these behaviours are not the definition (unless you are a behaviourist) of thoughts and experiences. — bert1

Right. So why not adopt a behaviourist position as the simplest model?

The definition of thoughts and experiences involves something other than a behaviour. — bert1

This seems to be the problem. You define thoughts as something ineffable and then act surprised that there's no physical explanation for them. It's the inevitable consequence of such a definition. If thoughts just are neurons firing, the problem is dissolved.

But, notwithstanding the above, the thing I really want to focus on here is how you would know that you have your sought after 'explanation'. What would it do that 'neurons firing' doesn't?

just one little piece of evidence. — Chester

Why buck the trend now, you'd been going so well without any whatsoever!

The explanation of this correlation is still wide open. — bert1

As I asked @schopenhauer1, what would an 'explanation look like? What properties of an explanation are missing from "neurons firing seem to cause what we experience as thoughts"?

frank specially presented a definition of God that he was working under — Hanover

No, he associated it with feelings "most" people have about the inconceivability of uncaused events. He claimed that it was the reason why God was used as a name for it, because of the connotations the word has. I'm denying that there's any evidence the word has those connotations for "most" people and therefore its very unlikely to be the reason why it is used in that way.

if "mind" "emerges" from the physical, the emerging from one type of realm (the physical), into a completely different kind (the subjective, internal, "feels like", qualia, mental) state needs to then be explained in non-dualist terms. — schopenhauer1

And what would such an 'explanation' look like? How would you recognise that some proposition constituted an 'explanation'? I ask because such avoidance seems to dog these kinds of discussions. Some physical relationship is proposed by the (non-panpsychic) physicalist, and the 'hard problem' crowd will inevitably respond with "but that's just a description of how, not an explanation accounting for it". What I've yet to hear is a reasonable definition of what such an explanation should be like.

We can do how - neurons firing seem to cause what we experience as thoughts.

We can do why - having the experience of thoughts seems to help integrate information better than letting individual circuits act independently.

What's missing?

This is why "God" is sometimes used as a name for it. That word connotes, among other things, an insurmountable mystery. — frank

Nah. It connotes a guy sitting on a cloud in charge of stuff. "God help us!", "God knows!", "Pray to God that doesn't happen", "God loves his children", "God said to Abraham..."... This ineffable mystery crap is just tacked on post hoc when we look at the top of the cloud and find it glaringly unoccupied.

@Banno

I'm finding it difficult to discern the issue you're focusing on here. Is it a) there are jobs which produce nothing of use, everyone knows this but it serves the wealthy to maintain such, or is it b) people have a perception of their jobs as having no value, or producing little of use?

When talking about the cause of this state of affairs, you seem to assume (a), but when drawn on the apparent subjectivity of 'producing nothing of value' you revert to (b).

Both issues are equally meritorious of discussion, but I don't think they're the same thing at all. One is political/economic, the other psychological. Which is it you're focusing on - or do you disagree with my separation?

communism, like it's foundational father socialism, is grounded in covetousness and theft. — Contra Mundum

And there's an economic system that isn't?

the incentive to work for one's wages in such a system dies when those who work far less are paid the same as those who put in far more effort. — Contra Mundum

Again, do you have a system in mind where wages are paid on the basis of effort?


I see your objection, but not your alternative.

Huge areas have virtually nothing in them. So I'm not so surprised that the UK is weird on a landscape fragmentation measure. — fdrake

Yeah, smaller countries tend to have higher indices of fragmentation because open space is at such a premium. I guess the CPRE must be a lot more powerful than anyone gave them credit for. 'God save the village green!'

The measure's also very local; it's not going to measure international connectivity or commuting/travel intensity within or between countries. — fdrake

Yeah. I hadn't thought of that because we were comparing countries, but of course you're right, a country's exposure to other networks will make a huge difference to the progress of the epidemic.

Something based on a population movement network, maybe?. The virus spreads along the interaction networks of people, so a decent connectivity measure for covid probably wants to track an interaction network rather than something that reflects land geometry. — fdrake

I agree, but I think for something like a global pandemic we need static measures, which means leveraging the inference within static networks to imply responses in the dynamic ones... Or, I suppose we could just use 'snapshots' of dynamic networks as a proxy. I like your idea of using flow models rather than network models though. I might see what data there is on that.

If I were Google I'd probably have a gigantic inter-and-intra national population flow database that spanned the globe and had second to second resolution. And I'd be keeping that quiet. — fdrake

Yes, all this proxy data analysis is obviously moot since Google know where all of us are in real time (as well as what we're doing, who we've met, and when we last bought a sandwich)! We could just ask, but then it wouldn't surprise me if Google cooked this whole thing up just to market some tracing app they'll bring out next year (obviously with help from the CCP, Huawei, the illuminati, the lizard people from the centre of the earth, and Uri Geller - who are all in on it together.... Now where did I put that tinfoil...? ).

If we find more evidence of huge numbers of people with antibodies, indicating a much higher proportion of asymptomatic cases than originally thought, for example, that would suggest when we come off lockdown, in the absence of a vaccine, we're going to end up in herd immunity territory anyway and the Swedish model of mostly voluntary distancing might look like a better idea than a straight comparison with its neighbours currently suggests. — Baden

Yes. The results from the widespread prevalence testing in Iceland suggest about 50% of those testing positive (for active disease) were asymptomatic. This tallies with quite a number of other smaller studies, such as the Vo’Euganeo one from Italy. Finding 50% asymptomatic at testing increases the liklihood that a considerably greater number will have antibodies (as we have to also include those who were symptomatic but did not seek treatment).

The key issue, I think is that, like it or not, there's an inevitable trade off between immediacy of lockdown measures and severity of lockdown measures. The less severe the lockdown needs to be, the easier it is going to be to implement it quickly next time. The more severe the lockdown measures need be, the higher the threshold of certainty that will be required to act - and that delay could prove fatal.

As I said way back, I think we need timely, targeted and confident action next time. To achieve that we need good, accurate data and - more importantly - people willing to follow good accurate data wherever it leads.

The trouble is people have already become so emotionally invested in flag-waiving for their favoured course of action (and who can blame them, given the stakes), that I don't have much hope that the politically expedient course will match the data.

The average number of voronoi cells per 1000km2 would probably track the amount of unpopulated/uninhabited/unconnected areas too. Those areas would have huge cells in them, that would massively pull down the average over the landmass area compared to what it would be if constrained to population centers. — fdrake

Agreed, but what I was trying to quantify was connectedness, but in a manner which included urban sprawl (so hub distances or connectivity measures wouldn't quite capture it). Voronoi meshes will take into account the open spaces, but it will do so in a way which biases in favour of accounting for network links (roads and railway). A single road connecting two urban areas will double the number of meshes relative to the same area without a road. I'm sure there are better ways of doing it, but I think the impact of a single road captures connectivity in a way which outweighs the bias toward open space. If it didn't, then France (good network but low population density) would come out lower than UK (higher population density but crap networks).

Although... As I said, I'm very surprised by just how much lower the UK was, so I might have to check my figures again.

Have you got any ideas as to how we might better capture the degree of connectedness?

They're not comparable due to different moments of starting measures and different types of measures. — Benkei

Yes, that's rather the point I was trying to make. It's not possible to tell if population density has an effect because we cannot control for those other factors. All we can see is that there's no noticeable effect to be explained.

The conversion to deaths per capita is meaningless. — Benkei

I agree. I was only using the figures that were being discussed, for consistency. The raw case numbers show no meaningful trend with population density either. Cases per capita is still a useful measure though. If local population density is relevant, then comparing raw case numbers (for the sake of assessing the effectiveness of responses) is not going to be accurate without some reference to the population. Consider the raw case numbers for New York compared to Eritrea. Cases per capita is flawed, but no less so than raw cases. It depends on what you're trying to get the data to tell you.

Population density has an obvious effect on R0 because it increases the number of contacts an average person is likely to have, which creates an opportunity for the virus to infect another. — Benkei

As your cited study makes clear, this is only relevant on a local scale. The degree of connectivity between relatively closed networks will be far more important on a national scale, which is the scale the figures are being compared on.

As I can imagine you've been on the edge of your seat waiting for the actual data...

Anthropogenic landscape fragmentation based on voronoi mesh density (meshes per 1000km2) - a measure of how dense network meshes are created by roads, railways and urban sprawl.

UK 2 - 4 (which really surprises me)
Netherlands 35 - 75
France 20 - 35

Still very different between countries. Still doesn't seem to have an effect on case numbers.

Incidentally Sweden's is less than 1, same as Finland. Both more than Norway at lower than 0.1 (they basically don't seem to have any roads at all in Norway!)

It's crude measure. Consider that the highlands of Scotland Mid Wales, and the Pennine hills have a density of maybe 1/km2. And France has the Alps, the MassifCentral and the Pyrenees. What one wants is a sort of mean distance between habitations — unenlightened

Absolutely.

Voronoi triangle sizes, or Katz centrality measure for each conurbation is what we need. I didn't have those figures to hand.

'm hypothesizing that if they had had better behaviour pre-lockdown they'd be more like Norway now. I'm also hypothesizing that if Sweden had locked down after (having had good pre-lockdown behaviour), they'd be more like with Norway now. — Baden

OK. I understand what you're saying now, but I'm not sure what you're basing it on. If we're trying to establish the effectiveness of lockdowns we can't assume the effectiveness of lockdowns as part of our hypothesis. You'd need to control for the other factors, which, if you have done so, I haven't picked up on in the way you've presented your theory.

Edit - just to be clear, I mean effectiveness of types of lockdown. There's so much conspiratorial garbage being spouted I thought I might need to make that clear. The effectiveness of some kind of lockdown is, I think, beyond question. I just think it's very important for next time (and there will be a next time) that we properly learn from the experience rather than just justify post hoc whatever it was we advocated most strongly for at the outset.

I think population density is an important factor where the rate of infection is concerned. — Benkei

UK population density 274/km2.
Netherlands population density 419/km2.
France 123/km2.


UK 2685 cases/million.
Netherlands 2368 cases /million.
France 2580 cases/million.

It doesn't seem to have made any difference here. A near doubling of population density in each, with no noticeable change in case rates.

So, comparing like with like, it makes more sense to put Sweden up against other Nordic countries. They've got ten times as many deaths as Norway, for example, with just under twice the population. — Baden

That doesn't follow. If the reason why Ireland has similar per capita cases to Sweden (despite different lockdown strategies) is because the Swedish essentially locked themselves down, then their different lockdown strategy can't also be to blame for the differences between them and other Nordic countries.

Either the Swedish had an effective lockdown or they didn't. If they didn't, then the fact that they have nonetheless managed the same caseload as Ireland is surprising. If they did, then the fact that they have a higher caseload than other Nordic countries becomes surprising. It can't explain both.

that does not define the experience you have of looking at the pencils and thinking there's three of them. It merely defines how the experience is produced. — ernestm

What kind of thing would 'define' the experience. You seem to be using 'define' in an odd way here. If I 'define' my experience of eating a pizza I usually mean by that some verbal account of the senses and thoughts that occurred. That's easily done with your pencil example. What more of a 'definition' are you looking for?

It's the more general point that we have this phenomena, we have these developing lines of enquiry in physics, so we should be exploring all options. — RolandTyme

Ahh, then I'm afraid you've lost me as far as what these "developing lines of enquiry in physics" are, if not quantum physics. Perhaps you could expand on what other areas of physics you think might be profitable lines of enquiry for studies of conciousness?

If I've understood you correctly, this sort of idea was explored by Roger Penrose in the 80s and early 90s. He wrote a book on his idea of how vibrations of microtubules could maintain quantum superposition. See here.

Most neuroscientists thought that it was impossible and duly it was calculated that any such state could not possibly last long enough to transmit a message, here.

I think the idea has been shelved since, but I may not be entirely up to date on it, others may know something more recent.

in the modern world, the physical traits/characteristics that we are born with no longer seem that necessary for us to survive. We no longer need to be athletic in any sense of the word to survive and reproduce. The same can generally be said for our intelligence and even health to a large extent. — Pinprick

The only reason why this is the case though is because we have a medical and economic system which dramatically increases infant survival rates. I don't see how 'memes' (by whatever definition you're using) are any more required to perpetuate that than genes are. What's required to perpetuate that (in a modern setting) is mainly cheap food, clean water, vaccines and antibiotics. To a lesser extent, but still significant, an established medical institution. I'm not seeing the link between these things and 'memes' at all.

I was talking about the written information, not the logistics of the library. — unenlightened

OK, we can limit the examples to that. So "trees are green, the sky is blue" contains more information than "trees are green, trees are green".
But still, this is only true with reference to the question "what colour is stuff?". With reference to the question "what propositions do those sentences contain?" they both have the same information (in terms of entropy reduction). Think of it in terms of knowledge. In the case of the first question, I started out uncertain about two colours. After reading the first sentence I'm now certain about two colours (100% reduction in uncertainty). After reading the second sentence I'm only certain about one colour (50% reduction in uncertainty - less information in Shannon's terms).

But in the second case, I'm uncertain about what the author will say in each half of the sentence. After reading the first sentence I'm now 100% certain what the author will say. After reading just the second sentence I'm now also 100%certain what the author will say. Same amount of information (in Shannon's terms). The repeat in the second sentence is still something the author said and I still had 100% uncertainty about whether they were going to say it prior to reading it, which was reduced by 100% after reading it.

Quantity of information depends on the question and so requires a questioner. We can't therefore measure the amount of information in something without reference to which uncertainty is being reduced.

Is this controversial? — unenlightened

No, I don't think it's at all controversial... in a world of people for whom the relevant information is what the words in the books say. What I think is controversial, at least to me, is...

the information content cannot change except by the destruction of the image. — unenlightened

If the information content cannot change, then I can only interpret that as meaning all the information (the location and state of all the matter in the picture), regardless of the expectations of an observer (otherwise the information would change - it would be different for a different observer).

Consider a pattern of raised or coloured dots. Each is either blue or raised but never both. The dots spell out the letter A in either raised or blue dots (the background dots are neither raised nor blue). To most people the information is not random. They recognise the 'A' pattern. To a blind man, the dots are random, he only distinguishes the raised ones. His expectation uncertainty for subsequent dots in the pattern has not been reduced by the state of the first few.

If you want to claim that this is merely the fault of the blindness, that the pattern is 'really' there and so also the information, then you'd have to say the same for every conceivable pattern (electromagnetic variation, electron spin, fibre angle...). And doing so, we're back to an ordered picture having no less information than a disordered one because it is only ordered in one context.

To bring it back to your library example. What you say about repetition not increasing information would be true for everyone except the auditor. For him every repeat of a book is a new piece of information. So, in the absence of either readers or auditors, we cannot say that the amount of information in the libraries is the same. Blind everyone and promote auditing to a religion and suddenly the information contained in your repetitious libraries has indeed gone up. The total count of books has become relevant, their content less so.

There's minimum description length: how precisely something can be stated in full — fdrake

Thank you. That's the name of the thing I was scrambling for when I was taking about 'efficiency of data compression'. I couldn't remember the correct term.

Each repetition of my sentence can be seen as further confirmation that I did not accidentally write "blind" when I actually meant "bland". — unenlightened

It can be, yes. I was struggling to marry that with your idea that the information was somehow in the picture (your first example), or in the paragraph here, regardless of the interpreter.

What you said above is only true if my expectation is that you're about to make a proposition. Then my uncertainty is resolved by the first sentence, and only resolved marginally more by the subsequent ones.

If, however, my expectation is that you are about to deliver a 200 sentence paragraph and I'm uncertain as to the content of each sentence, then neither the content of the first sentence, nor the pattern generated by the first few, reduce that uncertainty without my interpretation.

The information reduced by repetition cannot be both expressed in terms of Shannon's reduction in entropy, and be the same regardless of an interpreter. A pattern only reduces uncertainty by expectation. Someone needs to be doing the expecting.

I don't understand the connection you're making here. If the question I'm asking of the dataset is "what is unenlightened saying?", then the repetition carries no more information. If the question I'm asking of the dataset is "what state is each sentence in?" then the measure of information is the degree of decrease in probability for the total unit. Say each sentence could be either 'You are blind' or 'You are not blind', then each has a P=0.5. Every sentence which is specified by your post reduces that uncertainty. So the information is the total degree to which uncertainty is reduced (the total possibilities which are set to certainties). At least, that's how Shannon defines it.

What you seem to be describing is the efficiency of data compression, which is not the same thing as information quantity (in terms of probability reduction).

The only way I can see you conflating measures of information compression with measures of information quantity is with expectation. If the value of one sentence somehow dictates the following value, then my expectation is lower (less information has been carried). But you ruled that out by saying that the information was 'in the picture' and had nothing to do with the interpretation.

So I'm lost as to what you're trying to present here (yet it sounds interesting). Are you trying to explain Shannon - in which case you need to make clearer the difference between compressibility and total information, or are you presenting something new of your own thinking?

Have you tried writing a post on the topic yourself? (I tend to follow a very narrow range of threads so I may well have missed it).

Sorry if I'm being a bit slow, still not sure if you're talking about about meta-philosophy, or an attempt to tackle the entirety of philosophical investigation in one go.

I still get the impression that most people here aren't interested in the same kind of big-picture philosophy-as-a-whole thing that my interest is all about. — Pfhorrest

That's interesting. Could you say what sort of interaction would indicate that people were interested in 'big-picture philosophy-as-a-whole' stuff? I mean, there's hundreds of posts here on all sorts of different topics ranging from the origins of the universe to some specific quote in a particular academic text. I'm finding it very hard to understand what you could mean by interest in 'big-picture philosophy-as-a-whole' stuff that wouldn't fall somewhere in that spectrum.

Isaac DESTROYS evolutionary psychology. (Maybe). — fdrake

Ha! It was asking for it, it spilled my pint.

How I'm thinking about emotions in the natural kind flavour are that they are attractors in the dynamical system of active inference given the statistical regularities of our current lifestyles. — fdrake

Yes, that's it exactly. The essentialist rhetorically asks "Why are expressions of happiness (smiling, laughing, dancing...) similar across the world?" . But the answer is not because it is some natural kind as they might hope. It's because it's a statistical regularity of some basic aspects of human culture. The same can even be said even of biological regularities. Turing's reaction-diffusion equations for example are the basic explanation for most skin and shell pattering in nature - nearly. They explain them sufficiently to be regarded as the source of the regularity, but not quite sufficiently to be predictive with 100% success. If I were to bet on how a shell pattern might come out given the mechanics of it's growth, I'd bet on the patten generated by reaction-diffusion equations, but they're just a statistical aggregate, not complete description. So what I'm saying is that I don't think the re-imagining of classes as statistical summaries here is limited to cultural artefacts - as you mentioned to me earlier, this is all about population thinking.

It turns out, that one of the factors which disrupts the influence of reaction-diffusion equations on skin patterns is temperature. It changes the chemical reactions in the cells sufficiently to break them out of that particular pigmentation reaction. I think with emotion and emotional responses, we have a kind of Nash equilibria which we've learnt, so we're reluctant to change strategy, but similar to the skin pigmentation patterns, there exist environmental variables which shift us out of that algorithm.

A more complicated attractor might be whether an asteroid would enter into orbit around Earth. It'll come from some angle, and when it non-negligibly gets pulled by Earth's gravity, it might start to rotate around Earth. The attractor there would be the collection of all orbits around Earth that the asteroids take. — fdrake

Have you read any Jack Cohen? He applies the sort of maths you're referring to here to biology (from a mechanistic point of view - how mechanisms in biology like cell mitosis yield semi-chaotic results, but with strange attractors toward the familiar end results).

I'm unclear whether "state" refers to something like the state of a neuron, or whether it refers to something like the state of an environmental parameter, or whether at one stage in the process it refers to an environmental parameter (well, in its encoded form) and at others it refers to neuron states. — fdrake

I think the 'state' of an environmental parameter is outside of the Markov blanket, so we're only dealing with states of neural cortices. The initiation is from the signal (from the eye, skin, nociceptors, etc) which originates in the exterior world, but does not necessarily represent it (it will be an extract, biased by the the response from the previous inference). I'm aware of the fact that we have an unfinished discussion about the extent to which it does represent it - I don't think we quite agreed on it. Incidentally, @javra I think this is where you're misinterpreting what Barrett means.

my entire argument pivoted on decisions, thoughts, and some certain emotions not being perceptions – hence on our knowledge of these not being empirical. — javra

Barret is presenting a theory of cognitive processing, not epistemology. As such, none of the inputs are empirical in the sense you're using here. The inputs into the active inference models are all hidden variables outside of the Markov blanket for the system. Perception inference systems deal with raw signals (not raw data - that would require interpretation, which hasn't happened yet). The emotion related cortices are only dealing with signals fro other parts of the brain, so when Barrett refers to perception and interoception, she's referring to signals from parts of the brain responsible for predicting the cause of such raw inputs, not the raw inputs themselves. So introducing signals from other parts of the brain has little to no effect on her model.

Anyway,

The system described regarding habit formation in the Friston paper you linked doesn't have this "gets stuck there forever" property regarding habits though, a prior becomes change resistant by having its updates diminished by previous success using the policies (actions/worldly interventions, in the paper foraging strategies in a maze) it proposes. — fdrake

Indeed, but does what I've talked about above bring other biological systems a little closer the Friston's habits? I think there's not such a dividing line as all biological systems seem to be behave like this. Priors for predicting the patterns of a seashell seem stuck in reaction-diffusion equations - until temperature increases beyond the threshold for the model. But that's an aside - I see what you're getting at here.

So thinking of emotions (not core affect alone) as learned, they would need to be change resistant habits that activate based upon context similarity to the predictions (bodily-environmental model) their representations/encoded patterns generate. When evidence accumulates that the activating context for the habit is no longer present, the agent switches to an exploratory mode that yields the formation of new habits. — fdrake

Exactly. And how many other mental habits fit into this pattern. We have emotion and learning thus far. Logic? Embodied training (like riding a bike)?

Language seems to have the ability to prime which habits are simulated and enacted; and language as a cultural artifact/shared repository of symbols and meanings changes much more slowly than the fleeting associations that shape our emerging experience of emotions. It's a relatively time stable network of associations we partake in by analogous simulations. Moreover, language plays a mediating role in valuation of core affect. So: it changes slowly, it primes for which habits to activate by being a context, it mediates valuation in accordance with its own system of associations. It also seems to amplify predictions/interventions that are more typical of it when it's used as a prime (people primed with angry words report faces as more angry). — fdrake

Yeah, how can you express an emotion that's somewhere between anger and fear when your entire language, you're whole means of talking (and possibly even thinking) about the world doesn't contain a word for such a feeling? Like with Wittgenstein's talk about 'pain'. We're not only using the word to describe the socially shared experience, the mere existence of the word is acting as a resistor to updating or modifying that experience too quickly, which is necessary for communication to work.

What interests me here, is the extent to which this resistance to change from a predicable pattern actually serves a social function of it's own. Like the influence the word 'pain' has on our ability to express nuances of feeling actually serves a function (if we each had our own unique word to define 'our' pain, we'd never be able to talk about it). emotions are, at least in a large part, a means of social communication. It's possible that some of the restrictions society places of the classification are acting in a similar way - constraining private variety to make public expression meaningful?

They're still "optimal predictions" (in some sense) given their constraints and priors, but that does not mean the priors reflect the relational dynamics of the body and its environment, and their potential developments given the interventions I propose. — fdrake

I think that's really one of the important take-aways from the active inference model of emotional construction. Sufficient values from our environment are put into the priors for the model (and indirectly into our perception - but Barrett doesn't talk much about that) that we have to dismiss the notion of emotions being beyond judgement as in getting rid of the idea that "I'm just a feisty person", "I can't help myself" etc. Obviously there's a focus on self-help, but I think no less a focus should be put on societal influences. sociol-politics aside though, the point is that we can be wrong about our emotions not just in the sense that they're not suited to our modern world (that idea has been around for years) but that they're not 'suited' to any world, they don't come pre-packaged and suited to some set of circumstances predicted by evolution. They are an adaptation formed by the process of living within a social environment, learning from its cues - we don't get angry when the shopkeeper short-changes us because our bodies have an inbuilt system to fight off lions, we get angry when the shopkeeper short-changes us because getting angry in such a situation is a response which explains the entire situation (heart-rate, shop-keeper, coins, justice, monetary value, pride...) in a package which has produced least errors in the past.

I think one of the sources of confusion in many of the posts here (not yours) is in the false impression (Barrett is not particularly clear on this, I have to say) that when she talks about perception and interoception she's talking about pure data. Lot's of people seem intuitively turned-off by the model because they're seeing it as {heart-rate, image of a lion, nausea}='fear'. But this is, of course, not what she's saying. As we know, the forward-acting signals from the perception and interoception cortices are themselves predictive models, so the emotion constructing system is not getting raw data, it's getting interpreted, meaningful data

This information is available for later use by limbic cortices as they generatively initiate prediction signals, constructed as low-dimensional, multimodal summaries (i.e. ‘abstractions’); these summaries, consolidated from prior encoding of prediction errors, become more detailed and particular as they propagate out to more architecturally granular sensory and motor regions to complete embodied concept
generation

The way in which goal-oriented prediction errors might be fed back into the emotion system, as you're talking about with infelicity, is only hinted at by Barrett...

the salience network tunes the internal model by predicting which prediction errors to pay attention to [i.e. those errors that are likely to be allostatically relevant and therefore worth the cost of encoding and consolidation; called precision signals

...but Friston goes into it in more detail here (still not the right paper I promised you though!). The whole paper is really interesting, but the bit relevant to what you (and @StreetlightX) are talking about is section 5.2 (save you wading through the whole thing). It talks about how active inference models deal with (and better explain) the formation of habits which are in contravention of goals - ie ones which are no longer updating priors. It's quite complicated, but basically, the response from the habit (the forward-acting signal) become the expected input in an higher-order model. As such, so long as the habit is delivering the predicted output , there's no need to change the model-choosing model above it (in the hierarchy). it shows (or claims to) how an uncertainty reduction model better predicts habit vs investigative choices than a purely goal-directed model.

The point was a more general one (which I should perhaps have made clearer). There's tons of evidence, in general, covering when and how to use ventilation. More than enough to justify speculating on it in novel situations where there's pressure to act - contrary to what Hanover was intimating.

@Hanover

https://www.evidence.nhs.uk/search?q=guidelines+on+ventilated+patient

I was looking for a discussion, not an argument.

What set of core affects correlate to the cognitive state of envy? — javra

Here. An number of regions are identified.
https://www.sciencedirect.com/science/article/abs/pii/S1053811916303792

What I'm interested in is how you came to your conclusion. Obviously if you feel envy (or imagine yourself feeling envy) you don't have an fMRI scanner wired up to you, so what was your line of thinking that lead you to conclude there were no core affects?

what I've been upholding is that some emotions take place in the absence of core affects ("feelings" thus interpreted as interoceptive) being interpreted via emotion-concepts. — javra

How would you know?

Barrett's theory of emotions is an entirely unconscious or involuntary view of what's happening 'under the hood'. If this is right, then how does our conscious feeling of emotions tie into this? — Luke

That wasn't my intention, only to give you an idea of how both theories deal with involuntary responses. Both theories also deal with voluntary responses too. The degree to which you feel a decision is being made is not a distinguishing factor between the approaches. They really differ in the manner, felicity and the breadth in which the contributory factors are collected into a class.

Classic emotional theory has a consistent, entirely physiological collection of interoception states directly form an identifiable emotional state which then informs behavior (either via influencing concious choice, or directly).

Active inference theory has a varied collection of interoception states, together with perception states (from the environment) form a model predicting their cause. This model then initiates reactions (again, either direct involuntary or influencing voluntary choices). One of these actions is the labelling of the experience with a learned emotional label. These actions both then modify the environment and the perception/interoception states which modify the model, and so on.

If you wanted to read about it, IEP talks about Austin's use of the term; it's a more general failure/unsuccess category than "right" or "wrong", and he applies it to speech acts. My motivation for using it was to stress that goal/task relevance acts as a constraint in active inference, the "failures" we have with it are also rooted in comparisons to what we're trying to do. — fdrake

Thanks. I've had a look at the IEP and re-read the exchange in that light. It now makes a bit more sense, I think. You're saying that certain emotional responses resulting from the model might betray some higher goal even though they're the correct output from the model - so infelicitous, not "wrong"? That there could be a situation where we confuse the accurate function of the model for an accurate output? Like presuming that if a car is running really well it must be taking us where we want to go?

If so do you not think that the infelicitous output would simply constitue a prediction error of some higher model? Tom Fitzgerald has done some work with Karl Friston on active inference and habit formation which covers some of that ground. I might PM you with it though, I suspect the online equivalent of a series of blank looks if we start discussing it here!

We don't always have emotion experiences, even when we are being emotional. — jkg20

Then how do we know we're being emotional?

A belief about a probability is not identical to a probability. — Pneumenon

A documentary about documentaries is also a documentary. A poem about poetry is also a piece of poetry. A belief about a belief is nonetheless a belief. Something being about something else doesn't act as a universal argument that it is not also a type of the thing it is about.

So associating beliefs with a 'disposition to act as if', and associating a 'disposition to act as if' with probability (basic Bayesian position) is not undermined by the argument that sometimes things which are about things are not an example of the thing which they're about. Because sometimes they are.