Jonathan CW Edwards, University College London
The following paper is published in Journal of Consciousness Studies, Volume 12, No.4-5, pp60-76
We perceive colour, shape, sound and touch ‘bound’ in a single experience. The following arguments about this binding phenomenon are raised:
- The individual signals passing from neurone to neurone are not bound together, whether as elements of information or physically.
- Within a single cell, binding in terms of bringing together of information is potentially feasible. A physical substrate may also be available.
- It is therefore proposed that a bound conscious experience is a property of an individual cell, not a group of cells. Since it is unlikely that one specific neurone is conscious, it is suggested that every neurone has a version of our consciousness, or at least some form of sentience.
- However absurd this may seem it is consistent with the available evidence; arguably the only explanation that is. It probably does not alter the way we should expect to experience the world, but may help to explain the ways we seem to differ from digital computers and some of the paradoxes seen in mental illness. It predicts non-digital features of intracellular computation, for which there is already evidence, and which should be open to further experimental exploration.
The binding problem
The binding problem may be defined as the absence of an explanation for the ordered integration of many and varied sensory elements into a single subjective experience. Recent accounts of the importance of this problem to theories of consciousness are given by Chalmers (1995) and by Seager (1995). The problem is a complex of related problems (Hardcastle, 1994; Revonsuo and Tarkko, 2002), two of which I will distinguish, although they are entangled. The first, which I will call the information problem, is that of the nature of the pathways that bring signals arising at different sites in the brain together as information. (Drawing the flow chart without worrying about the physics.) The second, which I will call the physical substrate problem, is that of finding a substrate at the fundamental physical level which might support a subjective experience in which many elements are bound into a seamless whole.
There are further complexities. Binding of experiential elements is full of paradoxes and illusions, most clearly seen in brain disease, but also by normal perception. Experience is like a questionnaire about the world, with answers about both objects and their relations partly filled in. My aim is not to explain these anomalies, but to find a starting point for binding being possible. However, the anomalies suggest that experience uses a language quite different from that used by a television to form an image, in keeping with the sort of mechanism I shall propose.
The argument I will develop is that both information and physical substrate problems point to one solution; that consciousness is a property of a cell, not a group of cells. No one special cell is implicated. It is proposed that all neurones are conscious, or sentient, to a degree; that the single subjective ‘soul’ is a confabulation.
Starting from William James
Late in the genesis of this viewpoint I received a timely prompt from Paul Marshall to read Chapter VI of William James’s Principles of Psychology (1890). Not only did I find James discussing cellular consciousness, or polyzoism, but considering it the only explanation which is not self-contradictory. He cannot accept it but his further comments are amusing: ‘… metaphysics, not psychology, will be responsible for its career (having earlier defines metaphysics as ‘… nothing but an unusually obstinate effort to think clearly.’). That the career may be a successful one must be admitted … a theory which Leibniz, Herbart and Lotze have taken under their protection must have some sort of destiny.’
James abandons polyzoism for two reasons. Firstly, he assumes that it implies the existence of a unique ‘pontifical cell’,which he rejects. As indicated above, my view of polyzoism is democratic, not pontificial. The absence of a sense of ‘multiple subjectivity’ is implicit in James’s exposition; ‘Every brain cell has its own individual consciousness, which no other cell knows anything about…’ so one wonders why James did not entertain the democratic version.
His second objection is more serious, arising from his analysis of the binding problem in terms of nineteenth century atomistic physics. A composite structure such as a brain is merely a construct of an outside observer, and ‘non-existent as a genuinely physical fact’. The brain is a collection of little things, not one thing with the intrinsic identity that would allow it to have a single subjectivity. And ‘The cell is no more a unit, materially considered, than the total brain is a unit.’ At the time, these arguments were irresistible. However, physics has changed since 1890 and I suggest James might have argued differently today (see also Seager, 1995).
With James as foundation the exposition below develops three strands; (i) retracing of the arguments addressed by James, but in a computer age (ii) reclamation (hopefully) of the possibility of binding in the context of modern physics, but only within a cell, and (iii) exploration of the implications of cellular consciousness to address further obstacles that it may raise.
The information problem
In trying to identify a basis for consciousness, a definition of what is being looked for is needed. Individual sensory elements or ‘qualia’ (Chalmers, 1995) seem beyond analysis, but the binding problem provides a handle. Thus, the key functional requirement of consciousness, as I see it, is that something has simultaneous (cotemporal) access to many elements (of information) in defined inter-relationships (SAMEDI), i.e. access to a pattern. This requirement is not sufficient to define consciousness but perhaps covers sentience. Consciousness I will call sentience in which the accessible pattern includes a useful map of some other ‘outer’ environment, normally the outside of a human being, with a sense of time and, in its fullest form, adult consciousness, a sense of self (see Bolender, 2001; Frith and Frith, 1999)).
The information problem is how to find a neurophysiological unit in the brain that could have access to many elements of information as a pattern; perhaps 1000 elements in a single experience. Some would say much more information is needed but I am assuming the information coming in is not encoding every ‘pixel’ of the experience, but is built in an economical, modular way (Searle, 2000)
The answer seems simple. A single neurone has SAMEDI for signals arriving from elsewhere in the brain at its synapses. This is not meaningfully true of a net of cells, each of which receives a separate set of signals. Although it might be argued that the net ‘as a whole’ has access to the information as a pattern this is not the case, as James saw clearly. Simultaneity of inputs to a logic gate such as a cell, or semiconducting unit in a computer, is essential to the processing function of the gate (the signals need to be in operation at the same time) but the timing of events at other gates is irrelevant. Aggregate SAMEDI for several gates has no functional significance.
In an artificial ‘neural net’ in a computer that can ‘recognise patterns’ the pattern does not exist as a pattern in the net. There is no representation of the pattern in the sense of re-presentation. Nothing has SAMEDI for more than 0 or 1 and 0 or 1. Yet something in our brains does appear to have SAMEDI; we experience patterns and these patterns seem to be relevant to the way information is processed.
There is no mechanism for access to information held in several cells other than through signals converging on a single cell. There is no more reason why information should be shared between two cells a hundred microns apart in a single brain than between two cells in two brains a metre apart. The existence of a connection that might transfer information is irrelevant. Such transfer of information would change the pattern, if it indeed existed. Functionally, neurones are as separate as people, capable of exchanging information but not pooling it.
The idea that patterns are accessible as patterns in computers may have become accepted because computers can imitate us. However, as Koch and Segev (2000) point out, there may be a false premise here. It may be assumed that, because intercellular signals in the brain are discrete pulses, as in a computer, information processing in the brain involves only a mathematics similar to that of a computer. This is unlikely (Koch and Segev, 2000). Integration of signals inside cells is complex and can mimic multiplication. It may involve patterns. Thus although it might be argued that binding in a cell poses the same problem as in a net of cells, it does not have to. What we know of the brain indicates that binding could not occur between cells linked by discrete signals but might occur in a cell. A sophisticated substrate might be needed in the cellular ‘black box’ but if patterns are involved at all, this would need to be the place. Put another way, postulating a new, unobserved mode of integration between neurones begs the question why we need axons. A non-bitwise mode of integration within cells is probably already supported by experimental evidence.
In passing, I would make a comment made by many; that synchronised firing (Crick, 1994) of neurones cannot in itself create binding. Synchronisation of the traffic signals in ten cities does not mean that anyone observes it. Reading James I suspect that before computers the arguments must have seemed much more obvious. Consciousness means binding, which means integration, which happens in each cell separately.
The physical substrate problem: as seen by William James
It could be argued that binding occurs outside known physics, perhaps in another set of dimensions. However, neuroscience continues to extend evidence that experience depends on physical neural events at specific sites (Rees et al., 2002). Biology is full of peculiar things getting an ordinary physical explanation. It has also been suggested, in the functionalist view, that physical substrate is unimportant, that what matters is the flow chart (Chalmers, 1995). However, as Seager (1995) has pointed out, a brain has many different functions at different structural levels and it is not clear why one or other should be endowed with consciousness. Moreover, my argument is that the functions usually discussed will not fit. Ironically, my final conclusion is that I agree with Chalmers and, remarkably, with some of Dennett (1996) in that consciousness is a basic correlate of function, but a function that only certain fundamental physical substrates can subserve. To know the true function is to know the substrate.
The physical substrate problem is really the same problem as the information problem, but in sharp focus. As James (1890) points out, a cell appears to be just as much a collection of things as a brain. Access to many elements by a cell is no good if no one ‘atom’ of the cell has that access, and specifically simultaneous, or cotemporal, access.
I should point out here that pattern based information processing can be explained in the conventional physics of 1890. For a cell to respond to a pattern, no one part of the cell need be seen as having simultaneous access to the pattern. If incoming signals create electrical waves which interact in a way dependent on their pattern of phase relations, which in crude terms we know they do, cell output can be pattern-dependent. However, in a classical analysis no part of the cell has access to the pattern in the sense of simultaneous availability of all the incoming signals in defined interrelationships.
For experiential binding, however, SAMEDI appears to be an absolute requirement. Availability of either single bits of information in rapid succession or the single value outcome of an interaction between waves is no good. We cannot escape by saying our impression of simultaneous access to colour and shape is an illusion. The mechanism needed to create the illusion is just as difficult to find, even if it requires fewer elements than we tend to think we observe. The importance of simultaneous access to information may be unfamiliar but it has a long pedigree. It is implicit in Leibniz’s view of the universe (Woolhouse and Franks, 1998).
In classical atomistic physics ‘access to information’ can only really mean exposure to forces. Gravitation is irrelevant in the brain, so the forces will be mechanical or electrical. At the classical atomic level these are the same thing, so that an atom can barely have access to more than one bit of information at a time. A composite structure, such as a bit of brain, with different parts exposed to each of 1000 different forces will not support subjective binding because no one atom has simultaneous access to all these forces as a pattern. For subjective binding we need at least one indivisible part of the brain to have simultaneous access to the 1000 element pattern of forces, which classical physics does not allow. Hence James’s analysis that this is ‘a total brain activity which is non-existent as a genuinely physical fact.’ There are two facets of meaning to this statement. One is that atomistic physics does not allow forces to act as complex patterns. The second is that composite structures do not have the intrinsic identity that might allow them an individual subjectivity.
Fortunately, classical atomism would no longer be the place to look for something as fundamental as subjectivity. Classical physics is now merely a guide to what happens when things act as aggregates. As discussed below, modern physics restores the possibility of subjective binding, because indivisibles have access to many elements of information at once (Feynman, 1985; Bohm and Hiley, 1995; Seager, 1995). However, development of a coherent description of an observer in modern physics has been slow and painful and might yet benefit from a little help from Leibniz.
The physical substrate problem in the modern physics context
Many suggestions have been made about how quantum theory might explain consciousness (Hameroff, 1994; Penrose, 1994; Seager, 1995; Ho, 1996; Jibu et al., 1997; Globus, 1998; Esfield, 1999; Vitiello, 2001; McFadden, 2002) although not always in the context of binding. While reasons for involving quantum theory are often cogent, esoteric aspects are often invoked, such as indeterminacy, wave function collapse and Bose-Einstein condensation. Some of these are not even recognised in all versions of quantum theory (Bohm and Hiley, 1995). Much of the resistance to these explanation relates to the implication of events which tend only to occur under very limited conditions (Grush and Churchland 1995). More general aspects of modern physics would be more attractive as a basis for a biological phenomenon.
Like many biologists interested in subjectivity I must declare my modern physics to be largely self-taught, with advice from physicist colleagues. I have to take my courage in my hands and build a set of ideas that I believe both theoretically acceptable and accessible. An immediate problem is that conventional quantum theory is reticent on what might have ‘intrinsic existence’ of the sort that Leibniz, Descartes and James saw as important. Bohm and Hiley (1995), and Vitiello (2001) I have found the most help in this area, although I see that no perfect ontological view has yet been reached. For those who prefer a strict ‘Copenhagen’ approach to quantum theory I would simply say that a theory that needs an observer but cannot describe it or allow it to exist is no use to students of the mind.
Rather than seeing modern physics as ‘quantum’ superseding ‘classical’ my impression is that there have been a series of conceptual shifts over about 200 years, some of which may be rediscoveries, which I understand roughly as follows:
1. That waves, (oscillating fields) and matter are inextricably linked (Maxwell).
2. That waves need not be just patterns of movement of things, but may in themselves be things, or groups of things called quanta (Einstein’s photons).
3. That all elemental things are waves (electron diffraction etc.; quantum theory)
4. That all true waves are things or groups of things inasmuch as anything can be called a thing (modern field theory).
Put simply, the universe is populated by waves not billiard balls. These waves come with associated positions in space but these positions are more like the cursor on a line you draw with a computer graphics package than a site of a ‘lump’. Moreover, in an ontological analysis the wave and position must play quite different roles, an issue beyond the scope of this paper. A wave is a perturbation with amplitude and phase. It is not necessarily an ‘up and down’ oscillation. It may be more like a spinning or corkscrewing and not necessarily in ‘real’ space. It is also indivisible; you cannot have half a wave. (Try skipping over half a skipping rope.) If the indivisibles (for Leibniz, monads) forming the basis of both observables and observers are waveforms, then James’s arguments need revising. The chime of a bell is an indivisible just as much as an electron. James’s argument ‘[when] H2 and O combine… the ‘water’ is just the old atoms in the new position H-O-H’ is wrong. The waveforms occupied by electrons in a water molecule are new indivisibles.
Indivisibles that can be both observables and observers can reasonably be said to exist in the sense required by Descartes. In Bohm and Hiley’s (1995) terms these are be-ables. This makes it reasonable to say that a bell exists just as much as an electron, and that what defines this existence is a wave, field or quantised mode of perturbation, rather than old-fashioned signs like hardness, weight or opacity. The reader might question whether a chime is made of quanta, as is a light wave. The answer from quantum field theory appears to be yes; the quanta are phonons. There is a technical issue about whether the field or the individual quanta are the be-ables. It seems it is the whole field, which helps to move away from individual quanta which are unlikely to be of any biological relevance.
This discussion leads to the second reason for placing consciousness in a cell rather than a brain. For something to be associated with a quantised field or wave mode, that might give it the intrinsic identity a subject would seem to require, it needs structural homogeneity and clear boundaries. A bell supports such a field but a spoonful of porridge does not. The brain is forever porridge. It is structurally highly heterogeneous and has no well-defined boundaries. No perturbations with amplitude and phase can be expected to occupy totally and exclusively the brain, or any component neural network. Electrical waves in cells are exchanged for diffusional movements of chemicals and back again; it is all too divisible. On the other hand there is a chance that a wave might occupy totally and exclusively a neuronal membrane. It may be dangerous to jump to that conclusion and I shall explore it in more detail later, but it may be the only chance we have.
Access to information by indivisibles
Having implied that a neuronal membrane might behave as indivisible, or monadic, I need to return to the issue of SAMEDI. As indicated by Feynman (1985), the basic difference between modern and classical physics lies in Young’s double slit experiment. A single photon passing through two slits in a screen shows self-interference, indicating that it behaves as if it has simultaneous access to information about both slits. It is not that one part of a wave has access to one place and another part to another. However much a wave is pared down in intensity, even to a single quantum, it maintains its complete, indivisible, spatially distributed relationship to its environment. A photon passing through a ‘micro-colander’ with 1000 holes has simultaneous access to 1000 elements of information as a pattern: it fulfils the functional requirement of sentience. Moreover, this access to information informs the progress of the photon. It is the ‘active information’ that determines the behaviour of waves for Bohm and Hiley (1995). Put another way, if we are made up of waves we have no justification in thinking that our sentience is any different from the access to ‘active information’ of a wave.
We now seem to have a framework for indivisibles that can be both observables and observers which have this peculiar binding property that we find in consciousness. Both aspects of James’s concern about ‘genuinely physical facts’ are addressed. Note that the rules of access to information by indivisibles are not those which govern access to information about indivisibles, i.e. Heisenberg’s Uncertainty Principle.
Which indivisible perturbations might carry our consciousness?
A number of suggestions have been made for perturbations that might mediate consciousness. Several people have suggested that the brain’s electromagnetic field might help to explain binding, including Crick (1994), McFadden (2002) and Pockett (2002). However, there are several problems with ‘electromagnetic thought’ some of which Pockett herself has well described (Pockett, 2002). A brain’s electromagnetic field is not obviously separable from that of the spinal cord, nerves, skin, or even clothes. There is no evidence that the electromagnetic field generated by neuronal activity is involved in processing information in a way that could influence behaviour. The radiofrequency photons that form the field would be scattered and absorbed at random or pass rapidly out of the body. Much of the information they have access to would be about irrelevant brain ‘plumbing’ like blood vessels. We need perturbations that stay in a demarcated, ordered but potentially complex domain and are tuned to data mapping the outside world.
Hameroff (1994) has suggested that microtubules might be a substrate for consciousness. However, it is in the cell membrane that electrical waves generated at synapses, tuned to information about the outer world, are integrated. A translation of information to microtubules might seem excessive. Membranes can support more complex wave patterns than filaments, being essentially two, rather than one, dimensional. Anchorage to the cytoskeleton might, nevertheless, modulate membrane waves either mechanically or through ionic fluxes; the two structures might function as a unit, like ‘cello string and sound box.
In a sense we know that the integration of information in neurones involves the interaction of electrical waves spreading out from synapses, according to the rules described by Hodgkin and Huxley. We know that those waves operate within the demarcated, ordered semi-crystalline domain of the cell membrane. The simplest solution would be if these are the waves associated with sentience. But do they have the right properties?
The hypothesis seems to make two requirements of a wave that might endow the neuronal membrane as a whole with sentience linked to behaviour. Firstly, a wave with access to information about the state at all synapses would need its wavefront (or domain of non-trivial amplitude) to occupy the whole neuronal dendritic tree. This would seem to require a reverberating wave with time to make several passes – like the resonation of the bell. In Vitiello’s (2001) terms it would be a long range correlation. Secondly, to be describable as a quantised field it probably needs to be energy conserving, at least to a first approximation.
It is not clear that a purely electrical wave with these features exists. The basic Hodgkin-Huxley wave is a simple damped, dissipative biphasic packet. Fröhlich (1968) suggested an electromechanically coupled wave in which electrical and elastic potential were exchanged. He suggested that this wave might be a Bose-Einstein condensate, but this seems unlikely and is as far as I can see unnecessary. There is no doubt, however, that electromechanical coupling can occur in neuronal membranes. As shown by Iwasa et al. (1980), an action potential is associated with a mechanical wave. Petrov (1999) has shown that as polar planar liquid crystals, cell membranes generate biologically relevant voltages when flexed and vice versa. This is a form of piezoelectricity, called flexoelectricity, involving coupling of phonons to an electrical field. At least in isolated sheets of membrane modes of electromechanical perturbation can be established. Of note, there is increasing interest in electromechanical coupling as fundamental to the way cells sense their environment and respond through opening and closing ion channels (Zhang et al., 2001; Kumanovics et al., 2002). It has been suggested that this might be particularly relevant to sites of membrane curvature such as neuronal dendritic spines (Zhang et al., 2001).
Many attempts to relate consciousness to modern physics have sought to identify new mechanisms for information processing in the brain ((Hameroff, 1994; Penrose, 1994; Jibu et al., 1997; Globus, 1998; Vitiello, 2001), giving sentience a place in the causal chain. I was initially drawn by this line of thought despite being advised by people such as Andrew Huxley and Horace Barlow that new mechanisms look dangerously redundant. I then realised that although quantised fields may help to explain bound experience, that this in no way implies that a ‘field of sentience’ should contribute to output in an active way. In fact there are several arguments against, which I can only cover briefly here.
Assuming that the field of sentience is some sort of long range electromechanical (perhaps piezoelectric) correlation, then an active effect on information processing should be through local modulation of the parameters of interacting Hodgkin-Huxley waves; either membrane dimensions or capacitance or conductance terms. However, there is good evidence that these are constant. Moreover, active involvement in the electrophysiological process would require work and that would be at odds with an energy conserving field. Although primitive organisms and specialised cells like cochlear hair cells translate electrical effects into movement, neurones appear to be adapted to being an immobile substratum for electrical interactions. Neural spine design may actually minimise membrane movement.
The resolution of the problem seems to be that a field of sentience can just as much be bound into the causal chain by being inseparably associated with constant parameters for each cell. In this model the field of sentience can be compared to a chess board which is a passive but essential component of the causality of the game. The causal power of a chessboard may seem limited but the causal power of ten billion chessboards each with ten thousand squares, connected by trillions of pathways subject to feedback-related plasticity would not be.
It remains uncertain as to precisely what long range correlations exist in the cell membrane as a whole, or at least throughout the dendritic tree. Nevertheless, the key point remains that whereas resonant waves in a large chunk of brain are implausible, they are reasonably plausible in a cell membrane.
The above account may appear to leave something missing. It has been argued that experience itself must affect output if we are to talk about the nature of experience and its role. But I would argue that we only ever talk about the content of experience, which is carried by conventional interacting Hodgkin-Huxley waveforms, as it appears to a sentient entity similar to those from which it arose. My feeling is that the above does provide an adequate account of causality and that any apparent inadequacy arises from the dynamics being counterintuitive. Specifically, I would argue that if a paradox remains it cannot be resolved by suggesting another active mechanism in addition to the electrophysiology we already have evidence for. The proposal is, therefore, that there is a solution to the binding problem which requires nothing esoteric and nothing for which we do not have some experimental support, as long as it operates in each cell separately. If a cell can support modes of electrical/mechanical oscillation in its membrane, as a liquid crystal should, we have no reason not to expect it to have a sentience (SAMEDI) which mirrors the cell’s input. If the input maps the outer environment, the cell is conscious, and if the input maps the self, it has adult consciousness. Sentience remains awesome, but not at odds with physics.
Encoding the world into a conscious map
Could a neurone support enough information to explain the richness of visual experience? During the process of peer review I discovered that Steven Sevush has come to much the same conclusion as myself from a different standpoint, specifically addressing the neuroanatomical feasibility of cellular consciousness (Sevush, 2004). Each cell has thousands of synapses which can receive tens of messages a second. We think we are aware of hundreds of things at a time, but it may be much fewer (Noë, 2002). I suspect that what we think of as the full detail of an image is not downloaded into a cell, but that modules of experience embedded in the fine structure of the neurone are ‘called up’ by economical incoming codes. Moreover, our experience does not seem to be formed from ‘pixels’, but in a language of preformed elements (Searle, 2002). Binding of qualia based on electrical wave patterns might be more like transforming topologies than digital addition. We assume a cell could not support a whole picture because we are used to images built up with a paint brush or a VDU but that is not how waves work.
A worrying feature of Bertrand Russell’s point that all we are ever aware of is the inside of our head, is that it is hard to see why some part of the inside of a head should see in itself a landscape, an interior, or any other view of the world. If consciousness is in one cell the question becomes why a pattern of electrical phenomena in a membrane should give rise to sensations like yellow and lemon-flavoured, meaningfully interrelated. The initial answer would seem to be that we have no reason to think we can expect to predict how a cell membrane should experience itself or that it should not be what we do experience. At least it is a unitary physical substrate rather than the abstract ‘functional’ substrate of apparently indistinguishable signals passing along separate paths.
A number of people have looked for relationships or equivalences between subjective space and time and ‘outside’ space and time, raising points worthy of much further debate (Marshall, 2001; Bolender, 2001; Romijn, 2002). However, to include emotion, colour and taste, we have to assume considerable non-equivalence and it may be that outside space is encoded in both space and time and outside time similarly. Specifically, I would suspect that dimensions may be encoded in differentials in time (rates of change) since that is what waves are about. This might bear on the concept of the ‘specious present’, in which physical events dissociated in time are perceived as co-temporal but offset from an objective present. There may be no reason to think that ‘a moment’s thought’ lasts ‘a moment’.
The issues of time and differentials in time suggest a reason for the association of synchronised electrical signals with consciousness (Crick, 1994). While implication of synchrony in an intercellular model of binding seems untenable, synchrony should have a vital role in an intracellular binding based on patterns of wave interaction.
It may be possible to deduce the language for encoding the outside world into an internal map by considering constraints such as the need to integrate input from different senses in congruent dimensions. However, unlike normal languages, with meanings shared between individuals, a wave-based language of sentience would only be used by a cell to experience itself. Communication with other cells would be via discrete signals, not wave patterns. If two adjacent cells used a pattern in one case to be yellow and the other to be lemon-flavoured, it should not matter. All that is required is that the cell experiences a sensory input of yellow in the same code as a signal indicating the memory of yellow. Much more could be said on this subject but it may be that the rules governing the language of experience are more flexible, and usage less constant, than might be expected.
What is known about intracellular signal processing?
As reviewed by Koch and Segev (2001), integration of incoming signals in neurones has often been considered as a summative electrical process. However, Koch and colleagues have shown that this is too simple and that integration fits best a complex partly digital, partly analogue model including something equivalent to multiplication. This makes true pattern-based processing a serious possibility. Koch also points out the diversity of dendritic morphology in different types of neurone, suggesting that integration may involve combinations of processes with different emphasis on one or other process in different cell types. Apical, oblique and basal dendrites may have different roles in a single cell.
Summative integration is of no use as a basis for conscious experience. A thousand signals integrating in a summative fashion can generate two meanings; enough or not enough. With pattern-based binding the number of meanings could be 2 to the 1000th or more (about 1 followed by 300 zeros), enough for a separate meaning for every millisecond of life of every cell in every brain that ever lived. It would be a mistake, however, to assume that pattern-based integration is always advantageous. If it is possible at all, setting up waves with useful effects may require both very sophisticated cellular microstructure and very sophisticated regulation of information input. A digital computer can solve almost any problem with enough speed and memory. An individual pattern-based unit may only be able to solve one problem, or very few. Any one cell, may experience a panorama but only occasionally contribute to behaviour by identifying a local pattern match to which it is tuned.
A lot more information is needed about intracellular computation and its potential link to sentience. However, the hypothesis being explored makes a testable prediction; that at least in some neurones, complex patterns are involved.
A copy of consciousness in each cell: why not?
The picture I have painted is of two components to brain processing, acting in series rather than parallel. The intercellular part of processing is synaptic, non-bound and like a computer. The intracellular part has a synaptic input, is at least partly pattern-based, is available to a sentient field with SAMEDI that is a be-able in Bohm and Hiley’s terms, and leads to an electrical output. From both informational and physical substrate points of view this seems to be the only model that allows binding. The main difficulty may be cultural. We think we have a single sentience. I am proposing we are colonies of sentient cells, each with a hermetic unshared consciousness. We believe as a debating house believes, worship as a congregation. Is such a view consistent with experience?
Although the literature is dominated by the idea of a single centre of consciousness, the alternative idea of more than one centre of consciousness is not new. As indicated, William James considered a pontificial polyzoism very seriously and implied that the idea had a long tradition. However, in a democratic polyzoism there is no single ‘me’ cell; in at least some parts of the brain there may be millions that have the experience of being me. This has the advantage that it is consistent both with James’s fundamental requirements for binding and with the apparently distributed nature of consciousness. It is also very compatible with the ‘holographic’ aspects of consciousness described by Pribram (1991) in the sense that a version of the ‘experiential story’ is distributed widely throughout the brain.
If one neurone benefits from sentient pattern-based processing why not all? I suspect that 99% of what goes on in my brain is decided on somewhere other than the site of any subjective viewpoint that I see it from. This 99% is termed unconscious, but a lot of the clever things that brains do emerge from this unconsciousness. It would be easier if, rather than being unconscious, these events were in other cellular consciousnesses, or at least sentiences. This would resolve a problem with Penrose’s (1994) masterly argument that conscious thought is non-computable; that the best thoughts often seem to surface from somewhere else.
Although held in a cultural framework, the belief that we have a single self probably has an inbuilt biological basis. As indicated by Frith and Frith (1999) and by Gazzaniga (1998), there is evidence that one or more regions in the left frontal region are responsible for a story of self which makes sense of our relationship to the world and other selves. In individuals with autism this story may have different rules or may not be told at all. The story telling may be almost unique to Homo sapiens and may be the source of our success. It may be a useful myth.
How the story of self works is unclear. Certain cells in my left prefrontal area may be those that think they are me. It might also be that they regulate the flow of, perhaps synchronised, impulses which determine the focus of attention for other cells. The profuse branching of many axons would allow a pattern of signals to be distributed widely, such that large numbers of cells could receive a copy of the same story of self which they interpret as their own. Democratic polyzoism does not alter the need to answer questions about how intercellular pathways build up the story. It just suggests that these questions may have a different relationship to subjective experience in terms of site and copy number.
We are brought up to think that our brains are conscious as a whole but we are also taught that the cell is the living unit and that chemical processes are regulated individually in each cell. The Darwinian reasons for retaining separate biological machinery in each cell may apply equally well to consciousness. The existence of consciousness in each neurone is very much in keeping with what we know about recovery after brain damage. Patchy loss of grey matter or tracts of white matter would leave many cells to relay the contents of their consciousness to the outside world, with defects in certain areas relating to the lack of input from sister cells. A copy of consciousness in each cell might be the safest way to protect against threats such as birth trauma or measles encephalitis.
However wasteful it may seem, every cell has a copy of the genetic blueprint of our bodies. Why does each cell not hold a consciousness, or at least sentience, in its membrane? Each cell’s version would differ, more like the proteome than the genome. Red blood cells have no genes and are unlikely to make use of membrane waves. Skin and bone cells might have little use for sentience, but in a wound they undertake repair with considerable skill. They would not have access to the sort of mapping system generated in the brain but nevertheless, integration of selective information about the cell’s environment in the cell membrane might be used for regulating changes in cell shape and sites of secretion of tissue fibres. Sentience for them and for phagocytes might have no extended temporal reference. On the other hand some such cells may show behaviour suggesting memory or learning (Frost, 1996). Since each contains enough genetic information to build a brain we should not exclude the possibility that they have a ‘programmed self’. Neurones have a much more stable cytoskeletal structure and their form of consciousness may be incompatible with locomotion. Nevertheless, looking for evidence of integration by patterns and waves in single cells might be easier than in neurones. The effects of anaesthetics on single cells might be worth much more study.
Conversely, complex pattern-based integration might not be a major feature of all neurones. Neurones in peripheral nerves may be little more than linear electrical relays, less ‘intelligent’ than phagocytes. Neurones in the brain would generally be expected to use sophisticated integration and mapping, perhaps especially to generate abstract concepts, language and introspection but some cells may only deal with information from the retina or ear. The thoughts that we exchange in conversation may go on in rather few cells, but it is still possible that very many neurones are aware of a similar breadth of information but in a ‘dialect’ suited to their field of action. Perhaps the various shapes of dendritic trees hold clues.
Although consciousness is often related to the cerebral cortex, it is not absolutely clear that this is where a sense of self would reside. The thalamus is also crucial to consciousness and is densely supplied from overlying cortex. It is not inconceivable that this is where the ‘me’ cells might be. Parts of the brain stem, notably the peri-aqueductal grey matter, are crucially important to consciousness in the sense of being awake and responsive (Panksepp, 2002). Specialised cells in this area might be the ones to hold the consciousness that people write articles about. However, many would argue that these areas simply control wakefulness and sleep by sending out signals which regulate synchronous discharges elsewhere.
Why, if consciousness is in a single cell, do humans have such a large brain? Brain size is frequently related to intelligence, sophisticated behaviour, and everything we associate with the ‘higher consciousness’ of man. This, I suspect, is one of the most misleading premises of consciousness studies, as shown by the comparison of a bee and a cow. A bee has about a million times fewer neurones than the cow. The bee can talk to its hive mates in quantitative geometry and fashion hexagonal arrays. Cows seem to do little except eat grass. Brain size has little to do with sophistication of behaviour or social activities. An orang-utan may paddle a canoe and have a place in a group hierarchy. A crow with no cortex and a nut-sized brain can use a tool to get food from a jar and has a pecking order in much the same way. Brain complexity may give more scope, but it should not be seen as central to the consciousness issue. What may well be true of insects is that they do not have room for the networks needed for self-consciousness. Complex cellular consciousnesses may exist in greater isolation. The neurones controlling chomping mouthparts may not recognise as ‘me’ the leg being simultaneously chomped by another animal (Penrose, 1994).
One might ask why one’s experience belongs to ‘this’ cell, or why we wake up every day as the same cell. These are valueless questions, no more puzzling than why we are not turnips, or on Mars. The cell(s) you are at this moment refers to a congruent library of incoming memories. We would never be in a situation of saying, why am I a different cell from the one I was yesterday, or how come my body goes on talking when I have died of a stroke?
A consciousness in each cell might be relevant to aberrations in experience and mental illness. We talk of being in two minds, suggesting that although conflict within a consciousness may not be possible, conflict between conscious entities may be just below the surface. What is going on in those people who smile on one side of their face? Do people claiming to have more than one personality have segregated subgroups of self-narrating cells? Are hallucinations and other features of schizophrenia due to failure of regulation of the input of the self story into self-conscious cells? The equivalent questions are hard to answer with a global self.
In these questions there is a considerable danger in confusing the content of consciousness, input from other cells, and the way it is processed in a single cell. If consciousness in each cell should alter our interpretation of our experience I suspect it is in a subtle way which needs to be analysed with care. Nevertheless, the implications of looking at ourselves as colonies rather than individuals are not trivial. Even in terms of traditional neuroscience it cannot be controversial to say that decisions are not made by men and women but by cells. However, the idea that those cells might be the only conscious actors in the play may highlight just how relevant this is to real life. Politics and spin go on inside as much as outside as the Prince of Denmark and the Moor of Venice were well aware.
It must, however, be stressed that anarchy within a democratic polyzoism is not to be expected. Conflicts may chiefly arise when the self story is mismatched to the social environment. All the cells in the brain survive together, so there will be no pressure to compete. As in a general election, the retelling of the self story will be based on a consensus from the neural polls. If free will is a meaningful concept, which I personally doubt, every cell will make a very small contribution. Moreover, cells will have no knowledge of what their own contribution is. The cell’s output, the action potential, probably simply wipes the membrane clean of incoming patterns. All is designed for co-operation, as for the asexual workers in a beehive. Put another way, democratic polyzoism is functionally no more than those same co-operative interactions of individual cells the neurophysiologists have already identified – functionally it is the way we know it is.
Perhaps the most disorientating implication is that there may be no single subjective soul that controls our behaviour. We think of everything we do as being controlled by one being, and in general we see this being as something that communicates through language, including the writing of papers on consciousness. But the question of which cells ultimately control our conscious behaviour (Ramachandran and Hirstein, 1997; Frith and Frith, 1999; Crick and Koch, 2003) is likely to be too simple. Every action involves thousands of cells. Some communicate through talking or writing because they have verbal inputs. However, our perception of consciousness in others need have nothing to do with words. We sense that animals are conscious without them speaking. Eye contact tells us of consciousness, leaving the verbal mind straggling behind. Different members of our colony interact in different ways with animate and inanimate entities. Such interactions may involve consciousnesses which may be full of emotion but dissociated from speech. When we hear Casals we recognise across time the power of cells that speak without words. When we look at Bonnard we know that true intimacy is silent. Perhaps consciousness studies are the fruit of nerd cells that need to go out and get a life rather than just a second hand story from the real players!
If indeed every cell is independently conscious, the implications may be far reaching. Nothing I have so far encountered seems to make the concept incompatible with experience. The main problem seems to be our shared preconception of the existence of a single sentient self. The idea is potentially testable, since it predicts that intracellular information processing is based on complex patterns, at least in some cells. It also requires the existence of long range correlations in membranes, that might be amenable to probing via sensitivity to general anaesthetics. Until such time as it is tested experimentally it can only be judged on Ockham’s rule of parsimony. It is probably as parsimonious as any other suggestion. No new physics and little, if any, new biology are required.
Bohm, D. and Hiley, B. (1995) The Unidivided Universe. (London, Routledge).
Bolender J. (2001) An argument for idealism. Journal of Consciousness Studies 8, 4, 37-61.
Chalmers, D. (1995) Facing up to the problems of consciousness. Journal of Consciousness Studies 2, 3, 200-219.
Crick, F. (1994) The Astonishing Hypothesis. (New York, Scribners).
Crick, F. and Koch, C. (2003). A framework for consciousness. Nature Neuroscience 6, 2, 119-126.
Esfield, M. (1999) Quantum holism and the philosophy of mind. Journal of Consciousness Studies 6, 1, 23-38.
Feynman, R. P. (1985) Q.E.D. (Princeton, Princeton University Press)
Frith, C. and Frith, U. (1999) Interacting minds : a biological basis. Science 286, 1692-1695.
Fröhlich H. (1968) Long range phase correlations in biological systems. Int. J. Q. Chem. 2, 641-649.
Frost, H.M. (1996) Perspectives: a proposed general model of the ‘mechanostat’ (suggestions from a new skeletal-biologic paradigm). Anat. Rec. 244, 139-47.
Gazzaniga, M. (1998) The mind’s past. (Berkeley, University of California Press)
Globus, G. (1998) Self, cognition, qualia, and world in quantum brain dynamics. Journal of Consciousness Studies 5, 1, 34-52.
Grush, R. and Churchland, P.S. (1995) Gaps in Penrose’s Toiling. Journal of Consciousness Studies 2, 1, 10-29.
Hameroff, S. (1994) Quantum coherence in microtubules. Journal of Consciousness Studies 1, 1, 91-118.
Hardcastle, V.G. (1994) Psychology’s binding problem and possible neurobiological solutions. Journal of Consciousness Studies 1,1, 66-90.
Ho, M-W. (1996) The biology of free will. Journal of Consciousness Studies 3, 3, 231-44.
Iwasa, K., Tasaki, I. and Gibbons, R.C.. (1980) Swelling of nerve fibres associated with action potentials. Science 210, 338-9.
James, W. (1890) The Principles of Psychology. Chapter VI The mind-stuff theory. (Cambridge MA, Harvard University Press).
Jibu, M. and Yasue, K. (1995) Quantum brain dynamics and consciousness. (Amsterdam and Philadelphia, John Benjamins).
Koch, C and Segev, I. (2000) The role of single neurons in information processing. Nature Neuroscience suppl 3, 1171-1177.
Kumánovics, A., Levin, G. and Blount, P. (2002) Family ties of gated pores; evolution of the sensor module. FASEB Journal 16, 1623-29.
Marshall, P. (2001) Transforming the world into experience. Journal of Consciousness Studies 8, 1, 59-76.
McFadden, J. (2002) Synchronous firing and its influence on the brain’s electromagnetic field. Journal of Consciousness Studies 9, 4, 23-50.
Noë, A. (2002) Is the world a grand illusion? Journal of Consciousness Studies 9, 5-6, 1-13.
Panksepp, J. (2002) The primordial self. In Consciousness, Carter, R (Ed.) pp 186-188. (London, Weidenfield and Nicholson).
Penrose, R. (1994) Shadows of the Mind. (Oxford University Press).
Petrov A. (1999) The Lyotropic State of Matter. Gordon and Breach, Amsterdam.
Pockett, S. (2002) Difficulties with the electromagnetic field theory of consciousness. Journal of Consciousness Studies 9, 4, 51-56.
Pribram K.H.. (1991) Brain and Perception. Lawrence Erlbaum, New Jersey.
Ramachandran, V.S. and Hirstein, W. (1997) Three laws of qualia. Journal of Consciousness Studies 4, 5-6, 429-457.
Rees, G., Kreiman, G. and Koch, C. (2002) Neural correlates of consciousness in humans. Nature Reviews: Neuroscience 3, 261-270.
Revonsuo, A. and Tarkko, K. (2002) Binding in dreams. Journal of Consciousness Studies 9, 7, 3-24.
Seager, W. (1995) Consciousness, Information adnd Panpsychism. Journal of Consciousness Studies 2, 3, 272-288.
Searle, J. (2000) Consciousness. Annu Rev Neurosci 23, 557-78.
Sevush, S. (2004) Single-neuron theory of consciousness. Cogprints.org/3891/
Vitiello, G. (2001) My double unveiled. (Amsterdam, John Benjamins)
Woolhouse, R.S. and Franks, R. (1998) G.W. Leibniz, philosophical texts. Oxford, Oxford University Press.
Zhang, P-C., Keleshian, A.M. and Sachs, F. (2001) Voltage-induced membrane movement. Nature 413; 428-31. /HTML