I think you’ll find writers like Clark and Barrett declaring much closer allegiance to the pragmatists than to Skinner and Watson. It only took the field 80 years to catch up .The problem wasnt empirical validation , it was making the conceptual shift. — Joshs

I agree. I used to take a behaviourist approach way back at the beginning of my academic career (methodological, not Skinner), and over the course of it I've shifted to a more cognitive approach. That there are such conceptual shifts can be used as justification for any potential one though.

In that light, I have another link for you, — Joshs

Thanks. I'll have a read (with hopefully an open mind!), but your quote does sound like more of the same vague philosophical readings as in the previous piece. I don't object to these views at all (I'm on a philosophy forum after all), but they are only frameworks, not models. They don't have predictive power and their utility is not universal, it's personal. If you personally find that framework useful, then that's great, it does sound like it's not inconsistent with the empirical evidence (at least not universally so), but I don't think we should confuse the framework for the matter being framed.

I’m not going to let you get away with this ( he said half in jest). As a good empiricist you should know better than to pronounce a verdict on a theory without first demonstrating that you know what it is saying. — Joshs

Yes, fair enough, but that cuts both ways. Would you like to stand by a claim that your understanding of the neuroscience behind active inference models was better than my understanding of phenomenology?

The point is these are not two equal fields of enquiry. Both have methods and aims about which it is possible to be wrong, but science also has a body of knowledge about which it is possible to be wrong. There's no such body of knowledge in phenomenology.

I notice you didn’t comment on my observations concerning the incommensurability of rival meta-theories concerning agreement on what constitutes empirical evidence. Maybe you could start there. Would you be able , for example, to justify the Kuhnian claim that one scientific theory ( for example, phenomenologically oriented enactivism) can replace a rival one without invalidating -disproving any of that rival theory’s empirical predictions? — Joshs

If you'd read any of the arguments I've had about epistemology on here you'd see that I've already done so (to the best of my ability). I have no problem at all with underdetermination, I regularly make that case. What I dislike, however, is the move (often made) from approach A, with all it's empirical evidence, cannot found it's own premises, cannot demonstrate the validity of it's own frame...therefore approach B. It doesn't follow. None can. Yet we still must choose, and I'll take the one yielding the results.

We do create private language all the time , for instance
when we create new theoretical ideas.
‘ Private’ here is a bit of a mis-nomer though. — Joshs

...as is 'language'.

Right, but this is what you denied earlier, which is why I'm getting confused about your argument. You specifically said that moral language was not just providing the other person with some facts about the world for them to do with what they will, yet if moral language is just as you say above, then all it is doing is exactly that, providing facts (about the speaker's state of mind). So which is it? — Isaac

I'm trying to make clear a differentiation between different facets of the communicative act, what I called in that last post the "packaging" and "content". The most straightforward way of clarifying that differentiation is to consider first purely descriptive assertions. If I tell you "sequoias are a species of tree native to the California Sierras", my aim is not just to get you to believe that I believe that, but to get you to believe the same thing that I believe. If you accept my assertion, for whatever reason -- if you decide to agree with what I said to you -- then what you end up with is not just a belief about what I believe, but a belief about sequoias. Expressing what it is that I believe is merely the "packaging", in which I'm attempting to deliver the descriptive "content" about sequoias. I'm not just trying to talk to you about my mind, in that hypothetical speech-act, but about sequoias. Something descriptive about sequoias, specifically: something about their place in a model meant as a representation of the world. I'm trying to get you to adjust your representation-model to feature sequoias in the same way that mind does.

In parallel to that, if I tell you "a doctor should not kill one healthy patient to harvest his organs to save five dying patients", my aim is not just to get you to believe that I disapprove of that happening, but to get you to disapprove of that happening. If you accept my assertion, for whatever reason -- if you decide to agree with what I said to you -- then what you end up with is not just a belief about what I like or dislike, but a dislike of doctors doing that thing I'm talking about. Expressing my disapproval of it is merely the "packaging", in which I'm attempting to deliver the prescriptive "content" about murderous doctors. I'm not trying to talk to you about my mind, in that hypothetical speech-act -- nor about anybody else's minds, for that matter -- but about doctors murdering healthy patients. Something prescriptive about that, specifically: something about its place in a model meant as a blueprint for the world. I'm trying to get you to adjust your blueprint-model to feature murderous doctors in the same way that mind does.

No we don't. That would require a private language which would be impossible to learn. We have words which refer to public effects of what we take to be 'emotions' which we use to convey our own propensity to those public effect. If there were no mediating public effects we could not possibly learn the words. — Isaac

Packaging and content again. If you tell me you're feeling sad, do you expect that I merely take that as a description of some externally observable behavior you're doing, rather than remembering the way I feel when I describe myself as "feeling sad" and imagining that you feel that same way? Sure, we need shared public experiences to learn what the words mean, but once we've really learned the words, we understand them in terms of mental experiences -- unless you perhaps really have no theory of mind and are unable to attribute mental states to others, only to observe their behavior? That would seem a very strange deficit for a psychologist to have, but it could explain a lot, and I'm lead to understand that many psychologists go into the field because of interest in remedying their own psychological issues.

It's your technical definition that I'm trying to work with. It was this which distinguished intent from desire. I desire state X, I intend to do Y to get it. If intend refers to the state we want to be the case, then what's left for 'desire' to do? — Isaac

I don't know how I came across as thinking that intent was about what to do to get a state of affairs. I do differentiate between ends and means, of course, but "desire" and "intent" as I use them aren't about that distinction. They're about first- and second-order prescriptive attitudes toward states of affairs. I elaborated on that in the previous post:

In my scheme, an "intention" and a "belief" are each reflexive or second-order forms of "desires" and "perceptions", respectively. Each of them requires that you have awareness of your first-order states of mind, that you can perceive that you are perceiving and desiring certain things; and then also that you pass judgement on those first-order states of mind, that you desire to perceive and desire in that way or else differently.

So yes, on my account an intention, i.e. a "moral belief", is a second-order a desire, it's a desire that you desire to desire. This is more or less the same as Harry Frankfurt's conception of "will": your will is what you want to want. And yes, you could in turn have desires about your desires about your desires, ad infinitum, the more you thought over your decision-making process. Your intention, or will in Frankfurt's terms, is whatever the top level of that is: whatever you've concluded, after however much thought you've given it, that you want to want to want... etc.

Why to want things, and thus what to want to want, i.e. what to intend, i.e. what "moral beliefs" to hold, is a separate question from just what it is to have a "moral belief" / intention. Just like what to believe generally, descriptively, is a separate question from what it is to have a belief. (My answer in both cases, to the "what to think" questions, which we've been over and over already, could be summarized as "heed your experiences... and everyone else's too".)

(Again, think back like you are raising a child, or programming an AI. How do you want the child or AI to go about making these decisions, either about what is real, or about what is moral? How, generally, do you intend people to make those kinds of decisions -- regardless of how you believe that they in fact do make them? Now look at yourself in the third person, like you are parenting yourself, and ask: are you making those kinds of decisions in the way you want people in general to make them? Would you try to get someone else, who makes decisions the way that you do, to change the way they do that? If so, try to get yourself to change the way you do that, like you would anyone else.) — Pfhorrest

I also gave an example of the differentiation between "perception" and "belief" for analogy even earlier:

This is an important aspect of my differentiation between desire and intention. It's exactly like the analogous differentiation between perception and belief. You can perceive a pond of water in the desert, but because you know about mirages, disbelieve that there is actually a pond of water in the desert -- but that doesn't make you stop perceiving it. It still looks like there's a pond of water there, even though you have judged that perception to be incorrect.

Likewise, to have an intention, on my account, is to have a judgement about your desires, but that won't necessarily force them to change. — Pfhorrest

But if one of those facts about the world is about someone holding a prescriptive attitude toward some state of affairs, and the other fact is about the odds of that person's attitudes towards states of affairs being the correct ones to hold, then what you end up with is the adoption of a prescriptive attitude toward a state of affairs — Pfhorrest

People are not automatons. Those two factors alone would not be sufficient to end up adopting the belief. We have a thousand other factors coming in to play at once. — Isaac

The point of the passage you're responding to is not that those two factors would be sufficient (nor necessary) to cause the adoption of the prescriptive attitude toward the state of affairs, but that in the event that the communicative act is successful, and the listener adopts the view that the speaker is trying to get them to adopt, then the view that the listener ends up adopting is a prescriptive one, not just a descriptive one. Packaging and content again: the packaging may be "speaker holds this view" plus "speaker is reliable", but if the view is a prescriptive one, and the the listener accepts that package and unpacks it, they will end up with a prescriptive view, not merely a descriptive one.

If I tell you "you shouldn't do X" and all you take away from that is the fact that I don't want you to do X, without taking away any intent of your own not to do X, then you haven't accepted or agreed with what I told you. Which might be fine, you don't have to agree to everything everyone tells you, that's not the point here. The point is that if you do agree, you're adopting the same intention I'm expressing. If I told you "you shouldn't do X", and you said "got it, X would be bad", and later on I heard you tell someone else that they shouldn't do X -- and so on, such that it really seems like you have agreed with my moral assertion -- and then later still I find out that you did X, and in surprise I ask "I thought you agreed that you shouldn't do X?", and you say "yeah but I thought you wouldn't find out, sorry you did" ... it'd be clear there was some big misunderstanding there. I said you shouldn't do X, and you agreed with that moral claim, but then you thought it was fine to do X so long as I didn't find out, because you thought the moral claim only meant that I don't like X? That sounds to me like you never actually agreed that you shouldn't do X at all, but only lied about agreeing to placate me.

If I say "murder is wrong" and you say "oh yes of course, murder is horrible!" but then think to yourself "mental note: don't let Forrest find out about any murders I do", that's clear that you don't actually agree with the moral claim, and all you took away from my asserting it is that I feel some way about something. Even saying "I disagree, murder is fine" instead, as morally abhorrent as that would be, at least would show more comprehension of the speech-activity we were even engaging in. Even saying "I don't mind murder" would be clearer communication, because you'd at least be showing that you understand that what we're exchanging here are our prescriptive attitudes about things. OTOH if you said "right I understand, murder is wrong, but I have no objection to it", that would be a confusing response: you "understand" that it's "wrong", yet you have no attitude of disapproval toward it? Are you just saying that you're aware that other people disapprove of it? Cause that's not what we're talking about here. We're talking about whether or not to (dis)approve of it, not whether or not anyone does (dis)approve of it.

The picture you're trying to paint of moral judgement not only is woefully simplistic, but even as a goal it would be throwing away millennia of evolved, finely tuned mental processes in hubristic favour of something you came up with. — Isaac

Just like in my epistemology, I'm not at all suggesting that people should completely abandon their natural processes for deciding, either about what is or what ought to be. I'm not even putting forth a complete and precise mechanistic process for how to do either of those things. I'm putting forth reasons why -- when those natural processes fail us and we find ourselves trying to figure out things that aren't coming easily to us and sort out disagreements between each other about what's "obvious" according to those natural processes -- there are some broad limits on the kinds of processes we should turn to to resolve those quandaries.

Namely, in either case: that we shouldn't disregard the relevance of our experiences; that we should try to regard everyone else's as equally as we can too, and figure out something consistent with all of them; that we should be willing to toss away any suggestion as to what that something might be if it's shown to fail at that; but that we shouldn't demand that any such suggestion prove itself immune to all failure or else be tossed away immediately, but rather let float different such suggestions so long as none of them has failed at that yet.

There is a lot of wiggle room inside those broad limits, I'm not specifying exactly where in there is the best route and I'm not sure even sure there is a best route within there (only that somewhere within there is better than outside those bounds), and I expect most of the time in day-to-day life our natural inclinations will stay well within those limits. But it's when those fail us that the cases become philosophically interesting, when people get tempted to appeal to the supernatural, or to say that there is no right answer so shut up and stop talking about it, or that some answer is unquestionably the right answer, or that since nobody has yet proven beyond all shadow of a doubt that their answer is unquestionably right they're all wrong... that's when people start doing bad philosophy, that they think excuses them hide away from practical ways of working around those failures of our natural inclinations. That kind of bad philosophy is the thing that I find interesting and worth arguing against.

Hoo boy we're far from the OP. Nevermind. Completely merciless jargon use follows.

I particularly like their point that the use of Markov blankets and Bayesian theory in a psychological model is mot in itself problematic , the issue is HOW they are used. — Joshs

I think this point in the linked paper is worth throwing some words at.

, any claim that the sensory-effector system is (must be) any organism's Markov boundary depends on having already defined the knowing self, or agent, as whatever is Markov-bounded by the sensory-effector system. This both makes the argument circular, and introduces a highly problematic notion of the knowing self. It is at least a step in the direction of supposing a homuncular self in the Cartesian theater (Dennett & Kinsbourne 1992), and is weirdly reductive insofar as it supposes the agent to have fewer parts than the organism. Even so, it may be true that to know the state of a brain it is sufficient to know its initial state, internal dynamics, and the states of its sensory and motor systems (although we note that we are, scientifically, extremely far from this possibility, so assuming its truth is a very generous stipulation). But assuming the brain to be the appropriate target worth knowing already places the enquiry within the traditional neo-Kantian cognitivist frame. In contrast, from the EEE perspective—at least one strain of which is influenced by the phenomenological critique of Kant (Kaufer & Chemero 2015)—it might be equally worth knowing about the state of one’s hand, the Markov boundary for which almost certainly includes items outside of the body. It also might be worth knowing the state of the tool one is wielding, which is physically external, but in at least some cases epistemically internal (i.e. phenomenally transparent) to the agent9

I'm gonna try and summarise the argument in my own terms.

So what's the brain being the appropriate target for this Markov-blanketing operation doing? It's placing the brain-body system inside a Markov blanket; and what that means is that the brain-body system has sensory inputs from the environment and the body, and effector outputs - your actions are the outputs, they are proposed relative to the inputs and the current task. That might seem like a theoretically inert operation, but it's very easy to go from this interpretation of the Markov blanketing device to an internalist+representationalist conception of the mind. What makes this suggestive?

Imagine this chain of states:

S->M->R->S->M->R

With latent inputs:
H1->S

And latent outputs
H2<-R


Where S is a vector of sensory states, M is a vector of mediating states and R is a vector of output response states. H1 can be thought of as the hidden states of the environment + body insofar as they impact sensory foraging, H2 can be thought of as the hidden states of the environment insofar as they are causally impacted by the agent's responses R.

The "internalist/cognitivist" interpretation thinks of the M states as constitutive of the agent, so the "Markov boundary" of the M states is the sensory states+the response states, and the equation of the agent with the M states then yields the "sensory veil" hypothesis and vulnerability to skepticism.

How to get around this?

Increasingly, evidence is pointing to the importance of the braingut connection, for instance (Cryan & Dinan 2012). In an additional illustration of the difficulty of drawing neat inner/outer distinctions, it may well be that to know about one’s brain one also needs to know something about one’s gut biome—which is biologically inside, but topologically and epistemically outside. Reflect on such examples for a while and it is difficult to resist the conclusion that epistemic internalism is not the conclusion of an argument based on neutral premises, but is in fact a hidden premise of the starting point.

Instead, while still granting the chain as an adequate description, which states are thought of as the agent are changed from only M states to M states + some S states and some R states, the directness of contact is restored, and the "Markov boundary" of the agent comes to include some H1 and H2 states.

Edit: effectively, that internalism seems to follow from the fact that M states have S and R states as their blanket turns on the proposal that the agent consists only of M states, whereas the model is also consistent with interpreting some S and R states as part of the agent.

The most straightforward way of clarifying that differentiation is to consider first purely descriptive assertions.... in parallel to that... — Pfhorrest

This is a repeated tactic in your thinking and I've not understood it from the outset. Simply saying that X is like Y does not make X like Y, yet this seems to be the substance of your argument. You say "like with perception and reality we can..." I've given probably half a dozen reasons why moral talk (or moral thoughts) are not like perceptions and models of reality, yet this seems to have had absolutely no impact on your use of this strategy. So, is there a thing that is not like perceptions and reality, for you? Can everything be likened to it, just by saying so, do you have any criteria at all for these analogies?

I'll try again.

If I tell you "sequoias are a species of tree native to the California Sierras", my aim is not just to get you to believe that I believe that, but to get you to believe the same thing that I believe. — Pfhorrest

Yes, because we're hard-wired to assume a shared exterior source of our sensations, so our common language uses that when making declarative statements.

if I tell you "a doctor should not kill one healthy patient to harvest his organs to save five dying patients", my aim is not just to get you to believe that I disapprove of that happening, but to get you to disapprove of that happening — Pfhorrest

No, because not being hard wired to expect an external source for our hedonic affects, our language does not make use of that, and we do not make moral declarations with that in mind. Some people might sometimes want you to disapprove, other times they might not care if you disapprove so long as you comply, other times they might simply be informing you and be indifferent to your reaction, other times they might be identifying their own values for group identity.

If you accept my assertion, for whatever reason -- if you decide to agree with what I said to you -- then what you end up with is not just a belief about what I believe, but a belief about sequoias. — Pfhorrest

Yes, because we expect sequoias to be something which exists outside of ourselves, to have consistent properties which do not vary depending on who is looking at them.

If you accept my assertion, for whatever reason -- if you decide to agree with what I said to you -- then what you end up with is not just a belief about what I like or dislike, but a dislike of doctors doing that thing I'm talking about. — Pfhorrest

No, because we do not expect our likes and dislikes to be set by values external to ourselves which are invariant between people. Firstly I could 'accept' your assertion in any of the contexts mentioned above, only one of which would result in my disliking doctors doing that, and second, my likes and dislikes, being internally sourced, are not something which is generated by higher thoughts like understanding what my friend has said about their view. Unlike perceptive models which areinfluenced in that way because we expect shared external sources of them.

I'm not just trying to talk to you about my mind, in that hypothetical speech-act, but about sequoias. — Pfhorrest

Yes, because we all believe that sequoias are external-to-us objects with properties that do not vary depending on our physiological state.

'm not trying to talk to you about my mind, in that hypothetical speech-act -- nor about anybody else's minds, for that matter -- but about doctors murdering healthy patients. — Pfhorrest

No, that would be to assume the rightness or wrongness was a property of the doctors not of the mind perceiving them and we do not assume such a thing. So you can't talk about the 'badness' of an event without including in that the state of mind {feeling bad about it}. You can talk about the 'tallness' of sequoias without including the state of mind {thinking that sequoias are tall} because we expect tallness to be an invariant property of sequoias in a way that we do not expect 'badness' to be an invariant property of doctors killing people

You can't just claim things can be treated the same without addressing the ways in which they are different and showing those ways to be irrelevant.

If you tell me you're feeling sad, do you expect that I merely take that as a description of some externally observable behavior you're doing, rather than remembering the way I feel when I describe myself as "feeling sad" and imagining that you feel that same way? — Pfhorrest

Sometimes, yes. You overestimate the narrative of what you think you're doing. But that wasn't the point. The point was that we cannot do this with thoughts that have no external behavioural reference because it would make them impossible to learn. What's the external behavioural referent which distinguishes thinking X is morally wrong and not wanting anyone to do X?

yes, you could in turn have desires about your desires about your desires, ad infinitum, the more you thought over your decision-making process. — Pfhorrest

If so, then how? Presumably you agree that all of this takes place in a physical brain, so if you want to assert that this is possible, you'll need to posit a mechanism.

You can perceive a pond of water in the desert, but because you know about mirages, disbelieve that there is actually a pond of water in the desert -- but that doesn't make you stop perceiving it. — Pfhorrest

It really does.

If I say "murder is wrong" and you say "oh yes of course, murder is horrible!" but then think to yourself "mental note: don't let Forrest find out about any murders I do", that's clear that you don't actually agree with the moral claim, — Pfhorrest

Again, you're assuming your conclusion in your argument. You've not established that this is what moral claims mean, so how can you say that I wouldn't have agreed with it on those grounds? If the moral claim "murder is wrong" is to mean "murder is something which people will generally punish you for" then me thinking "don't let Forrest find out about any murders I do" is exactly me having agreed with it.

Just like in my epistemology — Pfhorrest

I get it, but just like in your epistemology, you're underestimating the degree of underdetermination your method yields - namely that none of the views you'd want to eliminate will actually be eliminated by it.

...as is 'language' — Isaac

Why, because communicating with oneself is so profoundly different from communicating with others in terms of its goals? Only if you start from a separation of organism and world. Then talking to oneself
is idealist solipsism.

Would you like to stand by a claim that your understanding of the neuroscience behind active inference models was better than my understanding of phenomenology? — Isaac

Yes, absolutely! But I realize I’d have to prove that to
you. I’ve been reading more Barrett and Clark, and listening to her youtube lectures. I’m getting a pretty good sense of where she stands. Keep in mind , my graduate degree was in experimental a cognitive science. Even though that was the 1990’s there is an awful lot that is familiar to me in her model. Connectionism and parallel distributed processing had already come into vogue , and I had been familiar with the James-Lange theory of emotion ( we interpret physiological effects to determine causes in the world) , and with Schacter and Singer’s studies on how we attempt to interpret arousal aa emotions of one sort or another( or as something other than emotion) based on situation.Eleanor Roesch’s work on conceptual category formation I think also bears similarities with Barrett’s modeling of on-the -fly conceptualization. I was also familiar with the work of Neisser, Gibson, Mclelland and Rumelhart.

I don't object to these views at all (I'm on a philosophy forum after all), but they are only frameworks, not models. They don't have predictive power and their utility is not universal, it's personal. — Isaac

My expertise in psychological theorizing is concentrated in clinical psychology, psychotherapy and personality theory. Since Barrett ventures into this territory from
time to time , we could use this as a source of comparison with cbt and other approaches to psychotherapy. Of course the sort of evidence that must be accepted in this area is different from that which neuroscientific models make use of, but it is nonetheless does have predictive power ( it must since it is results oriented rather than just abstract theory).

What I dislike, however, is the move (often made) from approach A, with all it's empirical evidence, cannot found it's own premises, cannot demonstrate the validity of it's own frame...therefore approach B. It doesn't follow. None can. Yet we still must choose, and I'll take the one yielding the results. — Isaac

And yet you said that you abandoned S-R theory for cognitivism not on the basis of the empirical evidence but on the basis of a conceptual shift. Would it be fair to say that results matter because they speak to the ability of an approach to define what it stands for clearly , comprehensively and in a coherent manner ? In other words, if a pretender to the throne of new psychological paradigm impressed you on these terms, then you could embrace it even if it hasn’t yet been translated into a thriving research community? My concern is that William James’ work was likely ignored by many who used the rationale that it hadn’t produced clear predictions or a body of empirical results backing it up. It is often difficult and frustrating to make sense of ground-breaking new ideas in psychology or philosophy, precisely because they have to introduce a mew vocabulary , and because the empirical research stage has to wait for a community to form around the ideas.

To be fair , I think that Thompson, Varela, Ratcliffe and Chemero aren’t as far removed from Clark and Barrett as they would like to think, and that maybe any differences in empirical findings from them will be subtle and their significance may only be appreciated by linking it back to the phenomenology( kind of like a work of modern art which is indecipherable without the text on the wall next to it ). ( I’ve written papers critiquing Thompson, Varela and Ratcliffe’s reading of phenomenological authors).

Nevertheless , I think Ratcliffe makes an interesting and somewhat significant distinction between accounts of emotion that make it an interpretive construction oriented to bodily states , and an interpretive construction that takes into account these bodily states but is primarily oriented toward the world. It is the glass which feels cold, not my bodily sensation of it.

https://core.ac.uk/download/pdf/228165528.pdf

Yes, absolutely! But I realize I’d have to prove that to you — Joshs

Well I admire your pluck, if nothing else! But now it's on me to respond, so were I to have the courage (I don't) how might I rise to the equivalent challenge? How could you test my understanding of phenomenology in the way I could test your understanding of cognitive psychology? The people you cite as 'misunderstanding' phenomenology have read the relevant books, to no less a degree than you've read Barrett and Seth.

That their conclusions about aim and methods differ from your can't be held as a measure of understanding surely? Otherwise I could use the same measure to claim you don't understand Barrett ans Seth.

That their conclusions differ from mainstream interpretations cannot be used either, otherwise, again, I could use the same measure to claim you don't understand Barrett and Seth?

Maybe you could quiz me on what was actually said? But I could pass that no less than you could with a Google search - again we're no closer to understanding here.

The problem is, we're talking about apples and oranges. In the case of Barrett and Seth they're presenting experimental results and postulating the implications, then trying to find a model which fits the result, make predictions from it and test those. With phenomenology people are just postulating the model alone, or often not even the model, but a way of talking about the model. I'm not sure in this latter case there's even any body of information of which it is coherent to use the term 'understand'.

My expertise in psychological theorizing is concentrated in clinical psychology, psychotherapy and personality theory. Since Barrett ventures into this territory from time to time , we could use this as a source of comparison with cbt and other approaches to psychotherapy. Of course the sort of evidence that must be accepted in this area is different from that which neuroscientific models make use of, but it is nonetheless does have predictive power ( it must since it is results oriented rather than just abstract theory). — Joshs

Yes, that's certainly something which would contribute to the discussion, but again, it's apples and oranges. That a therapeutic approach works is a different evidential data-point from a neural response helping us understand why it worked. Otherwise we'd not distinguish between placebos and medicines.

And yet you said that you abandoned S-R theory for cognitivism not on the basis of the empirical evidence but on the basis of a conceptual shift. — Joshs

No, I abandoned methodological behaviourism because I felt it's tenets no longer applied. Methodological behaviourism was simply the admission that all we had available to us were stimuli and responses. It would therefore be unscientific to pretend otherwise. The study of psychology at the time was a study of what stimuli caused what response, not because we considered the 'black box' to be a simple switch, but because we had no accurate way to look inside it and so any speculation as to it's workings was unscientific. I gradually switched approach as the data coming out of neuroscience made that less and less the case. The more accurate data we have about the 'black box' the more we can make good models of computational cognition. I still think methodological behaviourism is appropriate for many areas of investigation, particularly with very young children.

if a pretender to the throne of new psychological paradigm impressed you on these terms, then you could embrace it even if it hasn’t yet been translated into a thriving research community? — Joshs

Yes. I think that whether one embraces a particular paradigm is a very personal and intuitive thing. It might be a coherency, it might be some quality of it's advocates, it might be because you fancy the lead researcher... I really don't think it matters. What I'm objecting to here is the idea that some paradigm which is entirely consistent with the evidence, can be called 'wrong' simply by pointing out that it has paradigmatic assumptions and then loosely hand-waiving and one or two of those and calling it a critique. considering the depth of, say, Barrett's modelling detail and the extent of physiological corroboration, I found the attempt to undermine with an anecdote about a Doberman pretty insulting really. It's something I find all too often where cognitive psychology and philosophy meet. Detailed intricate and complex models are dismissed with the most trite "well it doesn't feel like that to me"...as if it would!

Let's see some saccade studies, some time series EEG, some micro-electrode investigations...anything serious and detailed showing the way in which the outside world forms part of the Markov blanket for our mental models. Then there'll maybe be some cause to rethink the paradigm. The gut biome idea, for example is a good potential route, but it would need far more than the speculative connection given thus far. We'd need to see a clear single path from biochemical marker to specific neural signal, otherwise the biochemical marker is simply being modelled, not integrated and we're back to it's actual state being hidden.

I think Ratcliffe makes an interesting and somewhat significant distinction between accounts of emotion that make it an interpretive construction oriented to bodily states , and an interpretive construction that takes into account these bodily states but is primarily oriented toward the world. It is the glass which feels cold, not my bodily sensation of it. — Joshs

I think this misses the point of the place in which Seth, Friston and Barrett see their model. I'm not sure I've read it right, but I think this is the point has made above. That there is a risk, even a tendency, to associate the Markov blanket with the limits of agency has no bearing on either the utility or the accuracy of the model. what we talk about and the reality thereby created ("The glass feels cold") need not be reflected one-to-one in a model of how such talk comes about in the machinery we assume constructs it.

The way we feel our thought processes to take place and be oriented toward, is a result of the actual processes, not a report of them.

This is a repeated tactic in your thinking and I've not understood it from the outset. Simply saying that X is like Y does not make X like Y, yet this seems to be the substance of your argument. You say "like with perception and reality we can..." I've given probably half a dozen reasons why moral talk (or moral thoughts) are not like perceptions and models of reality, yet this seems to have had absolutely no impact on your use of this strategy. So, is there a thing that is not like perceptions and reality, for you? Can everything be likened to it, just by saying so, do you have any criteria at all for these analogies? — Isaac

The analogies are meant to be for illustration of what I'm trying to say, not for persuasion (except inasmuch as clear understanding is necessary for persuasion). I'm not saying "this is analogous to that because I said so". I’m just starting with very general principles that make no reference whatsoever to the direction of fit of the kinds of opinions or assertions they are applied to. Then I apply them equally to opinion or assertions with the two opposite directions of fit, and what do you know, out of that emerges familiar positions on philosophical topics both about what is real and about what is moral. So yeah everything can be linked to something with the other direction of fit: just take the same principles, that are fit-agnostic, and apply them to something with the opposite direction of fit.

I note that something like this at least subconscious seems to have been going on in other philosophers who never explicitly (to my knowledge) called out the parallels. Mill’s ontology is the same as mine, and his ethics, though focused too much on ends alone, agrees with mine on what the correct ends are. I get to both of those conclusions via the same principles of universalism and phenomenalism, applied in opposite directions of fit. Kant’s epistemology is quite similar to mine, and his ethics, thoughts focused too much on means alone, approaches means in a similar way to mine. I get to both of those conclusions via the same principles of liberalism and criticism, applied in opposite directions of fit. Etc.

Yes, because we're hard-wired to assume a shared exterior source of our sensations, so our common language uses that when making declarative statements. — Isaac

No, because not being hard wired to expect an external source for our hedonic affects, our language does not make use of that, and we do not make moral declarations with that in mind. Some people might sometimes want you to disapprove, other times they might not care if you disapprove so long as you comply, other times they might simply be informing you and be indifferent to your reaction, other times they might be identifying their own values for group identity. — Isaac

First of all, I’m not talking about empiricism or hedonism at all here yet. We've stepped back to the topic of what we’re trying to do when we tell someone something, not the (chronologically earlier, logically later) topic of how to decide whether to agree with something we’re told. Empiricism and hedonism are only part of my answer to the latter question, not the former. For all I'm concerned about this topic of language alone, the descriptive claims could be about supernatural things being real, and the prescriptive claims could be about ritual purity being morally obligatory, even though my principle of phenomenalism would say to reject both of those claims. They are still the kind of claims, about something being real or something being moral, that we're talking about here, even if I think they're categorically incorrect claims.

Secondly, I phrased those examples in the first person there for a reason. Other people might mean different things by the same words I would use than I would mean, but I’m telling you in those examples what I would mean if I said them, and what state of mind in you would constitute, from my perspective, agreement with what I had said. It’s of course totally possible that people could use words differently than this way, which is why I’m not making any claims about what particular people mean by particular words, but about a kind of speech-act that can be performed by such words or perhaps others.

I am a native English speaker though, and except in philosophical contexts where people seem to go out of their way to interpret things in line with their ideologies, I never have trouble reaching understanding with people by using these words in this way, so this is clearly a way that they can be used, and this is the kind of speech-act I am discussing here, not some other speech-act someone might try to perform with similar words.

(I know an anecdote is not data, but I polled my gf, who is not otherwise privy to this conversation, about what she would think if someone said that they thought something was morally wrong but that they aimed to do it anyway, and she said that would sound weird, that such a person seems like a sociopath who doesn't understand what it means to think something is wrong, and only understands avoiding retribution from others.)

Again, you're assuming your conclusion in your argument. You've not established that this is what moral claims mean, so how can you say that I wouldn't have agreed with it on those grounds? If the moral claim "murder is wrong" is to mean "murder is something which people will generally punish you for" then me thinking "don't let Forrest find out about any murders I do" is exactly me having agreed with it. — Isaac

See above about the first person.

you can't talk about the 'badness' of an event without including in that the state of mind {feeling bad about it}. You can talk about the 'tallness' of sequoias without including the state of mind {thinking that sequoias are tall} because we expect tallness to be an invariant property of sequoias in a way that we do not expect 'badness' to be an invariant property of doctors killing people — Isaac

You’re mixing up two different parts of what’s being communicated by the speech-acts here: the state of affairs being talked about, and what’s being said about it (that it is or isn't a true or real state of affairs, or that it is or isn't a good or moral state of affairs). We can talk about sequoia trees being short, for example, without saying that they are short, so saying that they are (or aren’t) is no less expressing a mental attitude than saying that they should be (or shouldn’t be).

This was the point about the example dialogue between Alice and Bob a few posts ago, where Alice initially just said the sentence fragment "people killing each other" and Bob didn't yet know what she was saying about that sort of event: that it happens? or doesn't? or should? or shouldn't? As it went, Alice expressed two different attitudes toward that same state of affairs: that she thinks it happens, and that she thinks it shouldn't.

You can't just claim things can be treated the same without addressing the ways in which they are different and showing those ways to be irrelevant. — Isaac

You can't just claim that the same principles don't apply to different situations without addressing what about the differences between them makes those principles inapplicable to them. I get that that's what you're trying to do, but so far you seem unsuccessful to my judgement, so I remain justified in thinking that they are applicable to both.

Some of the things you're saying would have been successful if I had been saying the weird things you seem to take me to be saying, but I'm not saying those things.

The point was that we cannot do this with thoughts that have no external behavioural reference because it would make them impossible to learn. — Isaac

Sure, and I agreed as much. We learn them through the behaviors people do when using words to report their states of mind, but once we have learned them we can take them to refer to the states of mind themselves.

And NB that there absolutely is a behavioral difference in holding a descriptive or prescriptive opinion, on my account, because descriptivity and prescriptivity are defined almost if not entirely by the role played in our behavior. By watching their behavior we can infer things about how people think the world is, from the things they seem to expect, and how they think the world ought to be, from the things that they seem to strive to make the case. By hearing the words they use, like "is" and "ought", etc, in correlation with the states that bear those relations to their behavior, we can learn that when someone says something "is" the case that means they expect to see the world be that way, and when they say something "ought to be" the case that means they aim to make the world be that way.

For either of those kinds of states, we can ask ourselves whether they're erring somehow in thinking what they think, in either of those two ways; and we can ask ourselves whether or not to think likewise, in either of those two ways. That asking what is or isn't correct to think, in those two different ways.

What's the external behavioural referent which distinguishes thinking X is morally wrong and not wanting anyone to do X? — Isaac

This is really getting at my whole point here, about moral language at least. I think that there isn’t any difference between those at all. (NB this is not the same as me taking X being morally wrong to be the same as anyone disapproving of it; I’m only equating two states of mind. To think X is morally wrong is to not want anyone to do it; for X to actually be morally wrong is for that thought that X is morally wrong to be correct, whatever that turns out to mean.)

So if I think X is morally wrong, I want nobody to do X. If I tell you that it’s morally wrong, and you agree with what I say, such that you now also think that X is morally wrong, that should this mean that you now also don’t want anyone to do X. Otherwise the thing you’d be agreeing to would not be the same thing I was claiming; you’d have misunderstood me.

If I take myself, Forrest, thinking X is morally wrong, to be the same thing as me, Forrest, disapproving of X, and I tell you that X is morally wrong, and you, for whatever reasons, take away from that something that you call yourself, Isaac, also thinking X to be morally wrong, such that you say you agree with me about that proposition as stated, but you take you thinking it’s morally wrong to be the same thing as you thinking that I, Forrest, disapprove of it, then you haven’t actually agreed with me. You haven't adopted the same attitude toward the same state of affairs as I have.

If I disapprove of something, and you agree, that means you also disapprove of it, not just that you’re aware that I disapprove of it.

yes, you could in turn have desires about your desires about your desires, ad infinitum, the more you thought over your decision-making process. — Pfhorrest

If so, then how? Presumably you agree that all of this takes place in a physical brain, so if you want to assert that this is possible, you'll need to posit a mechanism. — Isaac

Me speculating on the physical mechanism is beside the point. It's clear in the first person that the mental phenomenon happens. I routinely have desires, and also want not to have those desires.

Sometimes I have at least a third layer there: for example, I momentarily allow myself to feel uninhibited happiness or excitation, looking forward to (more of) some good thing I want that it looks like I'll have opportunity to get, in a kind of situation where I've been hurt before; then I realize that my emotional guard is down and I try to tamp down on that happiness so I'm not blindsided by the "inevitable" bad thing that's about to happen; then I realize that that secondary response is an unhealthy trauma pattern that I don't want to be doing anymore.

So I find myself wanting to not want to not want whatever thing I was initially inclined to want. I see no reason why that kind of stack of wants about wants would have a hard limit in principle, though of course there are physical limits of one sort or another on mental capacity and I would expect taller stacks to be rarer anyway.

In any case, the only reason I mentioned higher than second-order desires is because that's a common critique of Frankfurt's conception of will as higher-order desire: people ask why is it specifically the second-order desire that's your will, why couldn't you have a third, fourth, etc? I see no problem with the possibility of having more than two orders, and I don't specify precisely the second order of desire as "intention", just the highest of however many orders you happen to have. If you have only the first-order desire, then your intention and desire coincide, by my account. Likewise with perception and belief, which now that I think about it also accounts for you seemingly wanting to treat those as synonyms: in the most common case, of not questioning your perceptions but just running with them naturally, your beliefs and your perceptions do coincide. It's only when you doubt your own perceptions that they become separate.

You can perceive a pond of water in the desert, but because you know about mirages, disbelieve that there is actually a pond of water in the desert -- but that doesn't make you stop perceiving it. — Pfhorrest

It really does. — Isaac

I can only imagine you must be using words in a different way than I am here again, because if you take those words to mean what I take them to mean then you're denying something I've seen with my own eyes. I've seen what appears to be water on a hot road ahead of me that then disappears as I drive over it, and that continues happening over and over as I drive down the road. So of course I'm not constantly surprised that it disappears, because I know it's a mirage, I don't believe that there's really vanishing water all over the road. But it still looks like water, just as much as it did before; disbelieving that it's not water didn't make me stop perceiving something that looks like water.

Or consider for another example, someone's about to show you a deep fake of your favorite celebrity saying something they'd never say, and they tell you ahead that it is a deep fake. Then they play it, and sure enough, it looks like that celebrity saying the thing. But you already know it's a deep fake, or at least you believe it to be so, so you don't believe that that celebrity said that thing, even though you perceive something indistinguishable from that being the case, so far as you can tell.

yeah everything can be linked to something with the other direction of fit: just take the same principles, that are fit-agnostic, and apply them to something with the opposite direction of fit. — Pfhorrest

Why not aesthetics then? What's different about the statement "X is beautiful"? When I mentioned this before, I'm sure you said it was to do with what people expected (ie an actual psychological fact about the world). Now, when I point out actual psychological facts about the world with regards to morality, you say it's nothing to do with such facts and only about what you personally mean in the abstract. So perhaps I've misunderstood your original objection, perhaps you could restate. What would be the problem with making this exact same analogy with "X is beautiful". "X is tall", "X is good", "X is beautiful", "X is the best!"... All could be treated in the same way - unless you're taking into account actual real-world facts about how we deal with aesthetics and personal judgement. But if we're to do that then we must also do so with morality.

First of all, I’m not talking about empiricism or hedonism at all here yet. We've stepped back to the topic of what we’re trying to do when we tell someone something — Pfhorrest

Yep, that's still what I'm talking about too. Our language is dictated by our models of life, so it reflects our psychology.

For all I'm concerned about this topic of language alone, the descriptive claims could be about supernatural things being real, and the prescriptive claims could be about ritual purity being morally obligatory — Pfhorrest

Yep. And the same problem would apply. In the former you're making a claim about an object (some supernatural thing) which we then expect to have invariant properties. In the latter you're giving an opinion which we expect to vary between individuals. If, instead one were to make the second claim something like "The bible says you should not X" then it would be a claim like the former. We don't expect what the bible says to be different for different readers.

I’m telling you in those examples what I would mean if I said them — Pfhorrest

You know the story of Humpty-Dumpty, yes? You've read Wittgenstein? There's no such thing as 'what you mean' by a word. There is only what the word means. It's a public definition, not a private one.

I polled my gf, who is not otherwise privy to this conversation, about what she would think if someone said that they thought something was morally wrong but that they aimed to do it anyway, and she said that would sound weird, that such a person seems like a sociopath who doesn't understand what it means to think something is wrong, and only understands avoiding retribution from others. — Pfhorrest

Ha! Unfortunately my wife's also a psychologist, so I don't think we'd have a fair sample if I asked her!

the state of affairs being talked about, and what’s being said about it (that it is or isn't a true or real state of affairs, or that it is or isn't a good or moral state of affairs). — Pfhorrest

See Ramsey on truth. "P" and "It is true that P" express the same thing (broadly). Saying that "sequoias are tall" is the same as saying "I believe that sequoias are tall". The property 'tallness' is being connected to sequoias by the statement. This is not the case with "Helping the poor is good", although you're still importing the same "It is true that helping the poor is good".
But 'tall' is a property of sequoias. 'Good' is not a property of 'helping the poor'. It's a state of mind, an attitude toward helping the poor, it's a property of your mind, not the event.

In one you're saying a tree has the property of tallness, in the other you're saying your own mind has the reaction of feeling goodness in response to thinking about helping the poor.

In the one the property would continue to be true even if you and everyone else in the world ceased to exist, it is an invariant property: sequoias will always be tall. In the other your mind is the vehicle of the property concerned, not the event that is the subject of the sentence. So unlike the first, where the subject of the sentence is also the vehicle of the property, it would not continue to be the case in the absence of your mind. without your mind to feel it's 'good', helping the poor would have no such property. Without your mind to believe it, sequoias would continue to be tall.

If I take myself, Forrest, thinking X is morally wrong, to be the same thing as me, Forrest, disapproving of X, and I tell you that X is morally wrong, and you, for whatever reasons, take away from that something that you call yourself, Isaac, also thinking X to be morally wrong, such that you say you agree with me about that proposition as stated, but you take you thinking it’s morally wrong to be the same thing as you thinking that I, Forrest, disapprove of it, then you haven’t actually agreed with me. You haven't adopted the same attitude toward the same state of affairs as I have. — Pfhorrest

What about if your statement means that you think of the action in a certain way 'wrong', but as a result of you saying it, I now think of that action in that same way? The property 'wrong' is still a property of our respective minds, not a property of the action. So the linguistic exchange has been about the states of our minds, not the action.

So with "the tree is tall" it's rightness is in the tree. It's right if the tree is, in fact, tall. We might judge that rightness with our flawed senses, we might use widespread agreement to better judge, but the truthmaker is the tree, an external, invariant object - whether it is or is not tall.

With "helping the poor is good" it's rightness is not in 'helping the poor'. It's in the mind of someone thinking about helping the poor. We still might judge that by our flawed senses (listening to them tell us they feel that way), but the truthmaker is not 'helping the poor', it's the mind of the person thinking about it.

The key difference. There's only one tree we all share. One truthmaker, therefore one truth.

There's not only one mind thinking about helping the poor, there's seven billion. Seven billion truthmakers, seven billion truths.

How could you test my understanding of phenomenology in the way I could test your understanding of cognitive psychology? The people you cite as 'misunderstanding' phenomenology have read the relevant books, to no less a degree than you've read Barrett and Seth.

That their conclusions about aim and methods differ from your can't be held as a measure of understanding surely? Otherwise I could use the same measure to claim you don't understand Barrett ans Seth.

That their conclusions differ from mainstream interpretations cannot be used either, otherwise, again, I could use the same measure to claim you don't understand Barrett and Seth?

Maybe you could quiz me on what was actually said? But I could pass that no less than you could with a Google search - again we're no closer to understanding here — Isaac

That’s a tough one.First, I should say that the disagreement I have with writers like Varela
and Thompson over their reading of phenomenology is what I’d call a family disagreement. Husserl was the one who introduced modern phenomenology around 1900, but each writer who came after him and who called their work phenomenological diverged from his approach in one way or another . For instance , many in the community might say Sartre’s version falls short of Husserl whereas Merelau-Ponty and Heidegger go beyond him. Many of those attempting to integrate phenomenology with cognitive science make use of a mixture of different phenomenologists. So I think that even though there are all sorts of internal disagreement about interpretation within that community, they would be in the same page concerning Chemero’s critique of Clark, because I think it’s general enough to capture what is common to all these versions.

I should also mention that it isn’t just phenomenology that would claim to find core philosophical pre-suppositions of Clark and Barrett problematic. A whole range of broadly postmodern schools of philosophy would do the same : hermeneutics , radical constructivism , post structuralism and deconstruction. From the vantage of this larger philosophical community, Barrett and Clark’s philosophical framework places them among the leading Neo-Kantian thinkers of the mid 19th century like Dilthey, Brentano , Helmholtz and Peirce. Not that this should mean anything to you. I’m just pointing out that the stakes are higher here than just a debate between phenomenology and pp.

I get the impression that it seems to you that phenomenology wants to avoid the ‘inner workings’ of the sub-personal domain, the ‘guts’ of the
system where I think you like to begin from, and just focus on apparent surface behavior.

I like doing that too. My B.A. was in cognitive neuroscience. What makes it tricky to see what phenomenology is about is that it begins from a claim that when we begin from even the smallest , most irreducible starting point for our model, for instance the neuron and its interconnections ( we could go further in the direction of ‘smallism’ : yes, smallism is a thing, and begin from the molecular or sub-atomic level, but then we’ve switched to an account which will hide everything useful from a psychological perspective) , we run smack to philosophical pre-suppositions that it is not the job of an empirical science to examine.

For instance, we owe the notion of empirical objectivity to a certain geometrization of the world into mathematical objects that took place between the time of Aristotle and Galileo. Implied in this formation of modern empirical science are assumptions such as the definition of the real world in terms of the calculable behavior of objects in motion. Phenomenology attempts to burrow beneath these assumptions in order to make explicit what is implicit in models like Barrett’s. An important point to make is that phenomenological analysis leaves intact all of the results of pp. It’s job isnt to refute or falsify , but to enrich.

what we talk about and the reality thereby created ("The glass feels cold") need not be reflected one-to-one in a model of how such talk comes about in the machinery we assume constructs it. — Isaac

Except that Ratcliffe’s point depends on understanding the machinery that constructs it in a difference way than does pp.

Pp’s machinery isnt all that mysterious and opaque. Even just from listening to a few of Barrett’s youtube lectures I already had a pretty good idea about what the machinery must be like in order to support her claims concerning behavior and why it runs counter to older views about pre-wired emotion modules in the brain.


I mentioned ‘smallism’ above. I think smallism and localism may apply to pp’s meta--theoretical assumptions:

“ It is our contention that even where smallism and localism are not openly endorsed, it is not the evidence but rather conceptual divergence that is responsible for the continuation of the extended cognition debate.
This is because, given the explicit or merely implicit commitment to smallism and localism (which are not empirical claims), no quantity or quality of evidence could possibly settle the debate. Rather than riding on empirical results and interpretations of those results, the extended cognition debate is therefore a debate about how to define the word “cognition”. For internalists in the debate (Rupert, Adams, Aizawa), cognition is defined in terms of computational manipulations of representations.But computational manipulation of representations is not part of the explanatory toolkit of those who gather evidence on extended cognition. These cognitive scientists work outside the paradigms of smallism and localism. When the debate is framed, as it usually is, as essentially an empirical one, then
critics react as if those working in extended cognitive science are dealing with defective evidence and/or incorrectly interpreting the evidence. But if the extended cognition debate is, as we claim, not empirical but one of definition, then critics might wonder: are advocates of extended cognition simply changing the subject? If the debate really is a disagreement about how to understand the word “cognition”, then the critic might think that, by adopting a different definition to begin with, those who go against mainstream smallist/localist cognitive science are not offering a competing account but are simply doing something else. This is a mistaken view: not only is extended cognitive science not changing the subject, but mainstream smallist/localist cognitive science itself might instead more justly be charged with changing the subject. We will elaborate on this point in the final section.”

https://content.sciendo.com/downloadpdf/journals/slgr/41/1/article-p9.pdf

You asked me how I could test your understanding of phenomenology. Why don’t we start instead with testing your understanding of the philosophical pre-suppositions grounding your own psychological
models . For instance, tell me more about how we can talk about what lies outside of a psychological system in its environment. I can see the influence of Kant on pp, but what is it about the difference between an outside and an inside that makes i it necessary for you to insist that they not be already co-implied in each other. Try to explain this without recourse to markov blankets but unsung a more fundamental language.

Try to explain this without recourse to markov blankets but unsung a more fundamental language. — Joshs

Also @Isaac.

I think that's a good challenge. Do you believe it relates to the point about cognitivism+internalism from the paper I elaborated on here? In which the state collection the Markov blanket is computed for is tacitly identified with the body (fine) and the causal paths which bodily processes mediate are thus construed as internal to the body (not fine simpliciter, but consistent), and the environment is then construed as "outside". I think it's a question of whether you see the coupled system of body and environment as shot through with causal chains that penetrate right through the body and go out the other side; being neither in it nor out it, vs whether you see it strictly in terms of the partition into bodily and environmental (and locale) states. The latter is an internalist construal (models inside the body = content inside the body), the former is an outside without an inside style externalism (undermining the existence of the inside, models inside the body = stages of already environmental relations).

Yes, that sounds right. Body-brain-world as single, reciprocally causal gestalt , with no piece alterable without affecting the configuration as a whole.

For instance , many in the community might say Sartre’s version falls short of Husserl whereas Merelau-Ponty and Heidegger go beyond him. Many of those attempting to integrate phenomenology with cognitive science make use of a mixture of different phenomenologists. So I think that even though there are all sorts of internal disagreement about interpretation within that community, they would be in the same page concerning Chemero’s critique of Clark, because I think it’s general enough to capture what is common to all these versions. — Joshs

Doesn't this fall foul of my first hurdle? We've already dismissed commonality as a measure, Remember Thomspon's 4 citations compared to Seth's 86 did not carry weight as a measure of understanding in psychology, so why would widespread agreement carry weight in any particular interpretation of phenomenology? Do you doubt that, should I trawl the papers, I could dig up an interpretation contrary to that general agreement and cite it, just as you did Thompson?

a claim that when we begin from even the smallest , most irreducible starting point for our model, for instance the neuron and its interconnections ( we could go further in the direction of ‘smallism’ : yes, smallism is a thing, and begin from the molecular or sub-atomic level, but then we’ve switched to an account which will hide everything useful from a psychological perspective — Joshs

Right. But when we drill down into that claim, I'm getting nothing by way of evidence. The claim, as you cite it here, invokes what is 'useful'. How are you judging what is 'useful' - especially "everything useful"?

we run smack to philosophical pre-suppositions that it is not the job of an empirical science to examine — Joshs

This is a common complaint. Again, one I've never really understood. Any scientific model has philosophical pre-suppositions which it does not examine. I get that. A philosophical model could examine them. I get that. But a philosophical model could not examine them without philosophical pre-suppositions. Philosophers are not super-human, they don't get to see things without those pre-suppositions. So how is it helping at all? We replace one set of pre-supposition infused ideas for another. Where does that get us?

For example...

For instance, we owe the notion of empirical objectivity to a certain geometrization of the world into mathematical objects that took place between the time of Aristotle and Galileo. Implied in this formation of modern empirical science are assumptions such as the definition of the real world in terms of the calculable behavior of objects in motion. Phenomenology attempts to burrow beneath these assumptions in order to make explicit what is implicit in models like Barrett’s. — Joshs

...Where the first half of the argument is itself a pre-suppostion on which the second half is based. As is the pre-suppostion that thinking about something can reveal a truth about it (the central pre-suppostion of philosophy).

tell me more about how we can talk about what lies outside of a psychological system in its environment. I can see the influence of Kant on pp, but what is it about the difference between an outside and an inside that makes i it necessary for you to insist that they not be already co-implied in each other. Try to explain this without recourse to markov blankets but unsung a more fundamental language. — Joshs

Well, @fdrake has already done a good job, so I won't repeat that ground. Instead I'll add...

In order to talk about 'a ball' we have to, at the same time as labelling it, know what it's boundaries are. To know it's boundaries is to know that it is 'a ball' because the boundary is the point where 'ball' stops and 'air' begins. The ball interacts with the air, but it is not part of the same thing as the air. If it was, it wouldn't be nameable. It's the same with 'me', you' , 'a person' etc. If we don't have boundaries we can't name the objects. If there's nowhere where 'me' ends and 'everything else' begins, then your statement that "the mind is intentional" has no referent. To make a claim about what 'the mind' is, you have to have a boundary to 'the mind' to know that you're making a claim about it and not about just everything. That boundary is the Markov blanket (damn, I nearly got to the end without mentioning Markov blankets!)

why would widespread agreement carry weight in any particular interpretation of phenomenology? — Isaac

Because the particular widespread agreement I’m talking about is informing the Chemero link I sent you as well as the ‘smallism-localism’ link. The latter paper mentions a thriving, productive extended mind research community oriented around shared conceptual assumptions, and this general interpretation of phenomenology is a core one
of theirs. You’ll notice Gallagher, Chemero and Slaby in the references, all of whom have written extensively on phenomenological authors. These same authors and others I’ve mentioned (Ratcliffe, Thompson, Zahavi) recur frequently in research papers of this community. Meanwhile, few to no citations of Clark or Barrett will appear. So this general interpretation carries weight in the sense that you will have a hard time understanding what this community is about , what conceptual difference they’re referring to with respect to the computationalist representationalist community, without knowing a bit about their shared phenomenological commitment.

But I suspect what you mean by ‘carrying weight’ is:
is it true ( not in some ultimate metaphysical sense)? My answer won’t be very satisfying to you. You simply have to do your best to read Husserl, Merleau-Ponty or Heidegger and decide for yourself. Either it will make sense and produce a gestalt shift in your thinking or it won’t. If it produces that shift , you won’t need an iota of empirical evidence in order to know what is missing from pp models.

Do you doubt that, should I trawl the papers, I could dig up an interpretation contrary to that general agreement and cite it, just as you did Thompson? — Isaac

Yes, but that won’t help you understand what shared conceptual commitment is guiding the non-representationalist, post-computational
extended mind community.

Philosophers are not super-human, they don't get to see things without those pre-suppositions. So how is it helping at all? We replace one set of pre-supposition infused ideas for another. Where does that get us? — Isaac

It is quite helpful when there are two competing research paradigms in psychology and one of them is claiming that the disagreement is a conceptual one rather than a dispute over evidence.

In order to talk about 'a ball' we have to, at the same time as labelling it, know what it's boundaries are. To know it's boundaries is to know that it is 'a ball' because the boundary is the point where 'ball' stops and 'air' begins. T — Isaac

Let me introduce Husserl’s analysis of the perception of a ball. I suggest at first you just treat it as a story and then we can see how it offers a different account of inside and outside and how this relates to Husserl’s notion of intentionality.

I can then point out what aspects of this ‘story’’ have had a profound influence of perceptual researchers like Varela, Thompson , Noe and O’Reagan.

Husserl attempted to chart a course between realism and idealism by grounding all experience in perception and grounding perception in structures of intentionality in which the subjective and objective aspects(what he called the noetic and noematic poles) are inextricably dependent on each other and inseparable. He was very much influenced in his project by the work of Franz Brentano, but went beyond Brentano's notion of inentionality by abandoning Brrentano's naturalism.

One of the key aspects of Husserl's approach was his explanation of the origin of spatial objects. Rather than defining an object in terms of its self-subsistence over time with its properties and attributes, he believed such entities to be , not fictions, but idealities. That is to say, what we , in a naive naturalist attitude, point to as this 'real' table in front of us, is the constantly changing product of a process of progressive constitution in consciousness. The real object is in fact an idealization.This process begins at the most primordial level with what he called primal impressions, which we can imagine as the simplest whiffs of sensation(these he calls actual, rather than real. Actual impressions only appear once in time as what they are. When we see something like a table, all that we actually perceive in front of us is an impoverished, contingent partial sense experience.

We fill in the rest of experience in two ways. Al experience implies a temporal structure of retention, primal impression and protention. Each moment presents us with a new sensation,the retained memory of the just preceding sensation and anticipation of what is to come. We retain the memory of previous experiences with the 'same' object and those memories become fused with the current aspect of it. A the same time, we protend forward, anticipating aspects of the object that are not yet there for us, based on prior experience with it. For example, we only see the front of the table, but anticipate as an empty horizon, its sides, and this empty anticipation joins with the current view and the memory of previous views to form a complex fused totality. Perception constantly is motivated , that is tends toward toward the fulfillment of the experience of the object as integrated singularity, as this same' table'.

Thus , through a process of progressive adumbration of partial views, we constitute what we call an object. It must be added that not just the sense of sight, but all other sense modalities can come into play in constituting the object. And most importantly, there is no experience of an object without kineshthetic sensation of our voluntary movement in relation to the thing seen. Intrinsic to what the object means as object is our knowing how its appearance will change when we move our head in a certain way, or our eyes , or when we touch it. The object is what it is for us in relation to the way we know we can change its appearance relative to our interactions with it.

In sum, what the naive realist calls an external object of perception, Husserl treats as a relative product of constantly changing correlated modes of givenness and adumbrations composed of retentions and protentions. The 'thing' is a tentative , evolving achievement of memory , anticipation and voluntary movement. We never completely achieve the objectivity of the object. But we can’t yet say at this level of constitution that what we have is something objective in an empirical sense. That requires a coordinating of our own experiences of the object and that of other people who have vantages on it that we do not. We then say that our own experience of the object is just an ‘aspect’ of the actual empirical
thing. Through this process of intersubjective
coordination we arrive at by idea of the empirical object , which is something that no one actually sees but instead is an idealization.

From this vantage, attempting to explain this constituting process in psychophysiological terms by reducing it to the language of naive realism is an attempt to explain the constituting on the basis of the constituted. The synthetic structure of temporal constitution is irreducible to 'physical' terms. On the contrary, it is the 'physicai' that rests on a complex constitutive subjective , and intersubjective, process that is ignored in the naive attitude. — Joshs

the particular widespread agreement I’m talking about is informing the Chemero link I sent you as well as the ‘smallism-localism’ link. — Joshs

Ah, then we have crossed-wires a little. I was referring to the slightly more general sense in which you claimed to 'understand' the arguments of Seth and Feldman Barrett (despite holding an unpopular interpretation of them). I'm making the analogy that I could make the same claim about phenomenology (or even a certain brand of it). Simply because I hold the (unpopular) view that it's methods and aims are such-and-such, cannot be held as evidence that I don't understand the approach. You can't cite more popular sources explaining that it's aims and objectives are not such-and such as evidence that I don't understand. Otherwise I can do the same for your unpopular interpretation of Seth and Feldman Barrett. Sort of veering wildly off-topic here, but it's a common theme and so of interest to me. A popular response (particularly in the more 'continental' philosophies) is to claim one's interlocutor doesn't understand. I've never really understood what measure of 'understanding' was being used.

My answer won’t be very satisfying to you. You simply have to do your best to read Husserl, Merleau-Ponty or Heidegger and decide for yourself. Either it will make sense and produce a gestalt shift in your thinking or it won’t. If it produces that shift , you won’t need an iota of empirical evidence in order to know what is missing from pp models. — Joshs

A nice answer. One I have a lot of sympathy with. But the last section seems out of place with the approach. I wouldn't 'know' what was missing would I? Knowledge is not gained by gestalt shifts, only perspective.

Yes, but that won’t help you understand what shared conceptual commitment is guiding the non-representationalist, post-computational
extended mind community. — Joshs

There's that 'understand' again. Grrr!

It is quite helpful when there are two competing research paradigms in psychology and one of them is claiming that the disagreement is a conceptual one rather than a dispute over evidence. — Joshs

OK, How so?

Let me introduce Husserl’s analysis of the perception of a ball. — Joshs

It sounds entirely consistent with active inference accounts of perception. I'm not seeing the difference. You might need to provide me with a little exegesis.

I wouldn't 'know' what was missing would I? Knowledge is not gained by gestalt shifts, only perspective. — Isaac

Not sure I’m following you here. Can we just say that a gestalt shift not only opens one up to a new approach but changes their interpretation of their currently held model?
Now suddenly , in the light of this changed perspective, that current model appears ‘lacking’. I realize there are implications here of a direct comparison between the older and newer model that would have to be justified.

I’m going to take a page from Kelly and say that when I achieve a gestalt shift in my thinking it allows me to ‘subsume’ the older model within the newer one. Kelly used the notion of subsumption in the therapist-client relationship. In order to be helpful, the therapist must subsume the client’s construct system as a variation within the therapist’s system. This doesn’t mean that the therapist must begin thinking like the client, only that they be able , from within their own perspective , to effectively anticipate the client’s ways of construing situations.
That’s how I’m using the word ‘understanding’ with regard to adjudicating between conceptual systems.

As a good postmodernist( or maybe I should say a good Kellyan) , I could claim to ‘understand’ phenomenology (or any other set of philosophical presuppositions) by either subsuming it within a more superordinate model of mine or using it to subsume other perspectives. In either case I would be treating it as a valid and useful body of knowledge that functioned as either a foundation for my preferred perspective or as the cutting edge of thinking for me. If instead I simply argued that it was ‘wrong’, incoherent, nonsensical, irrational or falsified , I would run the risk of being called a modernist or realist. Don’t know if that was helpful.

Let me introduce Husserl’s analysis of the perception of a ball.
— Joshs

It sounds entirely consistent with active inference accounts of perception. I'm not seeing the difference. You might need to provide me with a little exegesis. — Isaac

Here’s my attempt at exegesis. I’m going to use Clark as representative of the pp position. Let me know if that’s not a good idea. Clark declares that he does not believe in a correspondence theory of mind:

“ This perspective leads to a rather profound shift in how we think about mind and cognition- a shift I characterize as the transition from models of representation as mirroring or encoding to models of representation as control (Clark 1995) . The idea here is that the brain should not be seen as primarily a locus of inner descriptions of external states of affairs; rather,
it should be seen as a locus of inner structures that act as operators upon the world via their role in determining actions.”

So Clark maintains the idea of perception as a meeting of the inner with the outer, but replaces passive mirroring with the organism’s active navigation of the world. But I don’t get the impression that for Clark the organism co-constructs and co-defines the very environment that it navigates by virtue of its interactions with that world, except in certain circumstances, and then it is an emergent function. Usually, the organism selects what is salient to it from an environment that exists independently of it. Clark seems to consider cultural creations not so much as manifestations of human-environmental reciprocal determination but as only additional tools (external props) for navigating a world definable independently of the subject’s activities.

“ our behavior is often sculpted and sequenced by a special class of complex external structures: the linguistic and cultural artifacts that structure modern life, including maps, texts, and written plans. Understaading the complex interplay between our on-board and on-line neural resources and these external props and pivots is a major task confronting the sciences of embodied thought.”

How might this differ from Husserl?

Husserl writes:

“Certainly the world that is in being for me, the world about which I have always had ideas and spoken about meaningfully, has meaning and is accepted as valid by me because of my own apperceptive performances because of these experiences that run their course and are combined precisely in those performances—as well as other functions of consciousness, such as thinking. But is it not a piece of foolishness to suppose that world has being because of some performance of mine? Clearly, I must make my formulation more precise. In my Ego there is formed, from out of the proper sources of transcendental passivity and activity, my “representation of the world, ” my “picture of the world, ” whereas outside of me, naturally enough, there is the world itself. But is this really a good way of putting it? Does this talk about outer and inner, if it makes any sense at all, receive its meaning from anywhere else than from my formation and my preservation of meaning? Should I forget that the totality of everything that I can ever think of as in being resides within what is for me real or possible?”

Is Husserl just agreeing with Clark here that the world has meaning for me by virtue of my goal-directed actions in it ? I think the difference is more significant that this.

One way to look at what Husserl is after with his notion of the intentional act is that what the organism encounters in the form of an external ‘stimulus’ belongs to a gestalt that includes the oeganism’s activity.
Let me bring in this quote from Merleau-ponty:

“When Gestalt theory informs us that a figure on a background is the simplest sense-given available to us, we reply that this is not a contingent characteristic of factual perception, which leaves us free, in an ideal analysis, to bring in the notion of impressions. It is the very definition of the phenomenon of perception, that without which a phenomenon cannot be said to be perception at all. The perceptual ‘something' is always in the middle of something else, it always forms part of a ‘field'.” (Phenomenology of Perception, p.4)

That field includes my embodied mind. Any change in my perception , any introduction of a new stimulus is a deformation of the entire embodied perceptual field , a gestalt shift.

In sum , when I perceive , the ‘stimulus’ I perceive doesn’t stand outside of me , over against me , it isn’t ‘matched’ against internally generated
action -oriented representations. Rather, it appears directly as a figure standing out against but defined in its very meaning by its role with respect to that background field. Husserl’s intentional act is not a ‘ ‘combining’ of external and internal . It is the creation of a new dimension of sense composed equally of ‘internal’ and ‘external’ aspects. . Each intentional perceptive act is at the same time an anticipating forward from previous experience ( what Clark would call an internally generated prediction) and an occurring into that anticipation of the perceptual object. The anticipation co-shapes the objective sense of the perception at the same time that the objective ‘external’ aspect of it addresses the anticipation.

What is crucial to note here is that there is no object in itself and no anticipational ‘prediction by itself, no guess that takes place PRIOR TO encounter with an outside (and thus no moment of matching outer with inner) These are not separate moments, structures or processes:the internally generated prediction and the stimulus that enters from the outside. the subjective ‘prediction’ and the objective ‘outside’ are not separate entities or processes. They are the inseparable poles of a single intentional act. I anticipate into what intrudes upon me. That bifurcated act IS the intended object. An intentional act may have as its meaning a perceptual object that is recognizably familiar and identifiable, or one that appears foreign , unlike what came before , unidentifiable in some fashion. This is what Clark would call an error , but doesnt error imply a correct template to compare the erroneous to? For Husserl ,a surprising perception isnt an error in the sense of a failure to correctly match owns predictions to a supposedly independently existing external pattern. It is a positive and substantive new sense, which may or may not lead to a new form of harmoniously anticipated perception.

Husserl agrees with Clark that the organism strives toward harmonious anticipation of events , but this striving isnt oriented around a conformity of internal inferences with external ‘information’. Rather , it is a moving forward on the basis of the substantive meaning of the ‘failed’ perceptual act. If such ‘error ‘ is followed by a new harmony it will not be because néwly generated internal pattens formed a better match with pre-existing external information, but because a new mutually co-detemined subjective-objective gestalt shift produced such a harmoniously anticipatory situation.

Clark offers that there are situations in which the organism and environment exhibit emergent properties of a single non-decoupleable system, but he makes such circumstances secondary to those in which he still
feels that an internalist representationist model is appropriate.

“ Where the inner and the outer exhibit this kind of continuous, mutually modulatory, non- decouplable coevolution, the tools of information - processing decomposition are, I believe, at their weakest. What matters in such cases are the real, temporally rich properties of the ongoing exchange between organism and environment. Such cases are conceptually
very interesting, but they do not constitute a serious challenge to the general role of representation-based understanding in cognitive science. Indeed, they cannot constitute such a challenge, since they lie, by definition, outside the class of cases for which a representational approach is most strongly indicated.”

Can we just say that a gestalt shift not only opens one up to a new approach but changes their interpretation of their currently held model?
Now suddenly , in the light of this changed perspective, that current model appears ‘lacking’. — Joshs

Yes, I think that makes sense, but 'knowledge' seems a much bolder claim than that. A new perspective can indeed show areas where the previous one is deficienct. So now imagine a person whose first experience is with this 'new' perspective and who now comes across what, to us, is the 'old' perspective. Are they not going to have the same experience?

If instead I simply argued that it was ‘wrong’, incoherent, nonsensical, irrational or falsified , I would run the risk of being called a modernist or realist. — Joshs

Applies both ways though, yes? Active inference does account for the Doberman. It just does so differently, from a different perspective. Either we're realists about our models (in which case there's got to be a truthmaker), or we're constructivist, in which case the Doberman example is a category error.

I’m going to use Clark as representative of the pp position. Let me know if that’s not a good idea. — Joshs

Not that it's a bad idea, but I'm much more familiar with Friston, Feldman Barrett, and Seth, so I may not always be faithful to his ideas specifically.

I don’t get the impression that for Clark the organism co-constructs and co-defines the very environment that it navigates by virtue of its interactions with that world, except in certain circumstances — Joshs

OK, so here's the first hurdle. Friston, Feldman Barrett, and Seth definitely do see co-construction as central to active inference. Hidden states are just that. 'States'. Not objects, or scenes, or environments. All of those are constructed as part of the active inference process. I'd need to work more directly from Clark to see if the disparity is your interpretation or his version, but either way, it's not the version I subscribe to.

when I perceive , the ‘stimulus’ I perceive doesn’t stand outside of me , over against me , it isn’t ‘matched’ against internally generated
action -oriented representations. Rather, it appears directly as a figure standing out against but defined in its very meaning by its role with respect to that background field. — Joshs

What active inference accounts of perception are acting on is signals, not objects of perception. So that which is 'matched' is an edge, a lightness, a movement... I don't see how such perceptive elements could possibly be already 'defined in meaning'. They're handled by the V1 and V2 cores, these don't even have direct connections to semantic cores.

there is no object in itself and no anticipational ‘prediction by itself, no guess that takes place PRIOR TO encounter with an outside — Joshs

Again, this concept is core to Friston's active inference.

and thus no moment of matching outer with inner — Joshs

Not seeing how this follows.

In general, the more you explain phenomenological approaches to me, the more impressed I am by the way they presaged active inference approaches. Which, although not your objective, I'm very grateful for. But I'm certainly not seeing any divergence. Quite the opposite.

In general, the more you explain phenomenological approaches to me, the more impressed I am by the way they presaged active inference approaches. Which, although not your objective, I'm very grateful for — Isaac

That’s good. According to my definition of understanding as subsumption I’d have to say that you do understand phenomenology in some form in that you’re seeing it as a legitimate and substantive precursor or foundation for pp models. The question now is whether the more radical interpretations of phenomenology represented by Merleau-Ponty, Varela, JJ. Gibson and Heidegger can enrich pp accounts.
I singled out the above authors because Clark specifically mentions them as anti-representationalist , anti-computationalist approaches that share many features with pp.

I’m going to quote a long passage from Clark’s ‘Putting, Brain, Body and Mind back together again’ because I think you might find his take on these figures interesting:

“ Heidegger (1927) wrote of the importance of Dasein (being there) - a mode of being-in-the-world in which we are not detached, passive observers but active participants - and stressed the way our practical dealings with the world (hammering nails, opening doors, and so on) do not involve detached representings (e.g. of the hammer as a rigid object of a certain weight and shape) so much as functional couplings. We use the hammer to drive in the nail, and it is this kind of skilled practical engagement with the world that is, for Heidegger, at the heart of all thought and intentionality .A key notion in this analysis is the idea of equipment- the stuff that surrounds us and figures in the multiple skilled activities underlying our everyday abilities to cope and succeed.
Thus, Heidegger's work prefigures skepticism concerning what might be termed " action-neutral" kinds of internal representation, and it echoes our emphasis on tool use and on action-oriented couplings between organism and world . Some of Heidegger's concerns, however, are radically different from those of the present treatment. In particular, Heidegger was opposed to the idea that knowledge involves a relation between minds and an independent world (Dreyfus 1991, pp. 48- 51) - a somewhat metaphysical question on which I take no stand. In addition, Heidegger's notion of the milieu of embodied action is thoroughly social. My version of being there is significantly broader and includes all cases in which body and local environment appear as elements in extended problem- solving activity.

Closer in spirit and execution to the present project is the work of the phenomenologist Maurice Merleau-Ponty, who was concerned to depict everyday intelligent activity as the playing out of whole organism- body- world synergies. In particular, Merleau-Ponty stressed the importance of what I have called "continuous reciprocal causation "- viz., the idea that we must go beyond the passive image of the organism perceiving the world and recognize the way our actions may be continuously responsive to worldly events which are at the same time being continuously responsive to our actions. Consider a lovely example, which I think of as "the hamster and tongs"
:
When my hand follows each effort of a struggling animal while holding an instrument for capturing it, it is clear that each of my movements responds to an external stimulation; but it is also clear that these stimulations could not be received without the movements by which I expose my receptors to their influence. . . . The properties of the object and the intentions of the subject are not only intermingled; they also constitute anew whole. (Merleau-Ponty 1942, p. 13)

In this example the motions of my hands are continuously responsive to those of the struggling hamster, but the hamster's struggles are continuously molded and shaped by the motions of my hand. Here action and perception, as David Hilditch (1995) has put it, coalesce as a kind of " free form interactive dance between perceiver and perceived.

" It is this iterated interactive dance that, we saw, is now recognized in recent work concerning the computational foundations of animate vision. Moreover, Merleau-Ponty also stresses the way perception is geared to the control of real-time, real-world behavior. In this respect, he discovers something very like the Gibsonian notion of an affordance- a notion which, in turn, is the direct inspiration of the idea of action-oriented internal representations discussed above in chapter 2 and in section 8.3.

An affordance is an opportunity for use or interaction which some object or state of affairs presents to a certain kind of agent. For example, to a human a chair affords sitting, but to a woodpecker it may afford something quite different. Gibson's special concern was with the way visual perception might be tuned to invariant features presented in the incoming light signal in ways that directly selected classes of possible actions- for example, the way patterns of light might specify a flat plain affording human walking. To the extent that the human perceptual system might become tuned to such affordances, Gibson claimed that there was no need to invoke internal representations as additional entities mediating between perception and action. In section 8.3 I argued that such outright rejection often flows from an unnecessary conflation of two properly distinct notions. One is the fully general idea of internal representations as inner states, structures, or processes whose adaptive role is to carry specific types of information for use by other neural and action-guiding systems. The other is the more
specific idea of internal representations as rich, action-neutral encodings of external states of affairs. Only in the latter, more restricted sense is there any conflict between Gibsonian ideas and the theoretical construct of internal representation.

Finally, the recent discussion of "the embodied mind" offered by Varela et al. (1991) displays three central concerns that likewise occupy center stage in the present treatment.First, Varela et al. are concerned to do justice to the active nature of perception and the way our cognitive organization reflects our physical involvement in the world . Second, they offer some powerful example of emergent behavior in simple systems.Third , there is sustained attention to the notion of reciprocal (or "circular") causation and its negative implications for certain kinds of component- based reductive projects. These themes come together in the development of the idea of cognition as enaction. Enactive cognitive science, as Varela et al. define it, is a study of mind which does not depict cognition as the internal mirroring of an objective external world . Instead, it isolates the repeated sensorimotor interactions between agent and world as the basic locus of scientific and explanatory interest.Varela et al. are thus pursuing a closely related project to our own.

There are, however, some important differences of emphasis and interest. First, Varela et al. use their reflections as evidence against realist and objectivist views of the world . I deliberately avoid this extension, which runs the risk of obscuring the scientific value of an embodied, embedded approach by linking it to the problematic idea than objects are not independent
of mind.My claim, in contrast, is simply that the aspects of real-world structure which biological brains represent will often be tightly geared to specific needs and sensorimotor capacities. The target of much of the present critique is thus not the idea that brains represent aspects of a real independent world, but rather the idea of those representations as action-neutral and hence as requiring extensive additional computational effort to drive intelligent responses. Second, Varela et al. (ibid., p. 9) oppose the idea that “cognition is fundamentally representation." Our approach is much more sympathetic to representationalist and information- processing analyses. It aims to partially reconceptualize ideas about the contents and formats of various inner states and processes, but not to reject the very ideas of internal representation and information- processing themselves.

Finally, our treatment emphasizes a somewhat different body of cognitive scientific research (viz., the investigations of real-world robotics and autonomous- agent theory) and tries to show how the ideas and analyses emerging from this very recent research fit into the larger nexus of psychological, psychophysical, and developmental research which is the common ground of both discussion geared to specific needs and sensorimotor capacities. The target of much of the present critique is thus not the idea that brains represent aspects of a real independent world, but rather the idea of those representations as action-neutral and hence as requiring extensive additional computational effort to drive intelligent responses. Second, Varela et al.
(ibid., p. 9) oppose the idea that cognition is fundamentally representation." Our approach is much more sympathetic to representationalist and
information- processing analyses. It aims to partially reconceptualize ideas about the contents and formats of various inner states and processes, but not to reject the very ideas of internal representation and information-
processing themselves. Finally, our treatment emphasizes a somewhat different
body of cognitive scientific research (viz.,
the investigations of real-world robotics and autonomous-
agent theory) and tries to show how the ideas and analyses emerging from this very recent research fit into the larger nexus of psychological, psychophysical, and developmental research which is the common ground of both discussions.

In general, the more you explain phenomenological approaches to me, the more impressed I am by the way they presaged active inference approaches. Which, although not your objective, I'm very grateful for. But I'm certainly not seeing any divergence. Quite the opposite. — Isaac

:up:

Also @Josh.

That's my take too, active embodied inference resonates very well with phenomenology. The nexus of conflicts between the two approaches; at least from the phenomenological side; seems to be the notion of perception as a representation. Heideggerians don't like representational accounts of perception very much for reasons of the task relativeness of representations, for reasons of alleged over reliance on cognitive categories and lastly for the Heideggerian commitment to holism. I'll write a little on those, though it is a hot take. The three are related.

Task relativeness: representations are typically construed as accurate/less accurate, purposiveness doesn't fit % accuracy, % accuracy instead is evaluable relative to a purpose - an analogy there might be the relationship of a salience map of the face to a face classification task.

Holism: that task relativeness is also a context sensitivity, and finding ourselves in the right context + recognising that we're in the right context for our acts is a meaning imbueing/discovering activity. Representation construed in terms of efficiency and accuracy alone can allegedly create a drought of semantic information; something has to make perceptual features and actions meaningful chunks of body+environment, not just accurate and task fit.

Over reliance on cognitive categories: a representation is theorised as a state vector tuned to be fit for a task, but there's little work on the vector comes to having moving parts - what work is done specialises on what parts of the brain are specialised for what task, not what parts of the active inference machine cotton onto meaningful causal structures. The over reliance on cognitive categories is that we end up feeding enough context into the active inference machine for it to work in a domain; like Friston's saccade experiments facial recognition study; but we don't learn how to evaluate which domain we're in using the same procedure. So we've fed in a cognitively demarcated context without paying attention to the demarcation of contexts, and we end up manipulating representations within a domain rather than tuning a representation generator that varies over domains - like what long term Friston's free energy approach aims to do but hasn't yet (@VagabondSpectre for central pattern generators being another framework).

The question now is whether the more radical interpretations.... — Joshs

As I was saying, the question now is whether the more radical interpretations of phenomenology represented by Merleau-Ponty, Varela, JJ. Gibson and Heidegger can enrich pp accounts. As you saw in my long quote , Clark thinks they unnecessarily close the door on internal
representations. If phenomenological and pp accounts have so much else in common, does this one difference
amount to anything significant? I suppose to some
extent it depends on the aims of the model. Clark likes to build machines , and I think it would be a lot more difficult to simulate psychological processes vi an A.I. system at present without invoking computations and representations. I think if Clark were a personality theorist, psychotherapist, researcher in psychopathology or social psychologist he might look at matters differently.

In this vein , there was something that struck me about Barrett’s youtube lectures on emotion. She decided to spotlight what I consider to be a relatively minor feature of emotion processing as a prime example of how pp differs from older, essentialist approaches to emotion. In her examples , the brain uses active inference to decide whether certain physiological sources of information amount to anxiety as opposed to indigestion , a heart attack or some other physical malady. I understand her aim is to show that deciding that one is experiencing an emotion is the end product of a complex process of prediction testing that takes into account as many sources of information as are available from the person’s interaction with the world as well as their interoceptive states. In enactivist approaches like that of Matthew Ratcliffe and Varela, the emphasis is not on WHAT is taking place when one has the sort of experience Barrett
describes, but on HOW one has it, in the sense of how one is finding oneself in the world, one’s comportment toward events. It is not that they are denying feedback from bodily states needs to be interpreted in order for one to have an emotion. I think it is that the various forms of input into affect , the hormonal , physiological-kinesthetic, behavior and social, are so tightly integrated through reciprocal causality that the question of WHAT one is feeling ( angina vs anxiety) is usually much less pertinent than the issue of how the world as a whole is altered for us when we are anxious or sad or elated. It isn’t that pp doesn’t have the tools necessary to account for mood as global attitude , but I wonder if beginning from computational representation turns integrated holistic comportment into a struggle rather than a given in most situations, something that has to
be wrung out from the data first as a what question and then as a how question. Representational models just seem to me to be clunky when it comes to handling full-fledged ongoing , real-time reciprocal causality.

When Barrett was describing the butterflies one feels when giving a public talk, instead of suggesting it could have been mere indigestion( which of course it could have) , she could have talked about how one’s heart races where one looks up at the crowd , and calms down when one quickly turns back toward the lecture notes , how it races again when looking back up and then calms when one remembers to imagine the audience naked, how one’s reflexes seem to be in overdrive at every noise from the crowd, how one’s legs seem primed to race one’s body out of the room. She could have talked about this constellation of thoughts , feelings, sensations as a coordinated dance, each component implying the next as a meaningful whole rather than a combination of arbitrary elements. Most importantly, she could have talked about the particular ways in which this anxious comportment shapes and orients one’s inclinations to relate o to other people. I recognize that the dance of emotion is composed of differences in equal measure to similarities , but representationalism seems perhaps to result in an emphasis on arbitrary difference at the expense of what makes the components of emotion belong together as a meaningful whole.

Task relativeness: representations are typically construed as accurate/less accurate, purposiveness doesn't fit % accuracy, % accuracy instead is evaluable relative to a purpose - an analogy there might be the relationship of a salience map of the face to a face classification task. — fdrake

I agree so far as object recognition is concerned, but this comes back to the point I made to Joshs about the features of perception being more fundamental that objects. active inference begins to work at things like edge recognition, contrast detection...and I just don't see how those sorts of things could be task oriented. By the time we get up to object recognition, I think salience matters, but active inference authors, certainly Seth, have already concluded the same and speak of multiple models. For Friston as well, the 'surprise' the model is trying to minimise is the signal it is receiving, which, for higher models like object recognition, will be inputs from other cortices, so task salience has already been subsumed in the model by then - one of the hidden states in the object recognition model will be signals from the parts of the brain tracking things like objective. Like if you're thinking "Where have I put my keys, I'm sure I left them on my kitchen table", the expectation that your keys are on the kitchen table will be an input into the system modelling the objects there.

Representation construed in terms of efficiency and accuracy alone can allegedly create a drought of semantic information; something has to make perceptual features and actions meaningful chunks of body+environment, not just accurate and task fit. — fdrake

I don't think this is the case, but I get that I'm straying away from the core of active inference in saying so. I don't think the chunks have to be meaningful. In fact I think they often aren't. I think 'meaning' is a post hoc activity of higher models to try and minimise surprise from the lower models. I don't see any use for it in the act of perceptions. I think it's sue comes in reviewing that act seconds later for efficient recall, or conversion into things like speech acts or object-oriented actions. Obviously the meaning-infused recall will then figure heavily in the next saccade, but only as one of many signals, not as an overarching control.

The over reliance on cognitive categories is that we end up feeding enough context into the active inference machine for it to work in a domain; like Friston's saccade experiments facial recognition study; but we don't learn how to evaluate which domain we're in using the same procedure. So we've fed in a cognitively demarcated context without paying attention to the demarcation of contexts, and we end up manipulating representations within a domain rather than tuning a representation generator that varies over domains - like what long term Friston's free energy approach aims to do but hasn't yet (@VagabondSpectre for central pattern generators being another framework). — fdrake

Yeah, that seems like a valid criticism. Perhaps it reflects the limits of a scientific approach. I can see the problems, but not necessarily the solutions in the lab. It may be time to let us wishy-washy psychologists loose on the subject, something more like Feldman Barrett is doing with emotion?

Besides... give us a questionnaire and a statistics suite and we'll prove anything*!

*(to p=0.05)

I agree so far as object recognition is concerned, but this comes back to the point I made to Joshs about the features of perception being more fundamental that objects. active inference begins to work at things like edge recognition, contrast detection...and I just don't see how those sorts of things could be task oriented. — Isaac

What makes you believe they're not task oriented? Or in other words - what makes the sensible default hypotheses non-task relativeness for edge recognition and contrast detection; or whatever broader category they lay in; when the rest of the procedure is task-relative?

I don't think this is the case, but I get that I'm straying away from the core of active inference in saying so. I don't think the chunks have to be meaningful. In fact I think they often aren't. I think 'meaning' is a post hoc activity of higher models to try and minimise surprise from the lower models. I don't see any use for it in the act of perceptions. I think it's sue comes in reviewing that act seconds later for efficient recall, or conversion into things like speech acts or object-oriented actions. Obviously the meaning-infused recall will then figure heavily in the next saccade, but only as one of many signals, not as an overarching control. — Isaac

So I can agree that it looks like the things we end up saying are meaningful upon reflection are post hoc constructions, but I do think there's another flavour of meaningful information which is embedded in the functionality. The type of semantic information being that perceptual features are foraged under some model of hypothetical cause; the hypothetical causal structure ascribes an explanatory space of meanings/reasons consistent with the act. EG, when someone's perceptually exploring a face, they look at the bits of the face which are most informative regarding its global structure assuming it were a face, you can see the general model of faces at work when looking at someone's scan paths over faces. As for why it maybe counts as semantic, it's like like instructional information.

While the extent to which perceptual feature formation is saturated by meaning "all the way down" is likely to be largely a philosophical dispute; the fact that it is largely a philosophical dispute should give us some pause.

Yeah, that seems like a valid criticism. Perhaps it reflects the limits of a scientific approach. I can see the problems, but not necessarily the solutions in the lab. It may be time to let us wishy-washy psychologists loose on the subject, something more like Feldman Barrett is doing with emotion? — Isaac

I dunno how to evaluate this!

What makes you believe they're not task oriented? Or in other words - what makes the sensible default hypotheses non-task relativeness for edge recognition and contrast detection; or whatever broader category they lay in; when the rest of the procedure is task-relative? — fdrake

Only that tasks (in the sense I think the phenomenologists meant it - 'doing the shopping', eating a sandwich'...) are modelled by areas of the brain several steps removed from the primary visual cortices. They'll have an influence by virtue of several stages of signal suppression, but it will be so watered down by that point that I wouldn't necessarily see it as a pragmatic influence.

I suppose one could take a much broader view of 'task', where a task might be to determine shape, or distinguish background from foreground - in which case I'd agree these can be task oriented, but From the quotes I've been given, that doesn't seem to be what the phenomenologists are on about. You'll know much better than I though, I may have gotten the wrong impression.

The type of semantic information being that perceptual features are foraged under some model of hypothetical cause; the hypothetical causal structure ascribes an explanatory space of meanings/reasons consistent with the act. EG, when someone's perceptually exploring a face, they look at the bits of the face which are most informative regarding its global structure assuming it were a face, you can see the general model of faces at work when looking at someone's scan paths over faces. As for why it maybe counts as semantic, it's like like instructional information. — fdrake

That's really interesting. I'd not thought of it that way, but I can definitely see the link. I suppose one concern I'd have is whether we lose something in lumping meaning as post hoc story-telling and meaning as sub-conscious purpose in the same boat. I think they're radically different in normal use, even though I can see the similarities here.

Yeah, that seems like a valid criticism. Perhaps it reflects the limits of a scientific approach. I can see the problems, but not necessarily the solutions in the lab. It may be time to let us wishy-washy psychologists loose on the subject, something more like Feldman Barrett is doing with emotion? — Isaac


I dunno how to evaluate this! — fdrake

Ha! Self-deprecation comes off sounding weird, I obviously think psychologists are brilliant (or am I being self-deprecatingly honest about my flawed egotism?)...sorry, I'll stop now.

The point was that I think neuroscience can only ever to categorised work. It's so fiendishly complicated that any attempt to draw wider, more holistic conclusions on the basis of it's results alone will fail (currently). Seth's lab at Sussex, for example, is getting more and more into the fine detail of perceptive processing (and now including interception too) and the papers coming out are really cool, but he'll never be able to work that back up again to integrate higher cortices using the methods he does. It's a huge computational task just to deal with the data in that very specific region. He'd need a new method. Maybe psychology, maybe something else.

Realised I'd missed replying to this.

Clark likes to build machines , and I think it would be a lot more difficult to simulate psychological processes vi an A.I. system at present without invoking computations and representations. I think if Clark were a personality theorist, psychotherapist, researcher in psychopathology or social psychologist he might look at matters differently. — Joshs

Yes. I think perhaps Clark is not such a good envoy for active inference. Friston is the Messiah of active inference, although some if his stuff can be a little impenetrable, it's worth persevering with, it'll give you a more solid foundation without what may be some idiosyncrasies of Clark's approach.

enactivist approaches like that of Matthew Ratcliffe and Varela, the emphasis is not on WHAT is taking place when one has the sort of experience Barrett
describes, but on HOW one has it, in the sense of how one is finding oneself in the world, one’s comportment toward events. — Joshs

I'm not sure I follow. 'How' in what sense? (I'm afraid 'finding oneself in the world' hasn't made it any clearer).

think it is that the various forms of input into affect , the hormonal , physiological-kinesthetic, behavior and social, are so tightly integrated through reciprocal causality that the question of WHAT one is feeling ( angina vs anxiety) is usually much less pertinent than the issue of how the world as a whole is altered for us when we are anxious or sad or elated. — Joshs

I think that question is subsumed under subsequent models of inference. Afterall, the world is not altered for us in any one unique way when we're anxious, any more than our physiological states are in any one unique set up. What Barrett is trying to say is that the way the world appears to change is one of the factors involved in the model.

Representational models just seem to me to be clunky when it comes to handling full-fledged ongoing , real-time reciprocal causality. — Joshs

Yes, I grant that. The models themselves would be way too complex, and using them in our conscious thought would be pointless repetition (we're already doing so subconsciously), but understanding that we process things this way as a 'big picture' can be helpful therapeutically, I think.

I suppose one could take a much broader view of 'task', where a task might be to determine shape, or distinguish background from foreground - in which case I'd agree these can be task oriented, but From the quotes I've been given, that doesn't seem to be what the phenomenologists are on about. You'll know much better than I though, I may have gotten the wrong impression. — Isaac

I don't really know how Heidegger would deal with objects showing themselves out of a background mechanically, certainly don't recall anything about it. The mechanics, so to speak, aren't the kind of thing a Heideggerian often wants to talk about. So I don't know if you've gotten the wrong impression or not. Perhaps @Joshs knows better regarding what produces the particular "objectivities of the objects" for a Heideggerian so to speak.

Only that tasks (in the sense I think the phenomenologists meant it - 'doing the shopping', eating a sandwich'...) are modelled by areas of the brain several steps removed from the primary visual cortices. They'll have an influence by virtue of several stages of signal suppression, but it will be so watered down by that point that I wouldn't necessarily see it as a pragmatic influence. — Isaac

That seems to be within task, like within "eating a sandwich", do you think previous task information is blocked from influencing the current task? Or, in other words, does the task unfold along with the doing, so to speak? I get that you can partition off the regulatory signals once you've fixed a task you're describing, and it becomes somewhat post hoc, but can you partition of the regulatory signals in the agent's history from informing them what the current task is?

I don't really know how Heidegger would deal with objects showing themselves out of a background mechanically, certainly don't recall anything about it. — fdrake

Heidegger agrees with Husserl and Merleau-Ponty that there is no such thing as a task-neutral sensation. But his reasoning differs from these authors. MP and Husserl ground experiencing in perception and argue that even the most primitive perceptual features are the result of constructive processes in which expectations are crucial. Heidegger, however, believes all new experiences are bound up so directly in holistically organized pragmatic aims and significances that trying to ground Being in perception produces an artificial abstraction. Instead, he founds all experiencing on what he calls the ‘as’ structure. We see something ‘as’ something , that is, as the contextual, pragmatic way it matters to us in relation to our ongoing concerns.

“Da-sein hears because it understands. On the basis of this existentially primary potentiality for hearing, something like hearkening becomes possible. Hearkening is itself phenomenally more primordial than what the psychologist "initially" defines as hearing, the sensing of tones and the perception of sounds. Hearkening, too, has the mode of being of a hearing that understands. "Initially" we never hear noises and complexes of sound, but the creaking wagon, the motorcycle. We hear the column on the march, the north wind, the woodpecker tapping, the crackling fire. It requires a very artificial and complicated attitude in order to "hear" a "pure noise." The fact that we initially hear motorcycles and wagons is, however, the phenomenal proof that Da-sein, as being-in-the world, always already maintains itself together with innerworldly things at hand and initially not at all with "sensations" whose chaos would first have to be formed to provide the springboard from which the subject jumps off finally to land in a "world." Essentially understanding, Dasein is initially together with what is understood. In the explicit hearing of the discourse of the other, too, we initially understand what is said: more precisely, we are already together with the other beforehand, with the being which the discourse is about. We do not, on the contrary, first hear what is expressed in the utterance. Even when speaking is unclear or the language is foreign, we initially hear unintelligible words, and not a multiplicity of tone data.”

“ But why is it that this as-structure is already present in a direct act of dealing with something? The most immediate state of affairs is, in fact, that we simply see and take things as they are: board, bench, house, policeman. Yes, of course. However, this taking is always a taking within the context of dealing-with something, and therefore is always a taking-as, but in such a way that the as-character does not become explicit in the act. The non-explicitness of this “as” is precisely what constitutes the act's so-called directness. Yes, the thing that is understood can be apprehended directly as it is in itself. But this directness regarding the thing apprehended does not inhibit the act from having a developed structure.

Acts of directly taking something, having something, dealing with it “as something,” are so original that trying to understand anything without employing the “as” requires (if it's possible at all) a peculiar inversion of the natural order. Understanding something without the “as”—in a pure sensation, for example—can be carried out only “reductively,” by “pulling back” from an as-structured experience. And we must say: far from being primordial, we have to designate it as an artificially worked-up act. Most important, such an experience is per se possible only as the privation of an as-structured experience. It occurs only within an as-structured experience and by prescinding from the “as”— which is the same as admitting that as-structured experience is primary, since it is what one must first of all prescind from."(Logic,The Question of Truth,p.122)”

“ Acts of directly taking something, having something, dealing with it “as something,” are so original that trying to understand anything without employing the “as” requires (if it's possible at all) a peculiar inversion of the natural order. Understanding something without the “as”—in a pure sensation, for example—can be carried out only “reductively,” by “pulling back” from an as-structured experience. And we must say: far from being primordial, we have to designate it as an artificially worked-up act. Most important, such an experience is per se possible only as the privation of an as-structured experience. It occurs only within an as-structured experience and by prescinding from the “as”— which is the same as admitting that as-structured experience is primary, since it is what one must first of all prescind from."(Logic,The Question of Truth,p.122)” — Joshs

For @Isaac, translation of the as structure and highlighting an issue. The as-structure is a component of active perception that takes a thing as a thing. That configuration over there? That's seen as a bird. It is a bird. Maybe an analogy for it is if we think of perception as an object classification task, the "as" is the classifier sorting the environment into perceptual categories. Classifiers are (arguably) representations though; the predictors on the right of the equation prescribe the classifier value for the thing on the left of the equation. The as structure is sort of how the sensorium counts as the world, but the elements of the as structure are not applied on like "paint" to indifferent objects, it's more like a negotiation - you've learned to see stuff as stuff by interacting with it, so the world impresses upon the sensorium just as the sensorium impresses upon the world - to the extent even that there's no point making a strict distinction between them in all contexts, what I'm typing on counts as a keyboard -> what I'm tying on is a keyboard.

The "as" is also not necessarily derivative of perceptual information, it's conceived of as a component part of perception - like how old perceptual features are priors for new ones. It's a component of an ongoing "lived flow", it's the name for the feature that chunks it into meaningful bits.

You can suppress that basic meaningfulness through an intellectual/practical act in a restricted way, as you might while considering unclassified objects in the classification task (hiding what the objects are classified as, what labels they have and so on), but it's a suppression of a basic functioning in a restricted way - stuff still counts as stuff elsewhere in the act.

As I've read @Isaac's points, he was talking about spatial and temporal aspects of objects; boundaries, relative positions, evaluations of where, the kind of low level object recognition stuff our brain does. What "counts as" an object in this regime is more like spatiotemporal locale than a contextualised role. To use the classic example, Heidegger's hammer includes its for-hammering as a component of what it counts as (a hammer), the ecological affordance of hammering is part of the hammer, whereas for the object recognition stuff Isaac was talking about, it's much more about the object as a "object in space and time", a spatiotemporal locale, a solid trajectory through spacetime, boundaries, edges, that kind of thing.

It seems to me the relevant dispute regarding perception is whether, when you functionally split off those low level things from the upper level things, do you render the account which uses that functional distinction inaccurate, since "upstream", higher in model's hierarchy, the two are actually integrated interactive processes? There's doubtlessly contextual questions; to what degree can you split them? To what degree can you split them in what contexts? What is lost/gained from considering things like that?

what makes the sensible default hypotheses non-task relativeness for edge recognition and contrast detection; or whatever broader category they lay in; when the rest of the procedure is task-relative? — fdrake

tasks (in the sense I think the phenomenologists meant it - 'doing the shopping', eating a sandwich'...) are modelled by areas of the brain several steps removed from the primary visual cortices. — Isaac

Let me suggest the way that Husserl and Merlea-Ponty would answer the question of whether there can be any such thing as a non task-relative sensation. But first, I’m wondering whether such a concept would fall under Sellars’s myth of the given.

Varela summarizes Goodman’s version of this:

“To be objective, the interpretationist points out, one would have to have some set of mind-independent objects to be designated by language or known by science. But can we find any such objects? Let us look at an extended example from the philosopher Nelson Goodman.

A point in space seems to be perfectly objective. But how are we to define the points of our everyday world? Points can be taken either as primitive elements, as intersecting lines, as certain triples of intersecting planes, or as certain classes of nesting volumes. These definitions are equally adequate, and yet they are incompatible: what a point is will vary with each form of description. For example, only in the first "version," to use Goodman's term, will a point be a primitive element. The objectivist, however, demands, "What are points really?" Goodman's response to this demand is worth quoting at length:

If the composition of points out of lines or of lines out of points is conventional rather than factual, points and lines themselves are no less so. ... If we say that our sample space is a combination of points, or of lines, or of regions, or a combination of combinations of points, or lines, or regions, or a combination of all these together, or is a single lump, then since none is identical with any of the rest, we are giving one among countless alternative conflicting descriptions of what the space is. And so we may regard the disagreements as not about the facts but as due to differences in the conventions-adopted in organizing or describing the space. What, then, is the neutral fact or thing described in these different terms? Neither the space (a) as an undivided whole nor (b) as a combination of everything involved in the several accounts; for (a) and (b) are but two among the various ways of organizing it. But what is it that is so organized? When we strip off as layers of convention all differences among ways of describing it, what is left? The onion is peeled down to its empty core.”

Now let me analyze a notion like ‘edge’ in the way that I think Husserl might. Imagine that we are a
creature recently emerged from the womb and just beginning to make perceptual sense of the environment via our various modalities of reception and action. Husserl begins with the assumption that we only experience a sensation as that sensation if it is meaningful to us, and its meaning is bound up with usefulness , that is, how perceiving something helps us to navigate our environment , to pick up and handle
objects, to recognize and pursue sources of food, shelter, danger, etc. It might seem obvious , even primitive, how the perception of edges are useful to us(there can be no object differentiated from a background without contrast). But is an edge
the same thing as a contrast? Let’s think about what is necessary in order to have an edge. First of all, an edge is not the same thing as a point in space. It implies a multiplicity of points or contrasts of some sort. Could we say the that it requires recognizing a surface? What is it that composes a surface? Our geometrical knowledge tells us that a surface has such and such characteristics, but isnt this a higher order abstraction? There can be no such thing as an abstract surface in nature any more than there can be a straight line. The point isn’t simply to question the primordiality in nature of perfect lines and surfaces but to question the very concept of a line or surface as a sensory given rather than a relative constructive hypothesis.


Surfaces are imperfect in shape, color , hue, brightness, texture, etc, notnbecause the are imperfect exemplars of the category ‘surface’ but because the very notion of surface as a unitary entity is an idealization subjectively constructed. And if this is true of a surface it is also true of its boundary. Often our visual sense cannot confirm a boundary that fades and disperses and gets lost and blended with changes in light, shadow, color , depth, etc. Sometimes only the recourse to movement and touch allow for a construction that leads to a notion of something like a boundary. And how many different meanings of boundary might there be, depending on how we are seeking to interact with it?
A ball has a boundary which appears as an edge visually but only when we attempt to interact with it do we discover the notion of sphericalness. There are boundaries between planes which pre-suppose
the notions of ‘in-front-of’ and behind. Recognition of such ‘edges’ protect us from falling off cliffs. Sometimes we don’t need to know what’s
behind or in front of. Instead we need to know what is above or below, to the left of or to the right of. Perhaps it is the boundary itself we are pursing in a directional fashion. And of course, the orientedness in space of a perceptual feature does not originarily take place in objective space but in the subjective space
of embodiment. My body is the zero point relative to which everything that I perceive is correlated and is orientated.

These orientation concepts are complex constructions , as we know from brain pathologies in which one loses the ability to process left from right. There are also brain injuries that cause neglect of one side of the field of vision or of the body. This is not due to damage to sensory reception but to a kind of apathy. (In Schizophrenia we often find a failure to demarcate where one’s body stops and the world begins. This bleeds over into the a failure to recognize boundaries of other objects. The issue here with edges isn’t one of sensory input but of significance, the relations between the object and my aims. If purposiveness becomes fragmented, the world and its contours fragments along with it. )

So are all these examples of goal-oriented tasks just different constructions based on a task-neutral sensory primitive called an ‘edge’? Husserl would say that none of them are, and that in fact there can be no such singular category like ‘edge’ that encompasses these varying contextual constructions. Each is telling the organism something original and invaluable to their present need to interact with an aspect of the world , about the way a new feature contrasts with the previous(aboveness, belowness, behind or in front of, sphericalness) in relation to one’s bodily position and in the context of how one is specifically intending to interact with an aspect of the world.

The question then becomes, from Husserl’s vantage, where do we get the idea that there are such subject-neutral things in nature called edges or points? His answer was that these are the product of an abstract idealization of the perceived world that was invented between the time of Aristotle and Galileo in the form of objective geometrical mathematization. This idealization was itself
designed to perform certain tasks, but has been taken as the foundation for the analysis of the perceived world, in the form of objective sensation primitives. The examples I gave above become imperfect variations derived from a gemotricized subject-independent space-time model of sensation.

That seems to be within task, like within "eating a sandwich" — fdrake

Yes, I agree, but I'm not sure how relevant the salient features of the task are by the time it's just one of many signals competing for attention - we're never just eating a sandwich. In fact the task we think we're engaged in is more often than not a trivial distraction as far as the rest of the brain is concerned - which is far more engaged with standing up, digesting, checking we're not under any threat etc...

do you think previous task information is blocked from influencing the current task? — fdrake

No. That's a good point. Basically goes back to what I said above. We're never doing one thing at once, there's never 'a task' for our brain to be holistically oriented toward. Any sense that there is is post hoc narrative creation, not a live modification. When I'm 'eating a sandwich' not only am I also engaged in the things I mentioned above, but I'm engaged in filtering and processing the memories form the previous task or set of events, I may be still creating that narrative and some of the salience-related filtering in perception will result from narratives of previous tasks being 'replayed' for the sake of memory creation, rather than the actual task at hand.

I get that you can partition off the regulatory signals once you've fixed a task you're describing, and it becomes somewhat post hoc, but can you partition of the regulatory signals in the agent's history from informing them what the current task is? — fdrake

You can, but only to the same extent. In that tasks are always legion. Rescheduling is a case in point here, I think. when you turn on the light switch, you actually see the light before you fell the switch click (light, and the signals from your occipital nerves travel faster than the feedback from your pressure-sensitive nerve endings), but the signals from your occipital nerves are 'held back' from their pathways to scene-construction until the signals from the pressure-sensitive nerve endings get to the brain (which, on it's modelling assumption, is expecting them). The signals then get sent to to areas like scene construction and event narratives in the right order "I switched the switch and then the light came on". Various optical illusion like the violet chaser mess with this feature. The point is, it works because I reached out to switch the switch (previous task). Knowing I did that, has rearranged the events of the current task (checking the status of the lighting/room etc).

Heidegger, however, believes all new experiences are bound up so directly in holistically organized pragmatic aims and significances that trying to ground Being in perception produces an artificial abstraction. Instead, he founds all experiencing on what he calls the ‘as’ structure. We see something ‘as’ something , that is, as the contextual, pragmatic way it matters to us in relation to our ongoing concerns. — Joshs

See above. How does this square with the multiplicity of tasks and saliencies we have? We don't see something 'as' something, we see something 'as' many somethings, relevant to many different tasks, we only unite them later when creating the unifying narrative. I'm sure I don't, with you, need to go through the acres of evidence for post hoc narratives as unifying memory initiators out of dissonant sensory or interoceptive interpretations do I.

It seems to me the relevant dispute regarding perception is whether, when you functionally split off those low level things from the upper level things, do you render the account which uses that functional distinction inaccurate, since "upstream", higher in model's hierarchy, the two are actually integrated interactive processes? — fdrake

I think this depends on the influence the upstream priors have on the accounts rendered by the lower level models. The answer, at the moment, seems to be fairly little - tinkering at the edges stuff.

Basics like foreground/background, container/contained are distinctions made very early on in visual processing and are powerful (by which I mean higher models are more likely to re-interpret to match these outputs than suppress those outputs to match their priors). There's even some speculation that the container/contained distinction suffuses much of our higher conceptual thought...bu that's another thing. Point is there's very little the higher models can do to switch a background/foreground decision, or a container/contained decision. There's a paper by Seth exploring some of the hoops the higher models will jump through when he uses virtual reality headsets to switch these basics.

I think a good analogy of the relationship between models in the hierarchy is a well-managed company. The CEO might not even know how the accountants are working out how much tax the company owes, but he'll definitely asset his influence in hiring and firing them, in showing displeasure at a high bill and pleasure at a low one, the accountants will be well aware what their aim is, but nonetheless, the CEO hasn't the faintest idea what they're actually doing, and certainly wouldn't personally re-arrange a few columns.

Let me suggest the way that Husserl and Merlea-Ponty would answer the question of whether there can be any such thing as a non task-relative sensation. But first, I’m wondering whether such a concept would fall under Sellars’s myth of the given. — Joshs

I think you've read 'task-independent' where I'm talking about 'not the task you think you're focussed on'. All signals have to be interpreted according to priors and those prior will be influenced by the current state (which, given the nature of the dynamic environment to which we must respond, will always be some task or other). The point I'm making is that we can't (as the phenomenologists would have us do) reverse-engineer this effect, because the 'task' that's relevant to the priors is not necessarily one we're even aware of, and certainly one on many going on at the same time.

The point isn’t simply to question the primordiality in nature of perfect lines and surfaces but to question the very concept of a line or surface as a sensory given rather than a relative constructive hypothesis. — Joshs

Again, I think you're misconstruing the active inference approach as implying objective properties when it's doing the exact opposite. It's using the exact same constructive hypothesis of 'edge' are you're describing here. The difference is the modelling assumption it's using is not a 'higher order' one accessible to introspection, It's a very basic one (probably at least partly hard-wired).