Well, if we're not 'overstating', you only know what you currently remember about what happened when you tested the model.

All thought is post hoc by at least a few milliseconds. — Isaac

There is no point questioning memory as such; if we have no faith at all in memory, then we can have no faith in any knowledge at all. Memory is the foundation of who we are, to question it in a general way would be absurd.

You say we know hidden states, via inference; that our actual experience just is inference, if I've understood you. This makes no sense to me. We infer that there are hidden states, and by definition, they being hidden, we don't know them in the common sense of "knowing'. We don't know what those inferred hidden states are; if we did they would not be hidden.

You haven't addressed this earlier response:

...is exactly what I'm arguing for. There is nothing whatsoever about these 'hidden states' which prevents us from naming them. In fact, I think that's exactly what we do. The 'hidden state' I'm sitting on right now is called a chair. It's hidden from my neural network because the final nodes of it's Markov boundary are my sensorimotor systems. It's not hidden from me, I'm sat right on it. — Isaac


If you are the body is it not, along with the chair, a hidden state (or as I would prefer to say hidden process)? Of course we can name them, but it seems we are doing so from within the familiarity which constitutes our common and also individual experience. — Janus

So, I am saying that naming is just a matter of fiat: we can call the hidden state a chair, or we can call it the unknown whatever that appears to us a chair. But if we want to say it is a hidden state at all, then it seems contradictory to say that it is "really" a chair, since we have already acknowledged that we think it is "really" a hidden state, and a chair cannot be anything but a familiar object.

I should mention that mindfulness awareness is not quite what Hussel or Merleau-Ponty had in mind — Joshs

I’m aware of that but it is discussed in The Embodied Mind, which draws on cognitive science, phenomenology and Buddhist abhidharma. So they talk about it.

All awareness is self-transformation, it is about something other than itself even when reflecting back on ‘itself’. — Joshs

I don’t think that should be taken as axiomatic. ‘Consciousness without an object’ is part of the lexicon of Eastern philosophies (and of American mathematician and Vedantist philosopher Franklin Mereill-Wolff (ref)

He is attempting to explain mental features such as gestalt perception as evolutionarily formed products of simple mechanisms of material reality — Joshs

Seems like a perfectly sound conjecture to me. And has the advantage of being at least compatible with evolutionary theory. What Pinter *doesn't* get into, in my opinion, is the sense in which h. sapiens transcends the biological, but as I said, that kind of subject is out of scope for the book.

Phenomenology dumps Pinter’s rule-based material and mental realities in favor of a united reality that is relationally relative through and through. This is what I mean by rebuilding the building. — Joshs

That's why you end up with the kind of relativism which I don't subscribe to. In that respect I'm probably a lot closer to Kant and neo-Kantians than to a lot of current phenomenology.

Pinter is closer to Dennett than you might think, — Joshs

I don't see it. He doesn't agree with materialist philosophy of mind. Dennett explicitly says that the mind is reducible to physical and chemical laws, while Pinter says:

The mystery is that in an age when physics has carried us into such a fantastic and unimaginable reality, we still balk at the idea that there are mental phenomena which do not follow the rules of classical physics. Why is it so hard to accept that in a universe in which space-time bends and curves, where particles of matter weave in and out of existence, and space itself is particulate—why would it be strange to accept that the mind of living animals is something complex whose laws are not the same ones that have been familiar to us for centuries?

I interpret this to mean that doesn't agree that the laws which govern the mind are continuous with physical and chemical laws.

Obviously our reading of Charles Pinter is very different, but I appreciate you have at least acknowledged this book, as it seems an important book to me. The fact that he's not a philosopher works in his favour in my view.

It's occured to me that you could say that the way the world exists in the absence of an observer is indeterminate. Whereas the realist picture is that the vast Universe exists before anyone is around to observe it, what that view doesn't understand is that to the extent this is a picture, even a scientifically-informed picture, the mind is implicitly the author of that (or any) picture or theory or measurement. It is the work of the 'unseen seer'.

Furthermore that the ground for the unreasonable effectiveness of mathematics in the natural sciences, is that mathematics conforms to formal structures within conscious experience and so overcomes the perceived separation between observer and observed that seems to be a basic fact of existence, but is actually not.

if we want to say it is a hidden state at all, then it seems contradictory to say that it is "really" a chair, since we have already acknowledged that we think it is "really" a hidden state, and a chair cannot be anything but a familiar object. — Janus

Its nature is indeterminate. And so it can't be said to exist, because what exists is determinate (i.e. it is 'this' or 'that'.)

mathematics conforms to formal structures within conscious experience and so overcomes the perceived separation between observer and observed that seems to be a basic fact of existence, but is actually not. — Wayfarer

Does mathematics overcome the perceived separation? It seems to me that it feeds into the illusion of a separation by treating extension and duration objectively, when they properly belong to the a priori structure of the mind.

Space and time have their actuality in our immediacy, yet they are mediated through mathematics, which also belongs to the immediate (apriori) structure of the mind, and they ultimately come to appear as something separate - something external that we apprehend and manipulate, when it is actually, to put it crudely, simple masturbation.

There is no point questioning memory as such; if we have no faith at all in memory, then we can have no faith in any knowledge at all. Memory is the foundation of who we are, to question it in a general way would be absurd. — Janus

We cannot know something we can't remember. Socrates was correct.

It seems to me that [maths] feeds into the illusion of a separation by treating extension and duration objectively, when they properly belong to the a priori structure of the mind. — Merkwurdichliebe

But obviously they have objective consequences. Almost all modern technology - no, not 'almost' - relies on the predictive and descriptive accuracy of mathematics, including the devices that are mediating this very dialogue. In Wigner's classic paper on the Unreasonable Effectiveness of Mathematics, the word 'miracle' occurs twelve times, along with the observation that

The miracle of the appropriateness of the language of mathematics for the formulation of the laws of physics is a wonderful gift which we neither understand nor deserve. — Wigner

There are interminable arguments in philosophy of mathematics as to whether maths is invented or discovered, whether it's in the mind of humans or is something real in the world. But I'm saying the regularities and rational relationships inhere within the conscious experience-of-the-world - so it's neither 'in the mind' nor 'in the world', and that this indicates a deep philosophical issue.

There are interminable arguments in philosophy of mathematics as to whether maths is invented or discovered, whether it's in the mind of humans or is something real in the world. — Wayfarer

And that debate is no different than the debate of whether space and time are only a construct of the mind or have external existence.

But I'm saying the regularities and rational relationships inhere within the conscious experience-of-the-world - so it's neither 'in the mind' nor 'in the world', and that this indicates a deep philosophical issue. — Wayfarer

I agree, its a very deep philosophical issue. It doesn't help when you speak in riddles. You mention "regularities and rational relationships inhere within the conscious experience-of-the-world", and then say "so it's neither 'in the mind' nor 'in the world'". I always thought the mind was associated with "conscious experience".

So then, where is this "conscious experience-of-the-world" if it is not in the mind nor the world?

Solipsism maybe? No, that can't be

The miracle of the appropriateness of the language of mathematics for the formulation of the laws of physics is a wonderful gift which we neither understand nor deserve. — Wigner

I say: The miracle of the appropriateness of the language of extension and duration for the formulation of the laws of physics is a wonderful gift which we neither understand nor deserve.

, isn't in/determinate epistemic, whereas being "this" or "that" is ontological?

(ah, nevermind me, time to turn in)

Hey thanks for clearing that up, then.

I've launched a new thread on this question.

Hey thanks for clearing that up, then. — Wayfarer

Piece of cake. :kiss:

You were talking about perception. If you're now talking about the future, then no, I don't think we can know the future (in general). I'm claiming we know that which we have a successful model of. The success obviously requires testing in the present.

I don't think we can know the future (in general). I'm claiming we know that which we have a successful model of. The success obviously requires testing in the present. — Isaac

Which can, at best, give us a more or less arbitrary approximation of the future, never any certainty.

If one expands “the network doing the inference” to include the sensorimotor systems, what happens to the hidden state? — NOS4A2

Then the hidden state would be whatever lay outside whatever nodes you had as the new Markov boundary. The existence of hidden states is just a mathematical outcome of the focus on a system.

If we're talking about a system (whether that's a person or a teacup), it has an internal and an external - that which is it, and that which is not it. Without those basics, we can't be talking about anything, we're simply talking about everything. The idea of hidden states is simply saying that with internal/external defined, you must also accept that there is a boundary, and with a boundary there must be (in information terms) a (set of) final node(s) at that boundary.

So that's a system.

Active inference occurs in self-organising systems (a subset of all the systems described above). The self organising bit meaning that they actively work against a probability gradient (the most probably distribution of our component parts is entropic - randomly scattered). The components of our body are a stage above in terms of less randomness than the components of the world we find ourselves in (which are somewhat more randomly dispersed). So we (our bodies) represent an active inference system, we are working against a probability distribution to maintain our arrangement against random dispersal, at a rate slightly higher than that of our surrounding system (the ecosystem).

As each system has final nodes (which cannot form part of the inference) we should look for final nodes in the human system (having identified it a s an active inference system above). Those nodes appear to be the sensory neurons (the passive element), and our motor actions (the active element - though many include secretion here too).

You can, of course, focus on any system you like with any boundary, but some systems would not count as active inference systems because they cross a threshold of the degree to which they are working against a particular probability distribution of the dispersal of their components.

can an activity that only organisms can be shown to perform—experiencing, thinking, inferring, believing, seeing—be isolated to a single part of it? — NOS4A2

I think so, yes. We can simply remove parts and see if the process continues without them (if we were evil experimenters - otherwise we can just wait until it happens anyway, then study the result). People without arms show absolutely no sign of reduced thinking, inferring, believing or seeing. So we can infer that arms are not necessary. People with damage to the occipital cortex cannot see properly, so we can infer that without an occipital cortex, you can't see.

Well, if you consider that each and every internal organ has its own purpose, function, relative to the existence of the whole, which is a living being, then you would understand that each of these organs has a reason for its existence. — Metaphysician Undercover

Not how I use the word reason. If you're talking about purpose, then fine. The purpose of neurons is not to represent the outside world. We know this because there doesn't appear to be anywhere to store that representation, nor do any of the areas involved in seeing, hearing, feeling etc seem to have inputs from areas outside the external-facing sensory organs which are sufficient to explain their activity. Hence we do not 'see' a representation, we 'see' the external world. I can literally watch a message go from the external world to the retina, to the occipital cortex and be recognised by the V4 region as being a certain colour. If I damage, or temporarily cut off that route the subject can no longer see colour. I can't think of a more clear demonstration that the seeing of colour is data from the external world, not an internal representation.

There is no point questioning memory as such; if we have no faith at all in memory, then we can have no faith in any knowledge at all. Memory is the foundation of who we are, to question it in a general way would be absurd. — Janus

No. That was my point (reductio ad absurdum, I believe it's called?).

You say we know hidden states, via inference; that our actual experience just is inference, if I've understood you. This makes no sense to me. We infer that there are hidden states, and by definition, they being hidden, we don't know them in the common sense of "knowing'. We don't know what those inferred hidden states are; if we did they would not be hidden. — Janus

That's not how we use the word 'know'. We use the word 'know' to refer to successful models of hidden states. I say something like "I know where the pub is", by which I mean that if I go to the place I believe the pub is, I will find it there. We don't use the word 'know' to refer to thinks we have a direct node-to-node connection with - I don't think anyone outside of cognitive computational neuroscience would be able to use the word 'know' if it were thus restricted.

I am saying that naming is just a matter of fiat: we can call the hidden state a chair, or we can call it the unknown whatever that appears to us a chair. — Janus

Yes. all language is by fiat. There's no book of what things 'really' mean.

if we want to say it is a hidden state at all, then it seems contradictory to say that it is "really" a chair, since we have already acknowledged that we think it is "really" a hidden state, and a chair cannot be anything but a familiar object. — Janus

Uh huh. And why can we not be familiar with hidden states? If we have good models of them, we can be very familiar with them.


Active inference is a scientific theory, that means it's an attempt to describe the world and be able to better predict how aspects of it work. It's not a normative theory, it's not saying we ought use words this way or that, it's describing what is happening in the events we already use words for. We do not need to change our words to match the science. The science describes better what the words are already referring to.

Active inference describes, for example, what 'seeing' is. The intention is not that we say "Ah so we don't really 'see' things then", what 'see' means doesn't change, we're just describing what goes on in the process in more detail. Same with the word 'know'. The process of 'knowing' is not what we (perhaps) thought it was, we have provided some more detail describing what the process of 'knowing' something involves. We've not proven that we don't 'really' know anything. The human activity/state that the word 'know' describes still exists unaltered, we just now have a better model of how it works.

were talking about perception. — Isaac

Sorry, I wasn't talking about perception. I was talking about hidden states in general. It makes more sense to me to say we have some degree of confidence in our inferences. That lets me distinguish hidden states from observations.

I was talking about hidden states in general. It makes more sense to me to say we have some degree of confidence in our inferences. That lets me distinguish hidden states from observations. — Tate

Ah, then we're using two different meanings of 'hidden states' which is causing the confusion. I'm using hidden states in its technical sense with regards to Markov bounded systems. An observation, in this sense, is an observation of a hidden state.

I know it's a bit confusing to call a state 'hidden' and then talk about us directly seeing it, but that's just the technical terminology, it will be impossible to read about active inference, nor understand any of the work in this field without understanding what is meant by a 'hidden state' in its technical sense.

But I'm saying the regularities and rational relationships inhere within the conscious experience-of-the-world - so it's neither 'in the mind' nor 'in the world', and that this indicates a deep philosophical issue. — Wayfarer

This is what Plato pointed to years ago, and it is why there is no interaction problem for dualism. "Ideas" actually exist within the medium between soul and body, and are therefore themselves part of that very interaction. In Aristotelian hylomorphism, and consequently Marxist materialism, ideas consist of both a formal aspect, and a material aspect (content, subject matter).

Not how I use the word reason. — Isaac

OK, but "reason" was my word, and it was used synonymously with "why":

The physical instantiation is the model. the thing represented by that model is neurons. The point being that we cannot determine the reason (why) for the thing, through reference to the reason (why) for the representation. So we cannot determine whether the neurons act representatively, through reference to the model, because the model represents how the thing behaves, not the reason (why) for that behaviour. — Metaphysician Undercover

So you giving "reason" a different meaning was nothing but trying to argue through equivocation. And now you attempt to explain away that equivocation, by implying that you were trying to change the subject and create a distraction from what I was saying (to talk about something different, to use 'reason" in a different way).

The purpose of neurons is not to represent the outside world. — Isaac

You seem to have missed what I was saying to Banno, and just butted in to the conversation using words in a completely different way. In no way was I saying that the purpose of neurons is to represent the outside world. I was actually arguing the opposite to that, denying the reality of an outside world. I was saying that the human 'system' (if you'll allow me to use this word in an informal way), creates images, ideas, and such things within the mind, as symbols, which have meaning.

What I argued is that the symbols, (just like words for example), need not be in any way similar to any "external thing". Dreams are a good example. So the images within our minds whether they have sense input, or are derived through the dreaming mind, or some other creative way, relate to other things in a way which is analogous to the way that words, as symbols, relate to other things. There is no need for any sort of similarity between the symbol, and the thing "represented" by the symbol, constituting the meaning of the symbol.

Now, I do not think I even used the word "represent". The word I chose was "symbolize":

Any pattern could symbolize something. And not all symbols necessarily appear like symbols to everyone. — Metaphysician Undercover

But Banno asserted that it is known that neural nets are not representational. So my root word "symbol" got replaced by Banno's root word "represent", and you use Banno's word in a completely different way, to assume an "outside world".

To be clear, I stated earlier that there is no need to assume any "external world" at all, and extreme skepticism which doubts the reality of an "external" world. is well justified. To expand on this, I will say that it is highly possible that all of our relations with the so-called external world, including sensations, and communications with others, is done internally. It could well be the case that we simply model these activities as occurring through an external medium, but our models are backward, upside down, as often turns out to be the case (geocentrism for example), and all these activities actually occur through an internal medium. What is external to us may be absolutely nothing, a wall of nothingness could constitute the external boundary to one's body, as 'outer space', nothingness. Then the real medium, as that which lies between you and I as a separation, and through which we communicate, may be through the internal, 'inner space'.

In the uses of the term I'm familiar with "observed" and "hidden" are very distinct. Could you give an example of the two being used to refer to the same thing?

To be clear, I stated earlier that there is no need to assume any "external world" at all, and extreme skepticism which doubts the reality of an "external" world. is well justified. To expand on this, I will say that it is highly possible that all of our relations with the so-called external world, including sensations, and communications with others, is done internally. — Metaphysician Undercover

This is simply not possible (where 'internal' applies to some self-organsing system). To recognise a system, a self organising one, there has to be an 'internal' and an 'external' otherwise you're just referring to 'everything', and a self organising system has to have a probability distribution function that is opposed to the Gaussian distribution, as this is just the definition of self-organising.

So simply by the definition of a discrete system we've got, of necessity, an internal state, an external state, a Markov boundary, and two different probability functions on either side of that boundary. We can then infer that for the probability function of the internal state to work against the probability function of the external state, which is required for self-organisation, as above, there needs to be a model of the probability function of the external state in the function of the internal state.

Could you give an example of the two being used to refer to the same thing? — Tate

See the papers I cited earlier. Or any papers on inference systems.

See the papers I cited earlier. Or any papers on inference systems. — Isaac

If you recall a specific case of a hidden state being referred to as "observed", could you point that out? As I mentioned, in Markov analysis in general, it's not used that way. I think you've misunderstood something along the way.

If you recall a specific case of a hidden state being referred to as "observed", could you point that out? — Tate

inference on models and their parameters given data) that considers hidden states

First mention, first paper. It is the hidden states which are being inferred by the process in question (in this case 'seeing') ie what we 'see' is the hidden state.

They are often referred to as hidden states because they are seldom observed directly.

Second mention, still the first paper. They are seldom observed directly (ie they are observed, indirectly).

Empirical priors (priors that depend on variables) relate to transitions between hidden states, which are encoded in the B matrix

Third paper.

Short of going through every mention in every paper I'm not sure what it is you want.

Why don't you start with a reference informing your opinion that hidden states are not observed.

Yes. An inference regarding a hidden state is never called an "observation." I'm glad to see papers aren't being written with that kind of confusion.

An inference regarding a hidden state is never called an "observation." — Tate

I've just given three examples in which the subject of observation is the hidden state. As I said, why don't we start with the papers from which you've arrived at the impression that what we observe are inferences, and we can see what the differences are.

https://royalsocietypublishing.org/doi/10.1098/rsif.2017.0792

Hidden causes are called hidden because they can only be ‘seen’ indirectly by internal states through the Markov blanket via sensory states. As an example, consider that the most well-known method by which spiders catch prey is via their self-woven, carefully placed and sticky web. Common for web- or niche-constructing spiders is that they are highly vibration sensitive. If we associate vibrations with sensory observations, then it is only in an indirect sense that one can meaningfully say that spiders have ‘access’ to the hidden causes of their sensory world—i.e. to the world of flies and other edible ‘critters’.

So if hidden states are only seen indirectly then what is it that is seen directly? The “internal states”/“sensory states”? What are they?

The example given of spiders seems to suggest what I’ve been saying: that it’s the “sensory world” that is being directly experienced, not the external cause.

I've just given three examples in which the subject of observation is the hidden state. As I said, why don't we start with the papers from which you've arrived at the impression that what we observe are inferences, and we can see what the differences are. — Isaac

This is just garbled.

if hidden states are only seen indirectly then what is it that is seen directly? — Michael

Nothing. 'Seeing' is a process of inference. Nothing is seen directly. Everything that is seen is seen indirectly. It's not a direct process, it has stages.

This is just garbled. — Tate

It's perfectly clear. I'm asking exactly the same of you as you just asked of me. The citations from which you've derived your view. If you don't understand my request then how did yours make sense to you?