduction are neither deterministic outcomes described, nor probabilistic, but 'non-computable'. The microtubule quantum superposition evolves linearly (analogous to a quantum at the instant but is influenced computer) of collapse variables in by hidden non-local logic inherent (quantum-mathematical are fundamental The outcomes or spacetime possible limited, probabili? geometry). ties set ('orchestrated'), feedback MAPs (figure 9)). by neurobiological (in particular, The precise outcome?our free-will actions?are chosen by effects of the hidden logic on the quantum at the of reduction. system poised edge objective a sailboat A Consider free for will. sailor the sail in a certain way; sets analogy the direction the boat sails is determined the wind on the sail. Let's the action of by the sailor is a a non-conscious robot-zombie run which pretend quantum computer by is trained and programmed to sail. Setting and adjusting of the sail, sensing the wind and position and so forth are algorithmic and deterministic, and may be analogous to the preconscious and intensity quantum computing phase of Orch OR. The direction of the wind (seemingly or unpredictable) to Planckcapricious, may be analogous scale hidden non-local variables 'Platonic' quantum-mathematical logic inherent (e.g. in spacetime The or outcome the direction boat choice, sails, the geometry). (the on the deterministic sail settings acted point on shore that it lands upon) depends on repeatedly wind. Our 'free will' actions could by the apparently unpredictable be the net result of deterministic acted on by hidden processes quantum logic at Phil Trans. R. Soc. Lond. A (1998)

1886

S. Hameroff Orch OR

(a)

Tubulin States

(b)

Fundamental Spacetime Geometry Space?time S wife*.-Separation

6[I*11 ^-"""""""^ (io-

(c)

Face Recognition

Carol

Amy (d)

Volitional Choice

Sushi

Shrimp

Figure 12. For description

Phil. Trans. R. Soc. Lond. A (1998)

see opposite.

Quantum Orch OR event. deterministic fashion, to ourselves. each

This but

computation can

explain

occasionally

in brain microtubules?

1887

do things in an orderly, why we generally our actions or thoughts are surprising, even

intrinsic to brain function such as the proposed quantum computation Biological Orch OR model can in principle address difficult issues related to consciousness.

6.

Are

microtubules

quantum

computers?

connected to fundamental The idea of biological quantum computation geometry The Planck is 24 orders of smaller than the seems far-fetched. magnitude length volumes of an atom. Approximately 1078 discrete Planck-scale diameter correspond to the space occupied by one protein, and 10105 such volumes to a brain. The energy of one proposed Orch OR (e.g. 25 ms) is only 10~28 J, or 10~~10 eV, whereas the energy noise (kT) is much larger at 10~"4 eV. How could near-infinitesimally of thermal effects in biological Orch have macroscopic systems? small, weak and fast processes in which of be viewed as a nonlinear OR may phenomenon particular configurations influence times second to confor? volumes 40 1088 Planck-scale protein per emerge balanced London forces. Conditions mation through supporting quantum delicately evolved consciousness would favour survival and have naturally states and primitive 19986). (Hameroff In another paper in this volume, Tuszyhski k Brown review the physics of micro? a critique and provide of the Orch OR proposal. tubules They raise several issues in A. which are discussed Appendix microtubules and technological Here similarities are sought between proposals A has been realizable' for quantum computer computation. quantum 'potentially Com? described systems. by Lloyd (1993) as '... arrays of weakly coupled quantum is effected by ... a sequence of electromagnetic pulses that induce transitions putation states ... in a crystal lattice'. between locally defined quantum In the Orch OR model, the microtubule to Lloyd's crystal corresponds assembly is proposed lattice. Rather than trapped ions or nuclear spins, quantum superposition states of tubulins, and the role of pulsed to occur at the level of conformational Frohlich excitations. transitions played by coherent in terms of a The Orch OR proposal to technological schemes may be compared 1996; DiVincenzo 'figure of merit M' (table 1) (Barenco 1995). M is the time Tdecohere until decoherence divided by the time ?eiem of each elementary and gives operation, the number of operations allowable per computational unit before decoherence. With and Tdecohere of, for example, 100 ms (EEG alpha), teiem of 109 s (Frohlich frequency) Figure 12. An Orch OR event (continued from figure 4). (a) (left), three tubulins in quantum super? position prior to 25 ms Orch OR. After reduction (right), particular classical states are selected, (b) Fundamental spacetime geometry view. Prior to Orch OR (left), spacetime corresponding with three superposed tubulins is separated as Planck-scale bubbles: curvatures in opposite directions. The Planckscale spacetime separations S are very tiny in ordinary terms, but relatively large mass movements (e.g. hundreds of tubulin conformations, each moving from 10_6 to 0.2 nm) indeed have precisely such very tiny effects on the spacetime curvature. A critical degree of separation causes Orch OR and an abrupt selection of single curvatures (and a particular geometry of experience), (c) Cognitive facial recogni? tion. A familiar face induces superposition (left) of three possible solutions (Amy, Betty, Carol) which 'collapse' to the correct answer Carol (right), (d) Cognitive volition. Three possible dinner selections (shrimp, sushi, pasta) are considered in superposition (left), and collapse via Orch OR to the choice of sushi (right). Phil. Trans. R. Soc. Lond. A (1998)

1888

S. Hameroff Table 1. Figure of merit M for different proposed quantum computing technologies microtubules (modified from Barenco (1996) and DiVincenzo (1995)) technology

?eiem (s)

Tdecohere (s)

Ma

Mossbauer nucleus electrons GaAs electrons Au trapped ions optical cavities electron spin electron quantum dot nuclear spin superconductor islands microtubule tubulins

IO"19 IO-13 IO"14 IO"14 IO"14 IO"7 IO"6 IO"3 10~9 IO"9

IO"10 IO-10 IO-8 IO"1 IO"5 IO"3 IO"3 IO4 IO3 IO"1

IO9 IO3 IO6 IO13 IO9 IO4 IO3 IO7 IO6 IO8

aIn 'units' of predecoherence the Orch conscious

OR model yields event occurs.

operations

and

per qubit.

a respectable

M of IO8 operations

per tubulin

before

a

seem to to the proposals According put forth in the Orch OR model, microtubules If be well designed quantum computers. so, technological ideally designed) (perhaps efforts can possibly mimic some of nature's design principles such as cylindrical lat? tice automata of and alternating isolation and communication. The massive phases and specific microtubule lattice follow? parallelism geometry (e.g. helical patterns Fibonacci error correction. However, quantum series) may also facilitate will emulate reduction be to which, it is argued, hard-pressed technology objective envisioned is required for consciousness. comput? Presently technological quantum nuclei or other small entities. To ers will implement of ions, electrons, superposition and useful time-scale, in a reasonable a fairly large super? achieve objective reduction While such a task seems formidable, will be required. posed mass (i.e. nanograms) with reduction it is possible. computation Quantum objective may hold the only ing

the

computers. promise for conscious turns out to be correct (and of whether or not the Orch OR proposal Regardless it is testable; unlike most theories of consciousness Appendix B), it is the type of needed to address the multilevel problem of consciousness. approach transdisciplinary Thanks to Roger Penrose, who does not necessarily endorse the newer proposals, for illustrations and Carol Ebbecke for expert assistance.

Appendix

A.

Reply

to

Tuszyfiski

Dave Cantrell

& Brown

k Brown review the physics of micro? article in this volume, In another Tuszyfiski tubules and give a critique of the Orch OR proposal. They raise several issues dis? cussed here. Gravitational processes. by the remaining effects should be entirely overshadowed to thermal noise The energy from an Orch OR event is indeed very small compared mech? drown in an aqueous medium. Isolation/insulation (kT) and would seemingly from thermal noise or any type of anisms are thus required to shield microtubules that quantum-coherent decoherence. The Orch OR model suggests environmental Phil. Trans. R. Soc. Lond. A (1998)

Quantum

computation

in brain microtubules?

1889

which are immediately surrounded occurs in microtubules by an insu? and encased in actin condensation gelation lating charge (cyclically) (?3). Cyclical and isolated isolation allows for alternating phases of communication (input-output) superposition

computation. quantum microtubule subunits must also be sensitive to In addition to isolation, (tubulins) other influences from tubulins and influences superposed non-computable quantum In questioning the robustness of proposed in Planck-scale effects, quantum geometry. k Brown ascribe the gravitational energy for a tubulin protein in Orch OR Tuszyhski between two masses given by the standard where G is to be the attraction Gm2/r, r the ra is the mass of tubulin and is the distance between the gravitational constant, This would k Brown take to be the radius of tubulin. two masses which Tuszyfiski two adjacent the gravitational attraction between tubulins describe accurately (or and yields an appropriately small energy of 10~27 eV. However, tubulin monomers), E of a superposed the relevant self-energy energy in Orch OR is the gravitational from itself by distance mass ra separated a, given (for complete by E = separation) this energy for three cases: In Hameroff k Penrose (1996) we calculated Gm2/a. its radius; of the entire one-tenth protein by separation (2) complete (1) partial ? of nuclei 2.5 fermi lengths); level each atomic the at protein's separation (a (3) at the level of each protein's nucleons complete separation (a = 0.5 fermi lengths). at the level of atomic nuclei, roughly Of these, highest energies were for separation at the level of, say, atoms or amino acids separation (although x 2 are involved in each proposed As 1010 tubulins roughly may yield higher energy). 25 Orch OR event (e.g. for superpositions lasting ms) the energy is in the order of or still IO-28 10~10 J, extremely tiny (kT is about 10~4 eV). However, eV, roughly within one Planck-time of 10~43 s. This the 10~28 J energy emerges abruptly, e.g. an of IO13 1 kW per to instantaneous W be jab roughly equivalent may (Js_1), 10~21 eV per tubulin

tubulin. The size of the tubulin protein is probably too large to make quantum effects easily size proteins such as tubulin Nanometre sustainable. (8x4x4 nm3) may be the for interface scale a 1991; Conrad 1994). optimal (Watterson quantum/macroscopic lack causal efficacy of structural conformational Smaller biomolecules protein changes functions. would be insufficiently for a host of biological Larger molecules responsible effects. distances effects are expected to involve of 10 A (Inm), larger Conformational distance than those called for in the Orch OR model. The superposition separation IO-6 nm in the case cited) is indeed much smaller than (e.g. one atomic nucleus, 1 nm. As described which in ? 2, proteins are conformational changes may approach their conformation is nonlinear and unstable, through relatively regulated 'quakes' sensitive

to quantum

London forces. quantum-level make it extremely to defend the temperature requirements difficult Physiological due to A biological the thermal noise. use of the quantum persistence regime of be from a feature nature may thermal state must noise, quantum isolated/insulated actin gelation and condensed have evolved in cytoplasmic charged layers (?3). Some evidence supports biological quantum states (e.g. Tejada et al. 1996; Walleczek 1995). mediated

through

thermal in biological to the Frohlich mechanism, may According energy systems mode. condense to a coherent ... we doubt that ... microtubules are extremely sensitive to their environment in ? 3, nature may have solved the problem can be shielded. As described microtubules Phil Trans. R. Soc. Lond. A (1998)

S. Hameroff

1890

isolation and communication by alternating cytoplasmic phases of solution and to sensitive environment, gelation classical) ('gel', solid, shielded/in? ('sol', liquid, Thus microtubules can be both sensitive to their environment sulated, quantum). and isolated/shielded ('gel' phase). ('sol' phase) ... states two (or possibly by a of tubulin are separated more) conformational as GTP stimulus sizable potential barrier which again requires an external (such of both

which can it. Tubulin has numerous conformations to overcome possible hydrolysis) is The two-state tubulin model a simpli? without GTP interchange hydrolysis (?2). of tubulin has recently been clarified (Nogales et al. 1998) so fication. The structure will soon be available. molecular simulations ... the 500 ms preconscious time may be directly related to the action processing axon the travel the transmis? time along plus potential lag time in synaptic refractory rather than to the quantum collapse time. In the Orch OR model the 'quantum related time-intervals collapse time' T is chosen to match known neurophysiological E to preconscious the and mass related processes; gravitational self-energy may then we have used 25 ms (e.g. in coherent 40 Hz oscillations), be calculated. For example, and 500 ms (e.g. Libet's preconscious threshold for 100 ms (e.g. EEG alpha rhythm) sion

low-intensity If quantum

sensory

stimuli). with preconscious correlates then dendritic superposition processing, axonal are to activities than be relevant to consciousness likely firings) (more (e.g. in dendrites are of mixed polarity Pribram 1991). Microtubules (unlike those in an arrangement conducive to cooperative computation. axons), raise valid milieu Brown states in a biological k objections; quantum Tuszynski to at be nature have evolved first However, specific appear unlikely. may glance conditions macroscopic

for isolation, thermal state. quantum

Appendix Here

major of the Orch Neuronal

B.

Testable

and

screening

predictions

assumptions (bold) and corresponding OR model are listed. microtubules

sensitivity (1) Synaptic activities in both pre-synaptic (2) Actions of psychoactive

are and

directly

Life

amplification.

necessary

of the

Orch

testable

for

itself

OR

may

be a

model

predictions

(numbered)

consciousness.

correlate

with cytoskeletal plasticity architecture/ and post-synaptic neuronal cytoplasm. involve neuronal micro? antidepressants drugs including

tubules. microtubule-stabilizing/protecting (3) Neuronal and other conditions. heimer's disease, ischaemia Microtubules

communicate

by cooperative

drugs

may

dynamics

(e.g. Vos et al. 1992) will demonstrate (4) Laser spectroscopy in microtubules. excitations in microtubule states networks vibrational (5) Dynamic

prove

useful

in Alz?

of tubulin

subunits.

coherent

gHz Frohlich

correlate

with

cellular

activity. of microtubule/cytoskeletal patterns (6) Stable of tubulin diversity ments) and intra-microtubule neural behaviour. Phil Trans. R. Soc. Lond. A (1998)

neurofilanetworks (including and states correlate with memory

Quantum

computation

in brain microtubules?

1891

to 'Bdendrites contain microtubules largely 'A-lattice' (compared (7) Cortical are preferable for information lattice' microtubules, A-lattice microtubules processing et al. 1995)). (Tuszynski Quantum

coherence

occurs

in microtubules.

in physics similar to the famous 'Aspect experiment' (which verified (8) Studies et al. 1982)) will demonstrate cor? non-local correlations quantum quantum (Aspect between microtubule subunit relations states on (a) the same separated spatially in in the same neuron; and (c) microtubules microtubule; (b) different microtubules different

neurons connected by gap junctions. with SQUIDs quantum (superconducting (9) Experiments such as those suggested by Leggett (1984) will detect phases in microtubules. from microtubules. will be detected photons (10) Coherent Microtubule ing

coherence

quantum

isolation

requires

interference of quantum

devices) coherence

of surround?

by cycles

actin-gelation.

in cortical dendrites and other brain areas are inter? microtubules (11) Neuronal in act cross-linked surrounded gels. by tightly mittently in neuronal occur concomitantly and dissolution cytoplasm (12) Cycles of gelation Hz activities in dendrites). with membrane electrical 40 activity (e.g. synchronized ions microtubules are The regulated by calcium cycles surrounding sol-gel (13) associated with microtubules. released and reabsorbed calmodulin by Macroscopic thousands

quantum of distributed

coherence neurons

occurs microtubules in hundreds/ among and glia linked by gap junctions.

of cortical neu? link synchronously firing networks gap junctions (14) Electrotonic rons and thalamocortical networks. occurs across gap junctions. tunnelling (15) Quantum microtubule states in different occurs between subunit correlation (16) Quantum in different neurons connected 'Aspect experiment' by gap junctions (the microtubule neurons). The

amount

proportional

involved tissue of neural to the event time by E =

of neural mass involved (17) The amount scious event (as measurable by near-future to the is inversely proportional preconscious

in

a conscious

event

is

inversely

h/T. in a particular task or con? cognitive in brain imaging advances techniques) reaction time (e.g. visual perception,

times). An

isolated,

quantum

unperturbed

system

self-collapses

according

to E =

h/T. to of

will self-collapse superpositions according quantum technological (18) Isolated E = h/T. discussions of such experiments involving superposition (Preliminary and Anton Zeilinger.) Roger Penrose crystals have begun between Microtubule-based

cilia/centriole

(19) Microtubule-based connect with retinal glial

structures

are

quantum

cilia in rods and cones directly detect cell microtubules via gap junctions.

Phil Trans. R. Soc. Lond. A (1998)

optical visual

devices.

photons

and

S. Hameroff

1892 A

critical

rudimentary tion.

of cytoskeletal degree had consciousness,

assembly, significant

coinciding on impact

with the

the rate

of onset of evolu?

with present-day biology will show that organ? (20) Fossil records and comparison isms which emerged during the early Cambrian period with onset roughly 540 mil? lion years ago had critical degrees of microtubule-cytoskeletal and size, complexity for quantum isolation capability (e.g. tight actin gels, gap junctions (see Hameroff 19986)). References Andreu, J. M. 1986 Hydrophobic interaction of tubulin. Ann. N. Y. Acad. Sci. 466, 626-630. realization of Einstein-PodolskyAspect, A., Grangier, P. k Roger, G. 1982 Experimental a new violation of Bell's inequalities. Phys. Rev. Lett. Rosen-Bohm Gedankenexperiment: 48, 91-94. Baldwin, R. L. 1994 Matching speed and stability. Nature 369, 183-184. Barenco, A. 1996 Quantum physics and computers. Contemp. Phys. 37, 375-389. Beck, F. k Eccles, J. C. 1992 Quantum aspects of brain activity and the role of consciousness. Proc. Natn. Acad. Sci. USA 89, 11357-11361. Benioff, P. 1982 Quantum mechanical Hamiltonian models of Turing machines. J. Stat. Phys. 29, 515-546. Bennett, C. H. 1995 Quantum information and computation. Physics Today (October), pp. 2430. Bohm, D. k Hiley, B. J. 1993 The undivided universe. New York: Routledge. Chalmers, D. J. 1996 The conscious mind. In search of a fundamental theory. Oxford University Press. Churchland, P. S. 1986 Neurophilosophy: toward a unified science of the mind-brain. Cambridge, MA: MIT Press. Churchland, P. S. k Sejnowski, T. J. 1992 The computational brain. Cambridge, MA: MIT Press. Cianci, C, Graff, D., Gao, B. k Weisenberg, R. C. 1986 ATP-dependent gelation-contraction of microtubules in vitro. Ann. N. Y. Acad. Sci. 466, 656-659. of superpositional effects through electronic-conformational Conrad, M. 1994 Amplification interactions. Chaos Solitons Fractals 4, 423-438. Crick, F. & Koch, C. 1990 Towards a neurobiological theory of consciousness. Semin. Neurosci. 2, 263-275. Cruzeiro-Hansson, L. 1996 Dynamics of a mixed quantum-classical system of finite temperature. Europhys. Lett. 33, 655-659. Cruzerio-Hansson, L. k Takeno, S. 1997 Davydov model: the quantum, mixed quantum-classical and full classical systems. Phys. Rev. E 56, 894-906. Dayhoff, J., Hameroff, S., Lahoz-Beltra, R. k Swenberg, C. E. 1994 Cytoskeletal involvement in neuronal learning: a review. Eur. Biophys. Jl 23, 79-83. Dennett, D. 1991 Consciousness explained. Boston, MA: Little, Brown and Company. Dermietzel, R. k Spray, D. C. 1993 Gap junctions in the brain: where, what type, how many and why? Trends Neurosci. 16, 186-192. Deutsch, D. 1985 Quantum theory, the Church-Turing principle and the universal quantum computer. Proc. R. Soc. Lond. A 400, 97-117. Proc. R. Deutsch, D. & Josza, R. 1992 Rapid solution of problems by quantum computation. Soc. Lond. A 439, 553-556. Devlin, T. M. 1992 Textbook of biochemistry with clinical correlations, 3rd edn, pp. 74-75. New York: Wiley. Phil Trans. R. Soc. Lond. A (1998)

Quantum

computation

in brain microtubules?

1893

DiVincenzo, D. 1995 Phys. Rev. A 50, 1015. Engelborghs, Y. 1992 Dynamic aspects of the conformational states of tubulin and microtubules. Nanobiol. 1, 97-105. Feynman, R. P. 1986 Quantum mechanical computers. Found. Phys. 16, 507-531. Franks, N. P. k Lieb, W. R. 1982 Molecular mechanisms of general anaesthesia. Nature 300, 487-493. Franks, N. P. k Lieb, W. R. 1985 Mapping of general anaesthetic target sites provides a molecular basis for cut-off effects. Nature 316, 349-351. Frohlich, H. 1968 Long-range coherence and energy storage in biological systems. Int. J. Quan? tum Chem. 2, 641-649. Frohlich, H. 1970 Long range coherence and the actions of enzymes. Nature 228, 1093. Frohlich, H. 1975 The extraordinary dielectric properties of biological materials and the action of enzymes. Proc. Natn. Acad. Sci. USA 72, 4211-4215. R. J. 1991 Direct observation of Genberg, L., Richard, L., McLendon, G. k Dwayne-Miller, global protein motion in hemoglobin and myoglobin on picosecond time scales. Science 251, 1051-1054. Genzel, L., Kremer, F., Poglitsch, A. k Bechtold, G. 1983 Relaxation processes on a picosecond time scale in hemoglobin and poly observed by millimeter-wave spectroscopy. Biopolymers 22, 1715-1729. Glanz, J. 1997 Force-carrying web pervades living cell. Science 276, 678-679. Gray, J. A. 1998 Creeping up on the hard question of consciousness. In Toward a science of consciousness. II. The second Tucson discussions and debates (ed. S. Hameroff, A. Kaszniak k A. Scott), pp. 279-291. Cambridge, MA: MIT Press. Grundler, W. k Keilmann, F. 1983 Sharp resonances in yeast growth prove nonthermal sensi? tivity to microwaves. Phys. Rev. Lett. 51, 1214-1216. Halsey, M. J. 1989 Molecular mechanisms of anaesthesia. In General anaesthesia (ed. J. F. Nunn, J. E. Utting k B. R. Brown Jr), 5th edn, pp. 19-29. London: Butterworths. Hameroff, S. R. 1994 Quantum coherence in microtubules: a neural basis for emergent conscious? ness. J. Conscious. Stud. 1, 91-118. Hameroff, S. R. 1997 Quantum computing in microtubules. An intraneural correlate of con? sciousness? Jap. Bull Cognitive Sci. 4, 67-92. Orch OR model of con? Hameroff, S. 1998a Funda-mental geometry: the Penrose-Hameroff sciousness. In The geometric universe - science, geometry and the work of Roger Penrose (ed. S. A. Huggett, L. J. Mason, K. P. Tod, S. Tsou k N. M. J. Woodhouse), pp. 103-127. Oxford University Press. Hameroff, S. 19986 Did consciousness cause the Cambrian evolutionary explosion? In Toward II. The second Tucson discussions and debates (ed. S. Hameroff, a science of consciousness. A. Kaszniak k A. Scott), pp. 421-437. Cambridge, MA: MIT Press. Hameroff, S. 1998c 'More neural than thou': reply to Churchland's 'Brainshy'. In Toward a science of consciousness. II. The second Tucson discussions and debates (ed. S. Hameroff, A. Kaszniak k A. Scott), pp. 197-213. Cambridge, MA: MIT Press. Is the conscious mind subtly connected to a basic level Hameroff, S. 1998c? 'Funda-Mentality'. of the universe? Trends Cognitive Sci. 2, 119-127. Hameroff, S. R. k Penrose, R. 1996a Orchestrated reduction of quantum coherence in brain I. The first microtubules: a model for consciousness. In Toward a science of consciousness. Tucson discussions and debates (ed. S. R. Hameroff, A. Kaszniak k A. C. Scott), pp. 507-540. Cambridge, MA: MIT Press. Hameroff, S. R. k Penrose, R. 19966 Conscious events as orchestrated spacetime selections. J. Conscious. Stud. 3, 36-53. Hameroff, S. R. k Watt, R. C. 1982 Information processing in microtubules. J. Theor. Biol. 98, 549-561. Phil. Trans. R. Soc. Lond. A (1998)

1894

S. Hameroff

Hameroff, S. R. k Watt, R. C. 1983 Do anesthetics act by altering electron mobility? Anesth. Analg. 62, 936-940. Haroche, S. k Raimond, J.-M. 1996 Quantum computing: dream or nightmare? Physics Today (August), pp. 51-52, Hirokawa, N. 1991 Molecular architecture and dynamics of the neuronal cytoskeleton. In The neuronal cytoskeleton (ed. R. Burgoyne), pp. 5-74. New York: Wiley. Hunt, C. k Stebbings, H. 1994 Role of MAPs and motors in the bundling and shimmering of native microtubules from insect ovarioles. Cell Motility Cytoskeleton 27, 69-78. Jibu, M. 1990 On a heuristic model of the coherent mechanism of the global reaction process of a group of cells. Bussei Kenkyuu (Mater. Phys. Res.) 53, 431-436. (In Japanese.) Jibu, M. k Yasue, K. 1995 Quantum brain dynamics: an introduction. Amsterdam: John Ben? jamins. Jibu, M., Hagan, S., Hameroff, S. R., Pribram, K. H. k Yasue, K. 1994 Quantum optical coher? ence in cytoskeletal microtubules: implications for brain function. Biosystems 32, 195-209. Jibu, M., Pribram, K. H. k Yasue, K. 1996 From conscious experience to memory storage and retrieval: the role of quantum brain dynamics and boson condensation of evanescent photons. Int. J. Mod. Phys. B10, 1735-1754. Joliot, M., Ribary, U. k Llinas, R. 1994 Human oscillatory brain activity near 40 Hz coexists with cognitive temporal binding. Proc. Natn. Acad. Sci. USA 91, 11748-11751. Kandel, E. R, Siegelbaum, S, A. k Schwartz, J. H. 1991 Synaptic transmission. In Principles of neural science (ed. E. R. Kandel, J. H. Schwartz k T. M. Jessell), 3rd edn, pp. 121-134. New York: Elsevier. Karplus, M. k McCammon, J. A. 1983 Protein ion channels, gates, receptors. In Dynam? ics of proteins: elements and function (ed. J. King), pp. 263-300. Menlo Park, CA: Ben? jamin/Cummings. Leggett, A. J. 1984 Schrodinger's cat and her laboratory cousins. Contemp. Phys. 25, 582. Leibniz, G. W. 1768 Opera omnia (ed. L. Dutens) (6 volumes). Geneva. Libet, B., Wright Jr, E. W., Feinstein, B. k Pearl, D. K. 1979 Subjective referral of the timing for a conscious sensory experience. Brain 102, 193-224. Lloyd, S. 1993 A potentially realizable quantum computer. Science 261, 1569-1571. London, F. 1937 Trans. Faraday Soc. 33, 8. Maniotis, A. J., Bojanowski, K. k Ingber, D. E. 1997a Mechanical continuity and reversible chro? mosome disassembly within intact genomes removed from living cells. J. Cellular Biochem. 65, 114-130. of mechanical connections Maniotis, A. J., Chen, C. S. k Ingber, D. I. 19976 Demonstration between integrins, cytoskeletal filaments and nucleoplasm that stabilize nuclear structure. Proc. Natn. Acad. Sci. USA 94, 849-854. condensates. New Ideas Psychol. 7, 73, Marshall, I. N. 1989 Consciousness and Bose-Einstein 83. of Melki, R., Carlier, M. F., Pantaloni, D. k Timashefl, S. N. 1989 Cold depolymerization microtubules to double rings: geometric stabilization of assemblies. Biochem. 28, 9143-9152. Micevych, P. E. k Abelson, L. 1991 Distribution of mRNAs coding for liver and heart gap junction protein in the rat central nervous system. J. Comp. Neurol. 305, 96-118. Milloni, P. W. 1994 The quantum vacuum: an introduction to quantum electrodynamics. Boston: Academic Press. Miyamoto, S. 1995 Changes in mobility of synaptic vesicles with assembly and disassembly of actin network. Biochim. Biophys. Acta 1244, 85-91. Moravec, H. P. 1987 Mind children. Harvard University Press. Muallem, S., Kwiatkowska, K., Xu, X. k Yin, H. L. 1995 Actin filament disassembly is a sufficient final trigger for exocytosis in nonexcitable cells. J. Cell Biol 128, 589-598. Phil. Trans. R. Soc. Lond. A (1998)

Quantum

computation

in brain microtubules?

1895

Nagel, T. 1974 What is it like to be a bat? Phil. Rev. 83, 435-450. Neubauer, C, Phelan, A. M., Keus, H. k Lange, D. G. 1990 Microwave irradiation of rats at 2.45 GHz activates pinocytotic-like uptake of tracer by capillary endothelial cells of cerebral cortex. Bioelectromag. 11, 261-268. Nogales, E., Wolf, S. G. k Downing, K. H. 1998 Structure of the a/3 tubulin dimer by electron crystallography. Nature 391, 199-203. Penrose, R. 1971 In Quantum theory and beyond (ed. E. A. Bastin). Cambridge University Press. Penrose, R. 1987 Newton, quantum theory and reality. In 800 years of gravity (ed. S. W. Hawking k W. Israel). Cambridge University Press. Penrose, R. 1989 The emperor's new mind. Oxford University Press. Penrose, R. 1994 Shadows of the mind. Oxford University Press. Penrose, R. 1996 On gravity's role in quantum state reduction. Gen. Relativ. Grav. 28, 581-600. Penrose, R. 1997 On understanding understanding. Int. Stud. Phil. Sci. 11, 7-20. Penrose, R. k Hameroff, S. R. 1995 What gaps? Reply to Grush k Churchland. J. Conscious. Stud. 2, 99-112. condensation and liquid helium. Phys. Rev. 104, Penrose, O. k Onsager, L. 1956 Bose-Einstein 576-584. Pribram, K. H. 1991 Brain and perception. Hillsdale, NJ: Lawrence Erlbaum. Rasmussen, S., Karampurwala, H., Vaidyanath, R., Jensen, K. S. k Hameroff, S. 1990 Compu? tational connectionism within neurons: a model of cytoskeletal automata subserving neural networks. Physica D42, 428-449. Roitberg, A., Gerber, R. B., Elber, R. k Ratner, M. A. 1995 Anharmonic wave functions of proteins: quantum self-consistent field calculations of BPTI. Science 268, 1319-1322. Rovelli, C. k Smolin, L. 1995a Discreteness of area and volume in quantum gravity. Nucl. Phys. B442, 593-619. Rovelli, C. k Smolin, L. 19956 Spin networks in quantum gravity. Phys. Rev. D52, 5743-5759. Shimony, A., 1993 Search for a naturalistic world view. Natural science and metaphysics, vol. 2. Cambridge University Press. Shor, P. W. 1994 Polynomial time algorithms for discrete logarithms and factoring on a quantum computer. In Algorithmic number theory. 1st Int. Symp., ANTS-1 Proc. (ed. L. M. Adleman k M. D. Huang). Berlin: Springer. Singer, W., Gray, C, Engel, A., Konig, P., Artola, A. k Brocher, S. 1990 Formation of cortical cell assemblies. Cold Spring Harbor Symp. Quant. Biol. 55, 939-952. Smolin, L. 1997 Life of the cosmos. Oxford University Press. Spinoza, B. 1677 Ethica in Opera quotque reperta sunt (ed. J. van Vloten k J. P. N. Land), 3rd edn. Netherlands: Den Haag. Stapp, H. P. 1993 Mind, matter and quantum mechanics. Berlin: Springer. Cell Tissue Stebbings, H. k Hunt, C. 1982 The nature of the clear zone around microtubules. Res. 227, 609-617. Tejada, J., Garg, A., Gider, S., Awschalom, D. D., DiVincenzo, D. P. k Loss, D. 1996 Does macroscopic quantum coherence occur in ferritin? Science 272, 424-426. Tuszynski, J., Hameroff, S., Sataric, M. V., Trpisova, B. k Nip, M. L. A. 1995 Ferroelectric behavior in microtubule dipole lattices; implications for information processing, signaling and assembly/disassembly. J. Theor. Biol. 174, 371-380. Vernon, G. G. k Woolley, D. M. 1995 The propagation of a zone of activation along groups of flagellar doublet microtubules. Exp. Cell Res. 220, 482-494. Voet, D. k Voet, J. G. 1995 Biochemistry, 2nd edn. New York: Wiley. Vos, M. H., Rappaport, J., Lambry, J. Ch., Breton, J. k Martin, J. L. 1992 Visualization of coherent nuclear motion in a membrane protein by femtosecond laser spectroscopy. Nature 363, 320-325. Phil. Trans. R. Soc. Lond. A (1998)

1896

S. Hameroff

Wachsstock, D. H., Schwarz, W. H. k Pollard, T. D. 1994 Cross-linker dynamics determine the mechanical properties of actin gels. Biophys. Jl 66, 801-809. Walleczek, J. 1995 Magnetokinetic effects on radical pairs: a possible paradigm for understanding sub-kT magnetic field interactions with biological systems. In Electromagnetic fields: biological interactions and mechanisms (Advances in Chemistry 250) (ed. M. Blank). Washington, DC: American Chemical Society. Watterson, J. G. 1991 The interaction of water and proteins in cellular function. Prog. Molec. Subcell. Biol. 12, 113-134. Wheeler, J. A. 1990 Information, physics, quantum: the search for links. In Complexity, entropy and the physics of information (ed. W. Zurek). Reading, MA: Addison-Wesley. Whitehead, A. N. 1929 Process and reality. New York: Macmillan. Yagi, T., Kamimura, S. k Kamiya, R. 1994 Nanometer scale vibration in mutant axonemes of Chlamydomonas. Cell Motility Cytoskeleton 29, 177-185. Yu, W. k Baas, P. W. 1994 Changes in microtubule number and length during axon differenti? ation. J. Neurosci. 14, 2818-2829. Discussion P.

Marcer Machine the Cybernetic Group, Keynsham, (BCS UK). I welcome Penrose-Hameroff as microtubules candidate natural hypothesis concerning systems both classical and quantum modes. This is a testable possessing computational which should not be dismissed out of hand at this very early stage of hypothesis of the nature of understanding quantum computational systems. S. Hameroff. I agree. Let me reiterate Naturally, my belief that, even if Orch OR turns out to be incorrect, it is the type of multidisciplinary approach spanning and philosophy that will be required to understand consciousness. physics, biology efforts toward quantum Technological computation may be well served by studying the fine details of brain microtubules.

Phil Trans. R. Soc. Lond. A (1998)