Then I think you've either had little experience of the scientific community in my field or you've misunderstood my position. It's quite the most common view among my colleagues and those whose work I generally follow. — Isaac

Since you don't seem to know the difference between Hilton and Davis, I'm not convinced you have a field.

In neuroscience though, nobody thinks people are p-zombies. That's not up for debate.

That color and pain are models? — Marchesk

Yes. That the brain creates models to predict the outcome of the body's interaction with hidden states of the exterior environment. That these models are heavily socially mediated (factors like language and culture).

It is really the standard model in cognitive sciences.

Since you don't seem to know the difference between Hilton and Davis, I'm not convinced you have a field. — frank

Yep, you've called my bluff, I've just been winging it so far using Google.

In neuroscience though, nobody thinks people are p-zombies. That's not up for debate. — frank

Again, for emphasis.

I have a hard time believing that sensations being models is the majority view. What is red a model of? And what do neuroscientists have to say?

In neuroscience though, nobody thinks people are p-zombies. That's not up for debate. — frank


Again, for emphasis. — frank

You'd struggle to find so much as a handful of neuroscience papers which even mention p-zombies in anything more than a disinterested passing phrase, so I'm not sure how you might have formed this view. If you've got any supporting citations I'd be interested in reading them.

So you dont know what a p-zombie is either. Um. Have a nice day. :up:

I have a hard time believing that sensations being models is the majority view. What is red a model of? — Marchesk

The hidden state of some part of the external world.

The hidden state of some part of the external world. — Isaac

Alright, so does cognitive science have a proposal for how this model is generated?

The hidden state of some part of the external world. — Isaac

Here is a thing I've never managed to understand when talking with you about this. Do you agree with these things?

(1) The model's state is informative of the hidden state, but underdetermined by the hidden state.
(2) The model's state is directly causally connected with the hidden state but underdetermined by it.
(3) That underdetermination arises because of priors and task parameters.

When a hypothetical philosophical someone says "I see the apple", they're utilising the causal connection between their perceptual system and the apple. Do you believe they're seeing "apple models" or do you believe they're seeing what the apple models are modelling in the manner they are modelled (roughly, the apple)?

An alternative one could take is that the model puts one in direct access with the hidden state. Not sure how tenable that is, but if you wanted to ground scientific discovery in direct perception, that's a way to do it.

Yes, several. My favourite is Karl Friston's model.

Here's an introduction.

https://www.fil.ion.ucl.ac.uk/~karl/Active%20Inference%20A%20Process%20Theory.pdf

And here's one specifically about visual perception.

https://www.fil.ion.ucl.ac.uk/~karl/Scene%20Construction,Visual%20Foraging%20and%20Active%20Inference.pdf

...but there are others. Anil Seth's take is slightly different, but along the same lines...

https://osf.io/preprints/psyarxiv/zvbkp/

...is quite an interesting recent one.

Do you agree with these things?

(1) The model's state is informative of the hidden state, but underdetermined by the hidden state. — fdrake

Yes, definitely, but the extent will depend on the novelty of the environment and the developmental stage of the brain. It's one of the reasons babies are such interesting subjects.

(2) The model's state is directly causally connected with the hidden state but underdetermined by it. Underdetermined because there're priors and task parameters. — fdrake

Yes, but less so. I don't necessarily see any reason why a model might no become disconnected from the causal state which at one time formed it, I don't think there's anything neurologically preventing that.

When a hypothetical philosophical someone says "I see the apple", they're utilising the causal connection between their perceptual system and the apple. Do you believe they're seeing "apple models" or do you believe they're seeing what the apple models are modelling in the manner they are modelled (roughly, the apple)? — fdrake

The apple. There's nothing more that an apple is than the publicly agreed model. It's not that there's no apple, it's that that's what 'seeing an apple' is.

Edit - I should add that it depends on the context. In terms of challenging any objective authority to those public models I might be more tempted to say 'there's no apple'. It's a contextual language game thing about what we're using 'apple and 'model of apple' to do in some particular conversation.

↪fdrake An alternative one could take is that the model puts one in direct access with the hidden state. Not sure how tenable that is, but if you wanted to ground scientific discovery in direct perception, that's a way to do it. — Marchesk

I think directness is ultimately a question of whether there is a direct causal+informational relation between the hidden environmental states and the process of perception, not whether the whole process of perception is direct or indirect when regarding (properties of) the object. For me at least, a perceptual system is direct when there are no intermediaries between some part of it and hidden states.

Though I imagine that is unusual, since direct realists can be construed as believing when someone sees a red apple, the direct realist's perceptual system simply acknowledges that it is indeed a red apple, and there's a neat correspondence between perceptual properties and apple properties.

Yes, but less so. I don't necessarily see any reason why a model might no become disconnected from the causal state which at one time formed it, I don't think there's anything neurologically preventing that. — Isaac

I agree with that too. I think there's some ceteris paribus clause required - in normal circumstances the hidden states are directly causally connected with the perceptual process and the perceptual process is informative of the hidden states' status insofar as they are task relevant.

The apple. There's nothing more that an apple is than the publicly agreed model. It's not that there's no apple, it's that that's what 'seeing an apple is'. — Isaac

Are you throwing the hidden states into the public agreed model there?

Are you throwing the hidden states into the public agreed model there? — fdrake

Yeah, I think you'd have to. We can't escape this and look at it from a position where I'm outside of my modelling, but I don't see that as a problem (I know some people do). I'm ultimately a pragmatist and it seems to work.

For me at least, a perceptual system is direct when there are no intermediaries between some part of it and hidden states. — fdrake

So would what constitutes a 'perceptual system' have parameters other than the edge of their Markov blanket? I mean such that we're not simply making the above true by definition?

Yeah, I think you'd have to. We can't escape this and look at it from a position where I'm outside of my modelling, but I don't see that as a problem (I know some people do). I'm ultimately a pragmatist and it seems to work. — Isaac

So let's make a distinction between environmental and bodily hidden states. If a hidden state occurs as a part of a bodily system, then I'll call it bodily. If it doesn't, I'll call it environmental. Analogy, my current blood sugar level is a bodily hidden state, the position of the bottle on my desk is an environmental hidden state. Doubtlessly the two have feedbacks between them, and sometimes there is ambiguity regarding whether something is a bodily state or an environmental one.

I'm going to call whatever specifies the current overall task the body is engaged in task parameters, I'll throw in whatever task relevant actions are proposed in with task parameters. EG, I'm currently typing, part of that is motor control, part of that is cognitive functioning, where the keys are, what I feel the need to write, those are task parameters.

I'm going to call whatever a person's learned and is bringing to their current state from that learning - language stuff, habits, etc - priors.

And to spell out underdetermination, a system X is underdetermined by a collection of states when and only when that collection of states does not force X to produce a unique output. EG, if x+y=1, there's more than one solution to it. More metaphorically, if a corpus of evidence supports more than one conclusion, it can be said that the corpus underdetermines its conclusions.

Lastly, I'll call whatever bodily systems in their aggregate output our perceptual features (cups, pulses, warmth/cold, position, emotion etc) the process of perceptual feature formation.

Do you think the following are true:

(1) Environmental hidden states underdetermine perceptual feature formation.
(2) Bodily hidden states underdetermine perceptual feature formation.
(3) Task parameters underdetermine perceptual feature formation.
(4) Priors underdetermine perceptual feature formation.

?

For perceptual systems, (1) means something like "you need not see the same thing as someone else given the same environmental stimulus", (2) means something like "you and another person might disagree on whether 37.6 degrees celcius is normal body temperature or very hot", (3) is like "you and another person can agree entirely on the problem to solve and the solution but not do exactly the same thing" and (4) is like "you and another person bring different life histories to an event and so can interpret it differently". The possibility of those differences is underdetermination.

I think if you put the hidden states into the process of perceptual feature formation, it changes part of their causal relationship with perceptual feature formation. I think that's most clear with environmental hidden states. Putting the environmental hidden states into the process of "publicly agreed" perceptual feature formation, making them play an internal causal role, seems to make them causally fully determined by the process rather than informationally constrained and (possibly) causally partially determined by the process.

In active accounts of perception, people take exploratory actions to elicit data from their environment - that can be how your eyes track over a face to produce a stable image, picking something up and adjusting your body to distribute the load -. These exploratory actions elicit data streams nascent in the environment or cause the environment to behave differently and elicit data regarding the change - contrast exploring someone's facial features which are "already there" vs manipulating a heavy object and adjusting to distribute its load.

These exploratory actions interface with the hidden states - the layout of someone's face and its colour in the light conditions, how the heavy object responds to attempts to pick it up due to its distribution of mass, but they don't causally determine the state of the hidden state. EG, someone's face doesn't rearrange itself because it's looked at, the distribution of mass along the heavy object doesn't change when it's lifted. The environmental hidden states have their own developmental trajectory that we are perceptually exploring in an active, task relevant manner. When perception is functioning normally, though, our perceptual features do model the developmental trajectory of our environmental hidden states enough for our purposes. IE, normal functioning perception places informational/statistical constraints on the developmental trajectory of environmental hidden states, but it should not causally determine their developmental trajectory generically. Like exploring someone's face to form perceptual features of it doesn't actually change the layout of their face. That's a case of informational constraint without causal constraints.

Edit: to clarify, imagine a civil engineer's model of how a bridge bears loads is perfect, they will be able to tell exactly how much would be required to break it. If they had it in their computer, and put in inputs to the model that would collapse the bridge, the bridge would collapse. But the real bridge wouldn't collapse, it just would collapse with certainty if it was exposed to the same inputs. The model informationally constrains the development of the bridge given an input load, but it doesn't actually make the real bridge respond to a load. That'd be a situation where the model completely informationally determines the behaviour of the bridge, but has absolutely no causal relationship with the load bearing behaviour on the bridge.

When I pick up the heavy thing, I do determine its trajectory from the ground to some degree, but I don't do the whole thing - it might be unwieldy, I might bend too much, I might've overestimated the weight and pull it too high. But eventually I manage to stabilise the load. In that situation, the process of perceptual feature formation has attuned to the developmental trajectories of the heavy object and reached a fit for purpose relation - it's being held where it is stably, and I feel it being held there. That's a case of informational constraints with causal constraints. But I still don't causally determine gravity or the heavy thing's distribution of mass (hidden states) that play into the overall lifting action.

Edit2: We're in a more mixed situation than the bridge example with active perception. It's more like if the civil engineer realised that the bridge would collapse from peak Christmas traffic that year, an intervention to stop disaster would happen if the engineer told someone. That's more of the situation we're in - the models we make propose courses of action, so our models when accurate both propose worldly interventions given our current representation of the world and represent the world in some way, so they're causally connected to what they concern, but the content of the model doesn't determine how what it models will behave or develop, it places constraints on how it will behave or develop given the degree of accuracy of our model and our intervention.

I think if you throw the hidden states into the "public perception" of things, you lose the possibility of surprise and adapting to it. To be sure, there are public perceptions of hidden states , but those public perceptions don't causally fully determine the hidden states . It might be that it looks just like a Golden Delicious but it's really a Granny Smith.

If you make the environmental hidden states a part of the process of perceptual feature formation, you lose the ability to elicit underdetermined behaviours from them based on models; to be surprised by them at all. Since they may become fully causally, not just possibly informationally and partially causally, determined by the process of perceptual feature formation. How things look in public becomes what they are.

Ah, so you're another one for whom 'why?' apparently means something completely different with regards to consciousness than it does in every other field of inquiry. — Isaac

No it's the same old why. Just this time it's harder to answer. Because we cannot gather data about something private.

What use is just saying that the way things seem to you right now is completely impervious to any evidence to the contrary? — Isaac

That's not what I'm saying. I said that things seem to me a way. That is a fact. You keep saying things like "there is no phenomenological layer" or "you do not see red" but those are false. I do, in fact, have an experience. There is, in fact, a phenomenological layer. Me knowing how my brain works does not remove the phenomenological layer.

None of which demonstrates that it does so directly (ie, that you are experiencing the stimuli and not your culturally-embedded response to the stimuli). — Isaac

What does it even mean to "experience the culturally-embedded response to the stimuli" or to "experience the stimuli". That just sounds like word salad.

I know when the stimuli is removed, the experience is removed. I also know that when my brain is messed up in this particular way, the experience is removed. I therefore conclude that the brain processing of stimuli is causing the experience. Where is the issue with this line of logic?

Yes. That the brain creates models to predict the outcome of the body's interaction with hidden states of the exterior environment. That these models are heavily socially mediated (factors like language and culture). — Isaac

Ok so my experience is largely shaped by my language and culture. First off, no one is disagreeing (at least I'm not). Secondly, how does this undermine the claim that there is a phenomenological layer? It doesn't.

Conscious experience of red cups is what was in need of explanation... not an explanation of an explanation. — creativesoul

What? Idk what you're trying to say here.

I apologise for questioning your profession. I hadn't actually read many if your posts. But I have the same question:

Yes. That the brain creates models to predict the outcome of the body's interaction with hidden states of the exterior environment. That these models are heavily socially mediated (factors like language and culture).
— Isaac

Ok so my experience is largely shaped by my language and culture. First off, no one is disagreeing (at least I'm not). Secondly, how does this undermine the claim that there is a phenomenological layer? It doesn't. — khaled

I have long believed that culture and language influence experience. For instance, there's nothing in the visual information I get from standing in front of a tree that tells me it's a tree. It's all just shapes and colors. The tree is an idea.

So yeah, I see ideas. I think we all do. This doesn't conflict with the idea of qualia, though.

If your view does, how so?

So would what constitutes a 'perceptual system' have parameters other than the edge of their Markov blanket? I mean such that we're not simply making the above true by definition? — Isaac

This might be wrongheaded, but I think the perceptual system would not be direct if the process of perceptual feature formation didn't have direct causal contact to some hidden states. Isn't that the Cartesian theatre metaphor? We see "models" or perceive "aspects of aggregated sense data", rather than perception being a modelling relation. In those formulations, the models or the sense data are in direct causal contact with the environment, and all perception is of those things which are in direct causal contact with the environment. Two steps removed at all times (Cartesian Theatre) vs One step removed at some times (direct realism).

Okay yeah, but it's not an experience of a world outside the body, so ...

One could say the brain is generating a very immersive (but weird) VR-like experience when dreaming. — Marchesk

Yes, it's as though real life experience is written in neural code, blended all together in sometimes very weird ways and then relived.

And your position is that qualia exist (are a coherent ontological commitment), so saying their existence is 'obvious' is exactly the same as saying that your position is obvious. It's no different to arguing that 'Elan Vitale' is obvious, or that 'Aether' is obvious. — Isaac

The way the world seems is not a theory, is it? To try and put it another way, the biological machinery produces some end-product of consciousness, and that end-product is not theoretical, is it? It's a real end-product.

It doesn't matter what weird expression you use, they all end up empty. "What it's like...", "the way it seems...", "how it feels"...none of these expressions have any coherent meaning beyond behaviours and interoception of physiological states. There's nothing they describe that the aforementioned don't. — Isaac

With your last sentence it sounds like you accept that we have qualia but that you want to provide a physical explanation for them. I'm fine with that. It's your rejection of qualia (the end-product) that I don't understand.

There's no need to introduce an artificial "phenomenal layer" to account for that difference.
— Andrew M

I keep hearing this argument by all the Quiners here. I want to instead ask, what's the problem with introducing that layer anyways, even if we don't need to (not that I'm convinced of that)? What are y'all afraid might happen? What confusion have you been trying to avoid? — khaled

We're trying to avoid Cartesian dualism. That's the position of positing a container mind (the Cartesian theater), and then redefining ordinary words in terms of that container mind. For example, in everyday life watching a sunset, or kicking a football around with your kids, are experiences. Whereas for the dualist, an experience is instead the sense of redness, or the feeling of pain, confined to the mind and intrinsically private to a person. The connection between this "internal" experience and the "external" world is consequently mysterious.

I discussed this previously here. Cartesian dualism has no practical application in everyday life or in scientific inquiry. Concepts like qualia, p-zombies and the hard problem are purely philosophical inventions that derive from Cartesian dualism.

Understanding the difference between dualism and non-dualism is like understanding the difference between geocentrism and heliocentrism. People can look at the same world, but conceptualize it very differently.

Concepts like qualia, p-zombies and the hard problem are purely philosophical inventions that derive from Cartesian dualism. — Andrew M

Is non-reductive physicalism a form if Cartesian dualism?

Conscious experience of red cups is what was in need of explanation... not an explanation of an explanation.
— creativesoul

What? Idk what you're trying to say here. — khaled

You implied that it was necessary(for some reason unbeknownst to me) that my position explain phenomenology. I mean, I certainly can, but a better explanation of conscious experience does not require explaining a different explanation of conscious experience. Weird thing is that I believe I've given you more than enough to piece together as an explanation of phenomenology as well as conscious experience...

Phenomenology is a philosophical position that aims to explain conscious experience. It is an explanation. I find it overcomplicated. There's no need to use the framework to explain anything. Everything about conscious experience can be better explained without using a phenomenological framework. This has been done throughout this thread by those arguing against "qualia".

By the way, you're committing an equivocation fallacy with the word qualia. That's a nice way to say that your position has led to incoherency and/or self-contradiction. Banno has also noted the continual changes in your position.

The connection between this "internal" experience and the "external" world is consequently mysterious. — Andrew M

Loosely speaking, 'the connection' is the experience, on my view.

It consists of both internal and external, physical and non physical, subjective and objective. The problem I seem to see is that both sides miss this. Experience is neither objective, nor subjective; neither internal nor external; neither physical nor non physical...

It is both.

or at lest spooky emergentism. I recently listened to a podcast where a physicist explained why she thought information strongly emerged. But it was fundamental to understanding life:

...spooky emergentism. — Marchesk

Why do you say that emergentism is 'spooky'?

Do you expect conscious experience to just pop into existence ex nihilo style? Does it still seem magical to you? It's not the least bit mysterious, magical, or spooky to me.

Do you think the following are true:

(1) Environmental hidden states underdetermine perceptual feature formation.
(2) Bodily hidden states underdetermine perceptual feature formation.
(3) Task parameters underdetermine perceptual feature formation.
(4) Priors underdetermine perceptual feature formation.

? — fdrake

Yes, but again with caveats I'm afraid. I presume you're talking about mutually exclusive variables to an extent (again with ceretis paribus). In normal circumstances all four would collectively determine - ie there's no other factor - I want to leave aside the thorny issue of whether there might be some random factor for the moment as I don't think it's relevant (my gut feeling is that there might be at least a psuedo-random one resulting from the chaos effect of such a complex system).

If you make the environmental hidden states a part of the process of perceptual feature formation, you lose the ability to elicit underdetermined behaviours from them based on models; to be surprised by them at all. Since they may become fully causally, not just possibly informationally and partially causally, determined by the process of perceptual feature formation. How things look in public becomes what they are. — fdrake

I think if you put the hidden states into the process of perceptual feature formation, it changes part of their causal relationship with perceptual feature formation. — fdrake

If I've understood you right I completely agree. Of course, I think the case of the engineer's timely intervention is the most interesting part, but I totally accept that we have to have a foundation of the independence of hidden states in order to even make that interesting.

We may have got crossed wires. What I mean by saying that the thing modelled is 'the apple' which is a public model, is not intended as an entanglement of some hidden state with the public model. It's a limit of language (which is what I was trying to get at in my edit). The process of 'seeing' could be seen as essentially that of fitting sensory data to priors (filtering of priors being task dependant). So the meaning of 'I see an apple', might be 'the sensory input best fits the public model of 'apple'', but this is not that same as saying that we see 'model-of-apple', because that would be to make that Cartesian divide of 'seeing' into object>qualia>perception(of qualia). It's just that that's what 'seeing' is, so it's only correct to say we 'see the apple'.

If we wanted to phrase all this in terms of purely Markov Chains in the process of perception, then I don't think we can say any more than that the cause of of our perceptual feature has no name. We do not name hidden states, we only name objects of perception.

Edit - Another way of putting this (the language gets complicated) might be to say that we do name the hidden state (apple), but that these christenings then produce fuzziness on the hidden states we could possibly refer to in any given instance of perception - so the hidden state that is in direct causal relationship with our perceptual system will be only fuzzily identified by any word we apply. I'm not sure which approach is best (if any), I don't think we've really got the linguistic tools we need to develop theories about objects of perception.

I think the perceptual system would not be direct if the process of perceptual feature formation didn't have direct causal contact to some hidden states. Isn't that the Cartesian theatre metaphor? We see "models" or perceive "aspects of aggregated sense data", rather than perception being a modelling relation. In those formulations, the models or the sense data are in direct causal contact with the environment, and all perception is of those things which are in direct causal contact with the environment. Two steps removed at all times (Cartesian Theatre) vs One step removed at some times (direct realism). — fdrake

Yep, I think that actually a good way of putting it. I've described myself as an indirect realist before, but these are not terms I have a in-depth knowledge of, so I'm not attached to them. My question was really just getting at the issue of how we define the boundaries of a 'perceptual system'. Where does the perceptual system end and some other system take over (even if only in theory to show that it never does)? If we just say that the boundaries of the perceptual system are the edge of the Markov Blanket, then your version of direct realism is true, but only by definition (ie if some other process intervened between the hidden state and the perceptual system it would, by definition, either be a hidden state itself or part of the perceptual system). So to get a Cartesian Theatre problem (in order to disprove it empirically rather than definition-ally) we'd have to say that the creation of 'the play' out of some hidden states was not part of the perceptual system - the perceptual system was the bit watching the play. If we say the play-making mechanisms are part of the perceptual system then the system is in direct causal relationship with the hidden states (it's just that the description of the perceptual system is wrong). I don't see anything wrong here at all, I only wanted to clarify which way you were looking at it.