Mind/Brain and ME theorising

[Jonathan Edwards]

This thread is intended as a place to discuss how theories of ME/CFS are best formulated in terms of concepts like mind and brain and 'me' and neurons and perceptions and action potentials and 'representations' and all sorts of other things that when confused can lead to misunderstandings and obfuscations and guilt and irrational treatments etc.

 

[Jonathan Edwards]

@Woolie, @alex3619, @Snow Leopard, @user9876
We have been discussing Mark Edwards's theorising about predictive mechanisms and 'beliefs' in brains. I wanted to reply to Woolie's defence of levels of description. I agree with a good deal but want to disagree, or at least shift the angle, on certain things because I think further discussion may be constructive. What is clear is that (1) the BPS psychiatrists have no real idea what they think so can be forgotten about (2) the neuropsychiatrists like Edwards and Harrison are trying to generate explicit theories but are making a bit of a mess of it and (3) beyond that there is general confusion about how concepts of mind and brain relate so that almost everyone is likely to be talking at cross purpose. The dilemma is that we want theories that are helpful to patients but are also rigorous scientifically. I am not at all sure that one description can serve both purposes but at least we might be able to do better than at present.

I don't actually buy the 'ordinary language usage' line, exemplified by Gilbert Ryle's Concept of Mind, that says that you cannot say that brains think and bodies walk. If you are an engineering lecturer you can say a wooden beam holds up a roof. You can then say that bundles of cellulose fibres hold up the roof, and then that a combination of certain chemical bonds involving carbon hold up the roof. That may not be ordinary language talk but it is sound science. We say a person experiences pain. But person is like beam; it only works for interpersonal relations in sociology or purely behaviourist psychology. In the context of pathophysiology of ME we are interested in what is going on within the person - intrapersonal relations between pain and and memory and action and emotion etc. Pain is still experienced at this level of analysis but we need a finer grained level entity to ascribe it to. (If thinking is the totality of all these relations then brain probably does fine for thinking, but brain will be too big for experiencing pain.) I don't think levels of abstraction really come in to this. Pain is real and to have a causal effect it must be experienced by (or impinge on or influence) some physical component of brain. Anything else really is dualism.

Locke and Leibniz argued over how to describe at the intrapersonal level - sensations, memories, emotions, actions - without referring to neuroanatomy. So we do not need to talk neurons specifically, and should not if we are not sure what does what and want to lay down some general principles. But Locke and Leibniz realised that there must be a specific answer to 'which components of brain actually experience pain'. And as pointed out by an unknown scholar with the pseudonym Hyperaspistes, we do not know how many components experience pain at any one time in a brain so the idea that there is only one 'me' or 'I' that experiences is not well founded.

What seems to me particularly unfortunate is that terms like 'belief' and 'prediction' are used in neuropsychology in ways that do not take into account the difference between beliefs that are experienced as such and 'operational' beliefs that a computer might have (e.g. that its battery is 72% charged). The distinction is avoided because the current dogma is that there are no special parts of the brain that experience beliefs, that experience 'emerges' from neural network activity. I see that as incompatible with physics and neurobiology.

So my bottom line thought is that Edwards and harrison are trying to head in the right direction of getting a rigorous scientific account of what may be going on in these conditions but they are muddling their accounts because of the current muddled dogma.

 

[user9876]

You can then say that bundles of cellulose fibres hold up the roof, and then that a combination of certain chemical bonds involving carbon hold up the roof.

You would tend not to. But if you said bundles of cellulose fibres you should talk about the bundling mechanism than binds them into a beam. If I were to separate out the fibres and just loosely bundle them together with a bit of string they would not function as a beam. The point is that as you go down in a level of abstraction you need to add the detail. The use of the words 'wooden beam' abstract away from a lot of the detailed construction of the beam in a way that people have a common understanding but as you get into the detail you need to not just talk about the parts of the construction of the beam but all of it and the relationships at least at the next level of abstraction. Of course what often happens is that people know what detail is relevant to a scientific model and what isn't so that some of the pieces simply get missed out because they are not important in the context that they are used in. But a bad model can contain the wrong abstractions and mislead.

My biggest concern about a lot of the mind/brain discussions and papers is that they include a lot of fluff but not much in terms of carefully specified concepts. Yet it is the latter we need to make progress.

 

[Jonathan Edwards]

You would tend not to. But if you said bundles of cellulose fibres you should talk about the bundling mechanism than binds them into a beam. If I were to separate out the fibres and just loosely bundle them together with a bit of string they would not function as a beam. The point is that as you go down in a level of abstraction you need to add the detail. The use of the words 'wooden beam' abstract away from a lot of the detailed construction of the beam in a way that people have a common understanding but as you get into the detail you need to not just talk about the parts of the construction of the beam but all of it and the relationships at least at the next level of abstraction. Of course what often happens is that people know what detail is relevant to a scientific model and what isn't so that some of the pieces simply get missed out because they are not important in the context that they are used in. But a bad model can contain the wrong abstractions and mislead.

My biggest concern about a lot of the mind/brain discussions and papers is that they include a lot of fluff but not much in terms of carefully specified concepts. Yet it is the latter we need to make progress.

I agree that you need the detail, but that is why I said bundles of fibres rather than just fibres. What matters is the way they are bundled, indeed. I totally agree that we need detail - that is really the point of my further discussion. But there is little doubt that at the cellulose fibre bundle level roof-holding-up is still going on. You can still measure thrust forces at that level.

 

[user9876]

Pain is real and to have a causal effect it must be experienced by (or impinge on or influence) some physical component of brain. Anything else really is dualism.

To put a different view pain is not real it is just a word that people use to express one or more different mental states or signalling mechanism. Pain itself could be parameterised in terms of it may be local to a given area or it may occur under certain conditions (e.g. under movement). I would argue mental pain (in terms of distressing thoughts) is very different from physical pain. When the word pain is used the context of the usage will often imply more details.

So then if we separate out different versions of pain the we could start to see it in terms of signals, feeding into mental states or processes that process the signals and communicate stuff to other mental processes. I would say that these mental representations and processes are just connection patterns or activation states etc between neurons but I'm not convinced we understand enough of this mapping to really talk at that level of abstraction when things are happening within the brain.

When talking of mental representations and processes we need to hypothesis how they may act and look for experiments that may provide evidence for such. But that is very hard. With language processing there were various experiments which looked at things like recall of sentences, processing speeds etc to help form a hypothesis but it seems very uncertain.

I'm not sure how you would go about doing so in terms of some of the more basic control functions of the brain. I can see the idea of a 'blackbox' type analysis looking at signals in and out but I'm not sure we know how to measure everything. I kind of see the idea of looking at areas of activation but I think we need to be very careful in using language about what is found and what is inferred. So if the activation of an area of the brain increases with PEM then that is what we have found. We can then speculate about what it might mean but that speculation step needs to be separate and drawn out carefully made specific so we may have a hope of testing it in some way.

 

[user9876]

Interesting conversation. If I can make a recommendation: You may wish to agree on definitions before diving in too deep.

I do not think you can totally avoid this from devolving into an exercise in linguistic philosophy.

But this looks like serious fun.

I tend to think part of the initial conversation is about understanding the different words and what people mean by them which is a necessary precursor to any definitions. People on PR do come from very different backgrounds and have different ways of describing and approaching problems. Having worked a bit in multi-disciplinary teams one of the exciting (and frustrating) things is the way people come at a problem using different approaches and bringing different knowledge. If you are lucky this can lead to some interesting and creative thinking and solutions.

 

[duncan]

You are 100% right. I tried to delete my post but I was too slow. I'm like that dog in the movie: I see a word or phrase, and "Squirrel!"

Sorry for the diversion from an excellent topic.

 

[alex3619]

What matters is the way they are bundled, indeed.

This is a preamble to discussing the idea of mind etc. I thought I would get some things out of the way.

In living systems particularly the structural arrangement is critical to their function, and this includes dynamic factors and concepts like autopoiesis. Its also important to realize that models are not reality, reality tends to be messier than predicted by many models, and care must be taken in inferring from models. Models are abstract. Then there is the issue of dynamic stability. In a cloud of gas you don't need to worry about that, in a preserved timber beam you don't need to worry about that (much), but even in a viral swarm in a host you have to think about that.

Systems biology is the attempt (still in its infancy) to put the pieces together so we can figure out how they work together. The human brain project is in crisis, the brain is still largely outside of our grasp. I do not credit the notion that we cannot in principle understand the mechanisms of brain, that we are not smart enough to understand our own brains. That is trivially true, in detail, just like I do not understand my own city. I don't understand where all the screws, and bolts, and nails, and concrete lumps are in the city. I do understand the general geography and government bodies etc. I hold that the principles of brain and brain function are understandable in principle, but its not something we understand.

If you look at the brain from a complexity viewpoint there is nothing more complex that we are currently studying. Its beyond the complexity, by many orders of magnitude, of any other problem.

Let me start by saying that the DNA does not encode the structure of the brain directly. It encodes mechanisms that work as biochemical rules. The brain precursor cells just apply the rules locally, by mechanisms we partially understand (a scientist who figured out some of those was the inspiration for my PhD attempt). As a consequence, due to local variation and environmental factors, including the external environment, even identical twins do not have the same brain.

Yet its the structural organization of the brain, and the way they work dynamically, that are in principle understandable.

If I look at genetics and the genomics revolution, two things stand out as essential to progress. The first was basic technology, such as gene shears. Without certain advances in technology, other advances become very hard. The second is automation - the ability to apply technology rapidly and in a very controlled fashion. Those two paved the way for the genomics revolution. The issue that much of that revolution is hype is something else again, and a social problem - new fads are often mostly hype.

I do not think we yet have the fundamental technology to really understand the brain. Yet the only way to get there is to keep advancing the science piece by piece until we discover a few things that help, and then we can leverage them with automation.

It is my working hypothesis that mind (as a separate thing) does not exist. I do not argue that it cannot exist, only that it is unnecessary and unproven, and not the simplest hypothesis.

A lot of the arguments in this area are category mistakes. Assumptions are made. Those assumptions are rarely questioned. Science, to me, is about questioning everything. I regard everything in science as an hypothesis, including so-called Laws. A Law is just an hypothesis that is robustly supported. It can still be questioned if the data warrants that. Such as the old Law of Gravity. I also think that science can be applied to absolutely every topic, at least in principle. I do not think that is currently always wise though. Science is a method for asking questions of the universe, not a set of dogmatic facts.

Philosophically I am a pancritical rationalist, of which critical rationalism is a special case. A modern variant of that is empirical scepticism, though scepticism comes from different historical philosophy. Show me the data, and show me the reasoning and realize that progress occurs through questioning, and in science that means experimentation. Indeed in some ways I think questions are more important than answers. The right question, including research questions, can lead to a good answer. The wrong question may give you an answer that is not very good. Of course there can be debate about what is right or wrong or more or less useful, I am just trying to give the flavour of my position.

Much of what I call psychobabble is not because the ideas are impossible. Its because they are badly formulated, hypothetical, unproven, and rely on successive layers of data that only suggest the hypothesis is right, and never test it, and ignore contrary evidence. They seek to verify their ideas, and operate on the 19th century principle that all you need in science is to accumulate enough data so your data pile is bigger than that of the critics.

 

[Jonathan Edwards]

To put a different view pain is not real it is just a word that people use to express one or more different mental states or signalling mechanism. Pain itself could be parameterised in terms of it may be local to a given area or it may occur under certain conditions (e.g. under movement). I would argue mental pain (in terms of distressing thoughts) is very different from physical pain. When the word pain is used the context of the usage will often imply more details.

I agree we need variations and details and pains like grief distinguished from a stubbed toe.

But I think pain has to be real. It is a form of experience and all we can say about the 'physical world' is that it consists of those patterns of dynamic relation (causes) that directly or (more usually) indirectly determine the patterns of our experiences. So experience has to be real, it is the rest that we are less sure of - as Descartes says. That does not mean that the physical world is not there without human observers, just that without experience we could have no concept of physicality. P a i n is a word, but it must have a real referent. The referent does not need to be 'stuff' though; it is much more likely to be a dynamic relation - or perhaps an event.

So then if we separate out different versions of pain the we could start to see it in terms of signals, feeding into mental states or processes that process the signals and communicate stuff to other mental processes.

That seems spot on to me.

I would say that these mental representations and processes are just connection patterns or activation states etc between neurons but I'm not convinced we understand enough of this mapping to really talk at that level of abstraction when things are happening within the brain.

This is where I would want to nit pick. Connection patterns are there for hours, days or more often years. A pain is going to be some signals using the connections. Activation states are usually taken to mean cell firing in terms of action potentials and I think this is NOT what pain would be. Saul Kripke famously suggested that pain might be, for sake of example, 'c-fibre stimulation'. Fair enough, and he realised the stimulation was probably further up in the thalamus or cortex. But because neurobiologists are fixated on measuring activity, being all they can measure, David Papineau chinese whispered this into 'c-fibre firing'. But pain is a sensation. Sensation is an input. Experiencing is being influenced by the world, not influencing the world. So I would go back to your previous suggestion: 'signals feeding in to processes' - perfect. Of course the problem is that nobody can detect this in a brain with EEG or fMRI because these measure local output, not input. But inputs can be measured in patch clamped neurons in things like the cuneate nucleus of a cat using implanted electrodes (seminar last week on this) and so at least in principle they can be found.

When talking of mental representations and processes we need to hypothesis how they may act and look for experiments that may provide evidence for such. But that is very hard. With language processing there were various experiments which looked at things like recall of sentences, processing speeds etc to help form a hypothesis but it seems very uncertain.

As indicated above, this is indeed almost impossibly hard. But people like Mark Edwards, following basic neuroscientists like Patrick Haggard and Jakob Hohwy, are at least drawing flow diagrams of how they see their predictive coding or Bayesian models working and suggesting that 'box A' might be striate cortex, 'box B' might be prefrontal and 'box C' might be thalamus or whatever and saying where the predictions and inputs are compared. But they refuse to say which inputs to what are experiences - because they have this dogma that this would be 'Cartesian' and illusory. But if nothing gets the messages that encode pain as its input there cannot be any pain event.

I kind of see the idea of looking at areas of activation but I think we need to be very careful in using language about what is found and what is inferred. So if the activation of an area of the brain increases with PEM then that is what we have found. We can then speculate about what it might mean but that speculation step needs to be separate and drawn out carefully made specific so we may have a hope of testing it in some way.

So I guess I agree with your caution here in spades.

 

[Jonathan Edwards]

I agree with your general principles Alex, but I am leery of certain specific concepts.

In living systems particularly the structural arrangement is critical to their function, and this includes dynamic factors and concepts like autopoiesis.

To my (non)mind Maturana and Varela's theory of autopoiesis is wrong. The idea that subjectivity, or sentience, arises out of far from equilibrium self-sustaining thermodynamic states has no basis in anything else we know in science and fails miserably in the face of facts.

Similarly I see systems biology as a throwback to the Medieval thinking that Descartes, Locke and Leibniz de-bunked. Systems are arbitrary constructs in the minds of scientists, not individuals as von Uexkull or von Bertalanffy would have it. For sure everything is dynamics. And for sure we need to think in terms of complex systems dynamics, with feedback loops and attractors and metastable states and all that but we were doing that before anyone started writing grants with systems biology in the title. Systems biology has now taken over as a catch all buzz word that has to go into every grant and it mostly does so to cover up for any lack of intellectual content as far as I can see.

What I think is crucial is that the things we think of as 'mental' in terms of sensation or experience or conscious belief do not get magically explained away by non-local dynamic networks when to do so would require a violation of very basic rules of physical dynamics. If we want to suggest that there are some other rules of 'mental dynamics' that only apply inside heads we can, but that would be more dualist than Descartes and hard to justify without some evidence. And one of the key features of non-local theories is that they are more or less by definition untestable - always contingent on what is happening somewhere else. That suits the hand wavers but it is not science.

I agree that the human brain project is likely to be in crisis, but I suspect more than anything because nobody is sticking to physics! I agree that there is no separate thing called a mind, everything is physics, but we may need to think more carefully about the individual entities involved in that physics to explain subjectivity.

 

[user9876]

Systems biology is the attempt (still in its infancy) to put the pieces together so we can figure out how they work together. The human brain project is in crisis, the brain is still largely outside of our grasp. I do not credit the notion that we cannot in principle understand the mechanisms of brain, that we are not smart enough to understand our own brains. That is trivially true, in detail, just like I do not understand my own city. I don't understand where all the screws, and bolts, and nails, and concrete lumps are in the city. I do understand the general geography and government bodies etc. I hold that the principles of brain and brain function are understandable in principle, but its not something we understand.

In the past We've looked at modelling systems (involving computers, people and processes) with the idea of executable models where each part of a model describes part of the process or an interaction including resources that are being interacted with. This way each part of a model can be described relatively easily but the interesting behaviours come as the model is executed hence helping understand what behaviours the interactions bring. In this case the modelling was underpinned by a system of logic and process calculus which can then be run as a simulation of the system to understand combined behaviours.

So the idea that you say of understanding parts of a system is to my mind very necessary.

With any model of course it is important to understand that it is just an abstraction and that any predictions can be wrong. What is often interesting is to run variations of a model to see which it is sensitive to and which can be varied. This can give an idea of where detail and getting the abstractions right and where detail may be less important.

 

[user9876]

But I think pain has to be real. It is a form of experience and all we can say about the 'physical world' is that it consists of those patterns of dynamic relation (causes) that directly or (more usually) indirectly determine the patterns of our experiences.

I wasn't trying to deny the experience of pain. But I was trying to make another point about categories and descriptions. To me pain is a overarching concept that could be a distinction of a number of different more specific pain concepts. In itself I feel it is too general a category to usefully instantiate but rather we need to look at the different forms of pain, differentiate, describe in more detail and deal with these individual pain concepts. At some level of abstraction it is useful to talk of a generic concept of pain but when talking about reasoning about pain I feel we need to be more specific and talk about the different individual categories.

 

[user9876]

This is where I would want to nit pick. Connection patterns are there for hours, days or more often years.

Yes do nit pick. As I was writing I was starting to wonder what is a memory as opposed to what is an mental process that is happening, and what is a mental representation that is part of a reasoning chain vs a longer term memory and how are they represented in the brain. Naively I was thinking connections would basically encode longer term memories whereas activation patterns would be current working memory - But I have no real reason to think that apart from mapping on to artificial neural networks (as computer scientists would talk about them) which have nothing to do with the way the brain works.

But because neurobiologists are fixated on measuring activity, being all they can measure, David Papineau chinese whispered this into 'c-fibre firing'

I think that fMRI scans actually measure to deoxygenation of hemoglobin so this is probably activation in some sense but not necessarily. I have a vague memory that neurons have a kind of pulse finding which would suggest longer term activations rather than a transitory firing which (I assume) would be harder to pick up with fMRI.

But inputs can be measured in patch clamped neurons in things like the cuneate nucleus of a cat using implanted electrodes (seminar last week on this) and so at least in principle they can be found.

I assume this is a somewhat dangerous or destructive process.

As indicated above, this is indeed almost impossibly hard. But people like Mark Edwards, following basic neuroscientists like Patrick Haggard and Jakob Hohwy, are at least drawing flow diagrams of how they see their predictive coding or Bayesian models working and suggesting that 'box A' might be striate cortex, 'box B' might be prefrontal and 'box C' might be thalamus or whatever and saying where the predictions and inputs are compared. But they refuse to say which inputs to what are experiences - because they have this dogma that this would be 'Cartesian' and illusory. But if nothing gets the messages that encode pain as its input there cannot be any pain event.

I feel slightly nervous about strict flow diagrams because this tends to suggest a linear sequential style of processing which in itself is an assumption but it may be correct. There could be different processes linking to common mental representations. The other thing is to see feedback loops and loops to memories that may influence. But I think it is necessary to have detailed models as they seem to be developing and then to make predictions that may be in some way testable.

 

[Jonathan Edwards]

@User I agree with all of that.

And now you mention it the flow diagrams can't be the way they draw them. The brain is neither sequential nor parallel but constantly both divergent and convergent. I have a feeling that one needs some new trick of maths that can handle this statistically in a manageable way. The cat cuneate nucleus study was actually very neat. It shows that the touch receptor neurons at their first point of synaptic connection in the brain stem feed into cells that seem to fire in response to every possible combination in time and space of receptor firing. Even at this stage cells are picking up patterns that mean 'greasy' or 'woody' or 'woven' it seems so that the cortex is being fed high level data right from the start. And each of those signals is being sent to about 10,000 places at each connection cycle.

 

[alex3619]

To my (non)mind Maturana and Varela's theory of autopoiesis is wrong. The idea that subjectivity, or sentience, arises out of far from equilibrium self-sustaining thermodynamic states has no basis in anything else we know in science and fails miserably in the face of facts.

At its basic level autopoiesis has nothing to do with sentience. When they discuss this later on it was an enhancement of the basic theory and not that convincing. I actually discussed autopoiesis with Maturana in the early 90s, at a seminar. Autopoiesis is more akin to the notion of a strange attractor in chaos theory. It is stable states, what Maturana and Verella call structurally coupled states, that maintain equilibrium in biological systems. In one sense though this is trivially true. It is merely a label for something that happens, where stable biological systems have a higher chance of survival.

Now structural coupling is very relevant to brain function, but that is a whole other argument.

 

[alex3619]

Systems are arbitrary constructs in the minds of scientists, not individuals as von Uexkull or von Bertalanffy would have it

Almost nobody in Systems Theory thinks otherwise. The really old arguments are just that, really old. Nobody uses them. However when you focus on a small piece in a large dynamic system, and you devise hypotheses and experiments, you are creating arbitrary divisions. Drug companies do this in drug design, and often wind up with consequences they did not intend, such as the Vioxx fiasco.

PS In computer science this is similar to the closed world hypothesis and sometimes leads to unintended problems.

 

[alex3619]

Systems biology has now taken over as a catch all buzz word that has to go into every grant and it mostly does so to cover up for any lack of intellectual content as far as I can see.

I have discussed social fads in science elsewhere. What you describe happened with genetics and genomics. After some time people realized the limitations and changed how they did things. Some approaches are viable, others produce spurious results full of false positives. Sadly some of the genomics fad is still evident.

Systems biology is unlikely to be much different.

I do not disagree there is a lot of hype. For one thing, this kind of approach does not create answers. I do not think it should even be intended to. However when faced with large unstructured problem domains, such as disease states in the body, where do you focus? The low hanging fruit, that which is easy to find and do something about, has in many cases been addressed. Sometimes things got missed, but over time those will be resolved. So where do you decide to start experimenting? How do you reduce the risk of irrelevant experiments? Where do you put limited resources?

Systems biology, in my current view, is a partial answer to that. It can assemble interactions that suggest where to experiment, what to try. Its that experimentation that gets the results, not the systems biology. However its the systems biology that is important to more organized hypothesis generation. More imporantly it can identify possible confounders and associations, and hence refine experimental design.

 

[Snow Leopard]

The one thing that I've learned, at least with the (modest) experiments that I've done in my life is: be careful what you measure.

Just as questionnaire results are subject to bias and don't necessarily reflect someones true health state, we cannot assume that statistical correlations in fMRI data necessarily reflects increased perception of pain (or whatever), unless there is lots of strong additional data of other factors at the same time.

Neuroskeptic has written more than a a few articles on the issues with this technology:
http://blogs.discovermagazine.com/n...itives-in-fmri-spm-fsl-and-afni/#.ViamRytCi1c
http://blogs.discovermagazine.com/n...and-this-time-its-not-just-fmri/#.ViamaytCi1c
http://blogs.discovermagazine.com/neuroskeptic/2010/03/10/can-we-rely-on-fmri/#.ViamhStCi1c

What is really measured when using various imaging technology?


@user9876 mentions that fMRI is sensitive to deoxygenation of hemoglobin.

Obvious consequence: studies of parts of the brain that are nearby blood vessels, for example, the amygdala are subject to a high degree of confounding.
http://blogs.discovermagazine.com/n...mri-of-the-amygdala-all-in-vein/#.Viam1ytCi1c

My own knowledge of how these technologies work is quite weak.

So my suggestion, at least from an empirical point of view is to start with some questions about the fundamentals of the science.

How do each of the imaging technologies work?
What are the common biases?
Which are likely to be the most useful?
What new and novel technologies are being developed?

Then we can follow up with examples like examination of novel studies which have tried to investigate directionality of brain imaging findings (eg peripheral or high sensitivity within the brain)? and so forth.

 

[Woolie]

Hey there, I don't have a lot of time, but was just talking with @user9876 on the other thread so will add some relevant points here:

Here is the problem with confusing levels of description:

People tend to be seduced by "brain talk", and can think if a brain part is involved suddenly a psychological explanation is better, more satisfying. Have you seen this:

The seductive allure of neuroscience explanations
Advanced psyc students were asked to read explanations of a psychological study. Some of explanations threw in a "brain part", which had nothing to do with the explanation, just said what part of the brain might be involved (not how it was involved). The students rated the explanations with the brain bits as being more satisfying.

This is all interesting stuff, but there's a real reason for my worry: the psychobabblers are now onto this. Just cos something mentions "the brain" does not mean it isn't the same old psychological stuff repackaged, with CBT as its recommended treatment. For this reason, we should be very wary of the interoceptive network. I didn't say before, cos I wanted to hear what others think, but I suspect the outcome of such work in MECFS is very likely to be a recommendation of CBT or maybe antidepressants.

I guess I would argue that it is not sufficient to talk of mind or brain but proposals should be made in terms of specific neurological things or mental processes.

You should be able to explain the phenomenon at either level of description. That's the test of whether you're clear about what the brain-level stuff adds (sometimes, nothing, just a location in the brain where these mental processes might happen. But sometimes - just sometimes - it adds new insight).

@Jonathan Edwards, I think good cognitive neuroscientists are clear on the levels of description thing, and they understand that some questions involve knowing where/how at a neural processing level and some questions involve knowing what those insights can - and can't - tell us about cognition more generally. I'm more worried about how others outside the field may be misled by the blurring of levels.

 

[Woolie]

Okay, got a bit more time, so here goes...

What seems to me particularly unfortunate is that terms like 'belief' and 'prediction' are used in neuropsychology in ways that do not take into account the difference between beliefs that are experienced as such and 'operational' beliefs that a computer might have (e.g. that its battery is 72% charged). The distinction is avoided because the current dogma is that there are no special parts of the brain that experience beliefs, that experience 'emerges' from neural network activity. I see that as incompatible with physics and neurobiology.

There are sophisticated cognitive models out there that make very clear distinctions between these different kinds of "beliefs". You don't need brain talk to do that. Cognition is way ahead of neuroscience on these things - just look at Seidenberg and McClelland, who in the 1990's were developing computational models of how each individual learning episode contributes to, and subtly changes our knowledge base. Prediction, feedback loops, adjustment of weightings in response to unexpected feedback, all these things have modelled computationally without reference to the brain. But a lot of the neuroscience is being done by those who don't know this literature (and that's probably not the neurospsycs, that's the psychiatrists and the neurologists).

Sometimes neuroscience evidence can be useful in building models of how we think. For example, cognitive models discuss things like one-trial learning, incremental learning (and the reason why we might need each for different purposes). It turns out the hippocampus is a great model of one-trial learning, the neocortex of slow incremental learning and the medial temporal cortex outside the hippocampus as somewhere in between. Its all a question of knowing how to use this evidence in a way that isn't just neurobabble. And step 1 is - cognitive neuro 101 - is understanding the levels of description distinction. Like so many rules, only when you get it can you can know when its okay to violate it!

I don't expect there'll ever come a time when we can localise any particular type of experience to a particular brain part. Where do emotions happen? Ar they in the bodily changes that accompany them? (e.g., heat rate increases)? Or are they in the brain responses that occur? Early psychology tried to pitch these as either/or possibilities, but of course, they are very likely to be all true. Emotional experience is likely to reflect the sum activity of a constellation of interacting and constantly changing brain and body states.