Mind, body and ME

[Eddie]

We use impressions to get an idea of what the question is about but the sense that the proof must be true is not based on any particular impression. It is a sense that, for all those triangles we have never seen, it must be so.

Ok I think I understand what you mean. 'That sense' is some kind of intuition that probably has some evolutionary advantage. It is in a similar sort of sense to the way in which an animal has some intuition to avoid certain predators as you mentioned. So, if Hume would agree that that sort of 'sense' is knowledge then I think it is fair to assume that he was wrong as that impression is not directly had by that individual.

The problem I have with our brains containing patterns for every impression is that there are uncountably many impressions one could have. The experience of viewing an object is different at every moment of time and from every direction. I find it hard to imagine that all that information is already there. Why would it just not be the case that instead of it already being there, there are essentially uncountable many ways in which the brain's matter can be rearranged to store those impressions?

I see "knowing what they mean" in the same sort of way as I do with the brain containing patterns of impressions. It is not that we know what every impression means, but that there are some general rules which our brain uses to processes said information. One such rule might be that our brain treats objects as discrete entities. This isn't because we know what each entity means but because treating them as such has been useful.

The brain then learns from others that these discrete entities have labels that can be attached to convey information to others. I don't see how this is "knowing what those objects mean" it is more our brain has a certain way of processing incoming information. And this way of processing doesn't reflect how the world actually is, it reflects how to act in accordance with survival.

I'm sure I will have plenty more thoughts as I read though the rest of the posts, but I would ask what material you recommend reading on this. Who do you think are the people who most influenced your thinking on this outside of Leibniz? Before I got ME/CFS I wasn't studying philosophy and only recently started exploring these ideas.

 

[Jonathan Edwards]

So this logic would go all the way down then to the smallest possible unit of time in order for an event to be truly indivisible,

Right.

which has something to do with planck I'm guessing?

I don't think so.

Effectively, we are talking about 'quanta' of time. The central tenet of Bohr's quantum theory is that it deals with indivisibles - which at the time were (and continue to be) actions. Planck time was never proposed as an indivisible unit but as the level of precision of a measure of time beyond which you can never get a meaningful answer.

A number of people have talked about 'quantising time' or more often 'quantising spacetime', including Smolin, Rovelli etc.. But quantum theory never called for quantisation of metric. It is about indivisible actions or events. A quantum of time would only have identity if it was a discrete event or action, or causal connection within a field. And so we already have quantisation of spacetime - it is the quantisation of usage of spacetime by events. This is just like energy. Energy has no grain. A quantum of energy is just the energy used or owned by an action, event or in QFT terms, field excitation, such as a photon or s orbital. This is the general approach theat links with work by the philosopher of science Richard Arthur and the physicist Steven Savitt who talk of 'causal diamonds' - made up of an opening and a closing light cone (it may end up a bit more like an ellipsoid but this gets technical). Richard and I tend to agree on most things.

So the indivisible durations we are looking for are durations of field excitations. Photons in brains don't last long enough. S orbital electron modes tend to last for millennia. Valency electron modes tend to last between chemical reactions, which are probably still too slow. But once you get into electrical perturbations in conducting materials (and semiconducting materials) and move into the complexities of condensed matter physics you have a range of field excitations that can easily last milliseconds or seconds. So we can have the right time frames.

The physics is very complicated but has the merit that once you get into ordered semicrystalline biological structures like cytoskeleton and Nambu-Goldstone theorem you come out the other side with things like electrically coupled acoustic modes and the relevant dynamics collapse to old fashioned acoustics and Maxwell electrics as in a crystal set radio. The idea that interests me is that the cytoskeleton of a neuronal dendritic tree is effectively an aerial but one that responds to complex geometric patterns of electrical potential rather than just a temporal waveform.

Sorry to be so long winded but there is a lot to cover. I can walk you through the whole thing but it takes time.

 

[Jonathan Edwards]

So it would have to be something which isn't an action potential

We wouldn't want it to be an action potential, which is something that creeps along a very long fibre like tsunami. Integration of information occurs in dendrites, which respond much more instantaneously to integration over hundreds of potentials. That is where the physics looks reasonably promising.

 

[Jonathan Edwards]

It seems you experience things through a rolling shutter like you say, and relatively slowly, but where each frame represents one thing - a bit of one of an internal/external sound/image/feeling completely on its own.

The simplest explanation I can see is that experiences have the time frame of the relevant clocking rhythm - alpha to gamma - somewhere around 25-100 ms. Patterns of potentials give rise to new excitations/events over that time frame. There has been a long debate about whether experience is continuous or a 'string of pearls'. Introspection is no good as a guide because we know that the representation of time, space and everything in them is done with 'sentence-like' symbol packages. That means that your experience can include information about several time frames at once. That is how you can hear a note in a tune and at the same time hear that it is part of a familiar tune. There are probably useful insights to get from further experimentation with time perception but we have to interpret things very carefully.

 

[Eddie]

At first pass, yes, the knower is just the event of experience, if we have no legitimate cause to propose some 'enduring' knower that hosts these events. However, right at the heart of Leibniz's conception of the indivisible subject-as-perceiver-as-action is a problem with temporal divisibility. I think there is a technical solution to this in condensed matter physics but, yes, there is a mass of context here to cover first.

I have no issue with there being no enduring self. Conscious experience only exists at the current instant and must update with the same speed as that which the brain can process information.

It is my first time hearing about the combination problem. Why does non-locality enter the discussion? Could cells in the brain be in contact with one another at light speeds such that consciousness is delayed by a fraction from when things actually occur? The argument from wiring in series or parallel is intriguing, but in an embrace the wiring doesn't connect the two brains together.

I agree that emergent properties are strange and must exist at some level at the foundational level. I think we all agree that there is some property of a water molecule that makes it feel 'wet'. We just have no idea what that property is and since we can only see it when lots of water molecules get together is is called emergent. So I guess consciousness must to some extent exist on the level of the cell. But then it is interesting that we can only think one thought at once.

 

[Jonathan Edwards]

The problem I have with our brains containing patterns for every impression is that there are uncountably many impressions one could have.

I agree on this bit. It is not uncountably many. We might guess that we might want a system that can provide 10 ^100 impressions: 10,000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000

The number of degrees of freedom in the input to a neuron provides around that. I am not persuaded we need more! We probably want around 20-50,000 degrees of freedom and that is what neurons have.

So I am not suggesting that each of these patterns has been signalled beforehand, just that the options and their meanings are ready and waiting. We do not need to ask for matter to move around, beyond shifts of ions mediating synaptic potentials. The impression can be manifest as an EM field pattern that vanishes in 25 ms.

 

[Jonathan Edwards]

It is not that we know what every impression means, but that there are some general rules which our brain uses to processes said information. One such rule might be that our brain treats objects as discrete entities. This isn't because we know what each entity means but because treating them as such has been useful.

This is where we get into Frege's distinction between sense and reference, meaning as how it seems to mean and meaning defined by what it refers to in the outside world. Brains and computers need to keep to rules that respect reference, yes, but that is a different matter from there being anything that might 'know' or 'have a concept' which I think requires meaning as sense.

 

[Jonathan Edwards]

I'm sure I will have plenty more thoughts as I read though the rest of the posts, but I would ask what material you recommend reading on this. Who do you think are the people who most influenced your thinking on this outside of Leibniz?

That is a tricky question. The simplest answer is that I don't think anyone has really touched Leibniz even now. The problem with Leibniz is that it is often said, probably with good reason, that he is very hard to understand unless you have already wrestled with the same problems and come to much the same answer (as I had - I was told to read him because a paper of mine sounded like what he had said). I have written about 20 papers on this over the last 20 years. Steve Sevush has written a similar account in a paper and a book (The Single Neuronal Theory). Richard Arthur's books on Leibniz make him much more intelligible but Richard does not address conscious experience per se.

Email exchanges like this tend to generate cross purpose a bit and I don't think we are that far apart. I see sensed meaning as something that needs much more recognition than it gets because of its strange proposition ls structure if nothing else. But that is another tangent to go out on!!

 

[Jonathan Edwards]

It is my first time hearing about the combination problem. Why does non-locality enter the discussion?

The combination problem has kept many philosophers of mind in meal tickets for the best part of a century. Non-locality enters the discussion because the sort of non-locality needed busts open any testable physical theory. It is not a matter of faster than light. Quantum interactions override that in a sense, even if no two determinate events in causal sequence ever follow faster than light. It is the integration within a quantum event that ignores c. But the problem relates to locality in a much more fundamental sense - the sense that worried Newton long before anyone knew light had a velocity. It is locality I the sense that our theories depend totally for their predictions on each event having a domain address in spacetime. Any theory that tries to combine events with different address domains into a single event, such as an experience, simply cannot be tested in scientific terms. It involves a contradiction and as you probably know syllogisms with contradictory premises allow you to treat any conclusion you like as valid!

In the embrace the wiring does not connect the cells but as I then say, if you are relying on message sending through connections you have to combine events that are in sequence and that violates everything theory we have ever had in even worse terms.

 

[Jonathan Edwards]

I think we all agree that there is some property of a water molecule that makes it feel 'wet'.

I don't think so. Wetness is mostly feeling inappropriate coldness and low viscosity of many molecules. Water molecules have dispositions that allow groups of them to feel wet but then the 'feeler' is not the water, but 'you'. Arguments about wetness emergence have kept a lot of other philosophers in business but I don't think they are of any use in explaining subjectivity. As far as we know some form of subjectivity is everywhere, and we cannot know, so there is not a lot of point trying to explain why it should only be in certain places. Subjectivity that includes representations of red buses is unlikely to occur in peanuts but for reasons that are easy enough to understand - there are no integrating pathways that would generate and present such a representation in peanuts.

 

[Jonathan Edwards]

Leibniz's philosophy is distilled into his last major work, the Monadology of 1714. It sets out his beliefs, often without any explanations, which are better found in his Discourse on Metaphysic of 1686. I had a go at trying to explain what I thought Monadology was about and how much it still applies, in a long piece at Qeios: https://www.qeios.com/read/8RDKGI

It may still be incomprehensible, but that was my best shot.
For a complete collection of the key works the best is Woolhouse and Francks paperback GW Leibniz: Philosophical Works. I found them very easy to read and engaging. They are all quite short.

 

[Eddie]

Ok, I've been trying to fully grasp what “know what they mean” is referring to, so I wanted to try and give a more concrete example. I like the idea of the shape or geometry of the neuron as encoding the information so I’m going to use that.

The shape of the neuron might be dictated according to the incoming data via some set of rules that puts the information into a more usable form. Because the rules state that such inputs result in such a position, a neuron knows that if it is in 1 of the 10^100 possible positions it has received one specific impression out of all the possible impressions. An basic analogy might be that since I find myself in the position of enjoying pizza I can know that I have all the necessary factors that go into enjoying pizza. These two things don't have a direct correspondence, but if there were you could know exactly what these input factors were from the fact that you enjoyed pizza. And likewise the neuron knows that it has certain inputs because it is in a certain position.

Maybe then there are some other neurons that have been set up to collect the average positions of neurons who fall within specified boundaries. There could be one neuron that collects the average shape of a neuron when looking at a chair because all those neurons spread across time had similar but not identical positions. This sort of storage neuron could remain across time and produce some vague idea of a chair without needing the inputs. You could also have some neurons that have You could use this to produce the experience of knowledge, if a reacting neuron’s position matches enough of the relevant positions of the storage neurons it could be accepted as true.

There are problems with this, like why different impressions of similar objects would result in similar neuron positionings and what boundaries collection neurons use. But I think that these are problems that could have been solved through a long enough evolutionary processes. Even if this is off the mark, hopefully the idea of the neuron knowing what things mean is somewhat right. And even if the neuron knowing isn't to do with its positioning, I think the logic still holds.

I also think I agree with you on emergence. I wasn't trying to say that the subjective feeling of wetness is a property of water, just that there is some property of water (viscosity, its bonding structure, the molecular size) that does get perceived as wet for whatever reason.

 

[Eddie]

Leibniz's philosophy is distilled into his last major work, the Monadology of 1714. It sets out his beliefs, often without any explanations, which are better found in his Discourse on Metaphysic of 1686. I had a go at trying to explain what I thought Monadology was about and how much it still applies, in a long piece at Qeios: https://www.qeios.com/read/8RDKGI

It may still be incomprehensible, but that was my best shot.
For a complete collection of the key works the best is Woolhouse and Francks paperback GW Leibniz: Philosophical Works. I found them very easy to read and engaging. They are all quite short.

Thank you, I plan on looking through these and giving some a read when my brain isn't fried from too much thinking!

 

[Jonathan Edwards]

There are problems with this, like why different impressions of similar objects would result in similar neuron positionings and what boundaries collection neurons use.

Your account gets the gist of how Sevush, Somov and I see this working. The neuronal does not change shape but the analogy gets the basic idea. Neuronal dendrites have about 20,000 point synapses laid out along linear branches, each of which can generate an electrical potential. You then have a branching manifold decorated with a complex pattern of potentials that forms a detailed electrical field pattern that can inform shifts in actions in the cytoskeleton. Meaning would be encoded in the geometry of the relations between the potentials in that pattern. In effect the model is a more complex version of a violinist putting fingers on strings to produce harmonics. The geometrical relations between fingers and bridge and neck encode the sound produced.

You have also identified the need for different cells doing different jobs. We need cells that get fed invariant features of eating pizzas, or grandmother, or Jennifer Aniston, and we also need cells that get fed the immediate context of the bite on the cheesy bit by the gherkin, granny's blue jumper with a button missing and Jenifer Aniston in semi profile in a low cut dress with hair up at an awards ceremony. I have published what now seems quite a popular paper on this in Frontiers in Psychology 2016. https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2016.01537/full

What is exciting is that an Estonian neuroscientist called Aru has actually proposed a model for a basic pattern relation within the neuron - a comparison between basal and apical dendrite inputs - and related it to consciousness, although he misses some of the points I have been talking about.

 

[Jonathan Edwards]

also think I agree with you on emergence. I wasn't trying to say that the subjective feeling of wetness is a property of water, just that there is some property of water (viscosity, its bonding structure, the molecular size) that does get perceived as wet for whatever reason.

Yes, I agree. My exploration of both quantum field theory and contemporary academic philosophy has made me think that the philosophers are stuck with a Neo-Aristotelean account that doesn't work in modern physics. They argue about whether or not 'systems' can have emergent properties. As I understand it we no longer have 'things' with 'properties' but only instances of operation of property packets. The fundamental units that have indivisible identity are chess moves rather than chess pieces. They have no chess piece properties, they are just instances of operation move properties, as for playing chess online.

The problem with classical systems is that they are not indivisible individuals - by definition they can be broken down into sub-actions. But generalised quantum field theory brings us new individuals when collective actions are closely correlated and these new individuals are packets of new properties. When a bell rings zillions of proton, neutron and electron individuals with mass undergo correlated movement. And with that you get a new action, a sound, which is an individual with no mass and integer spin - a phonon mode. So contra people like Strawson, there is 'radical emergence' in physics, but not of property systems, but of new dynamic individuals.

In this context the wetness of water has always been an unfortunate example because wetness probably reflects ordered correlated actions of molecules but translational action doesn't quantise if I understand things rightly. So the sound of a violin would have been a much better example for the philosophers to argue about.

 

[Eddie]

Your paper on roles for the individual neurons is great. I had to spend a few days thinking about it, but the vast majority of parts I understand I agree with. The first few pages on causal reasons and why there must be inner receiving entities were very helpful. I won’t mention all the things I agree with, but I did have some points that I could use clarification on.

Can you explain what the firing of a specific neuron refers to. Is the firing the same thing as a neuron reporting its representation to another neuron?

I am also a little confused why you say the pattern of firing and non-firing is what matters. If meaning does occur at the neuronal level, surely there can be no meaning derived from a neuron that does not report its representation. Why can’t we just say that we experience the neurons who do report their representations? The non-firing neurons only matter in the sense that if they were firing we would be experiencing something else and other cells would get different inputs in the future.

What is the difference between a dog-pontifical cell who knows to fire when it receives the dog inputs and a whale-pontifical cell who knows not to. Both cells know what the input is in the sense that they know that it represents the inputs, but why does one fire and one not do so? Is there some rule that is different for each cell that tells it to fire when its inputs are represented a certain way?

And I think this factors into my final question on why you think there is some brute fact that explains the correspondence between the physical system and the meaningful experience. I am quite skeptical about any brute fact as even in Descartes’ case I don’t agree that he shows that brute facts exist.

I think there are a few ways in which we can say we know what something means. One such way would be that we can express a thing in some other way. For a neuron they are expressing the electrical and chemical information in some other fashion and therefore is producing meaning. But I am not sure that that is enough to say the cell knows what the input data really means.

Wouldn’t it be more accurate to say that the real meaning comes from whatever reason the cell has that tells it what representations to pass to other neurons. The dog-pontifical cell attaches meaning to its representation of its inputs because it knows what specific signal to produce. It is not the outgoing signal that has the meaning, but, however that cell does its job of taking the inputs and either verifying or rejecting that it is a dog. In the case of a whale-pontifical cell it derives some meaning from the input signals by the fact that it represents these signals in a new way. However, I think it is more accurate to say the meaning is not currently being produced in this cell because it does not trigger whatever process verifies a whale. Maybe this is what you meant all along and I was just confused. But if this is the case I don’t think we need the brute fact. The meaning is the physical process in which the cell determines if the representation of its inputs is something it would like to pass on.

I'll probably have to read through your paper and your posts here a few more times, but I am certainly starting to see how the pieces fit together.

 

[Eddie]

It's also amazing that if this is how the brain functions than what I learned in school about being made up of cells is so much more true than I realized. We are quite literally not one thing but a colony of neurons that have no choice but to get along and communicate with one another.

 

[Jonathan Edwards]

Thanks for the comments Eddie. It sounds as if you may be one of the very few people who see what Steve and I are on about. There may be more because I get notifications of people reading that paper on an almost daily basis now - just in the last year or so. But I don't know if they get it!

Your questions about meaning are crucial and remind me that on one issue I have changed my mind - the firing and non-firing. There is something fundamentally different about the math of the neural geometry/meaning relation from that of a QR code.

The other questions I think are answerable in my original model terms. I wonder if you are familiar with Paul Grice and Studies in the Way of Words? It was the one bit of contemporary academic philosophy which for added a major new insight into how we should do biology. Grice explains how subtle and sophisticated our use of meaning is.

I will need to go through each question systematically. They are very important and trying to answer may trigger some new thoughts.

 

[Jonathan Edwards]

Can you explain what the firing of a specific neuron refers to. Is the firing the same thing as a neuron reporting its representation to another neuron?

The firing of a neuron N signifies that the neuron has made some inference about something going on (usually out there but in introspection it can be in here). N is telling other neurons that 'there is red at three o'clock' or 'there is a London bus'. The signal has a propositional meaning in that it relays a fact that can be used in a logical inference, even though it has no internal syntactic propositional structure of the sort we see in sentences. The next neuronal along will receive a whole mass of such propositions or facts and these will provide it with a scenario as a representation of what is going on. A scenario is not a proposition but more a total collection of propositions/ premises from which a logician can draw various inferences. Each cell is designed to draw a different inference, as and when appropriate.

So each cell reports the inference it draws from its incoming representation and each cell only ever reports the same inference. That might seem to require too many cells to cover all inferences but note that intermediate level cells may be reporting abstracted inferences like 'P happened just before Q', with other cells providing information about what P or Q are.

 

[Jonathan Edwards]

I am also a little confused why you say the pattern of firing and non-firing is what matters. If meaning does occur at the neuronal level, surely there can be no meaning derived from a neuron that does not report its representation. Why can’t we just say that we experience the neurons who do report their representations? The non-firing neurons only matter in the sense that if they were firing we would be experiencing something else and other cells would get different inputs in the future.

This is a fascinating problem that I have not seen unpacked properly in the literature. A number of approaches to consciousness suggest that it lies in those cells that are firing - I think what you suggest. This seems to apply to Tononi's Integrated Information Theory but it is widely held. It falls foul of the combination problem. But a number of people, including Bjorn Merker (who I think has important insights), and arguably Tononi too, treat the array of signals rather like a QR code or digital map of what is referred to. The extreme version of thesis the idea that the signals are not just isomorphic to the events they represent but homotopic - they are laid out in some space in the same arrangement as what you are seeing.

The big problem with this is that Marr, Barlow, Blakemore and others in the 1970s worked out that the coding in neuron signs must be propositional in the way I have suggested, or at least 'adjectival', such that the signals can be integrated, on arrival, logically (or syntactically)rather than spatially.

A number of people, including myself, also came up with the thought that Tononi's idea of consciousness residing in the cells firing at each other at a given time couldn't work because in a QR code both black and white, or + and -, contribute to the meaning. (Tononi actually avoids committing himself to any specific causal event account, despite his axioms requiring it, which is the basis of a critical paper I wrote with a student, but others go for the QR code/map model specifically.)

But more recently I realised that propositional meanings do not have two options - it is, and it isn't. There are three options: it is, it isn't and 'no comment'. You are suggesting that A signal means 'It is.' and no signalman's 'no comment', and for at least some neurons I think that must be right. The caveat is that for other neurons with different jobs it might not always be. The implication is that, unlike a QR code, you can have mutually contradictory inputs. And maybe in a way that a computer built on binary arithmetic cannot. You need to have two synapses in the same dendrite: one for saying 'it is' and another for saying 'it isn't, and both could be activated by oncoming signals.

Only recently, I discovered that this is relevant to Leibniz's correction of Aristotle's logic that paved the way for set theory and modern logic. Leibniz set out to devise a 'Universal Characteristic' which encoded all thoughts in numbers, later as ratios. He failed but ended up tidying up Aristotle. My thought is that the neuronal dendritic tree is actually where the Universal Characteristic is seated, with the ratios being geometric relations between active synapses. Leibniz was usually right and only failed to prove he was when he had no access to the needed data. My hope is that before I die I might find some clue as to how to crack the Universal Characteristic, but it is the hardest problem I have ever tried to address.

So the problems with consciousness being where cells are firing are that it violates locality by creating a combination problem and so is incompatible with our understanding of causation and it will not work if it is assumed to work like a map/QR code (it needn't). The first problem is enough.