From an evolutionary standpoint, why does pain exist? The first naive answer most people reach for is that pain exists to make us take action to prevent damage. If we touch a hot stove, pain makes us pull our hand back.
But that’s not right. When we touch a hot surface, nociceptors in our hand send signals to the spinal cord, which often respond with a reflexive reaction, such as a withdrawal reflex. When the signal makes it to the brain, further automatic survival action patterns may be triggered, such as reflexively scrambling to get away.
But all of this can happen before, or independent of, the conscious experience of pain. So why then do we have the experience itself? It isn’t necessarily to motivate immediate action. The reflexes and survival circuitry often take care of that.
I think the reason we feel pain is to motivate future action. Feeling pain dramatically increases the probability that we’ll remember what happens when we touch a hot stove, that we’ll learn that’s it’s a bad move. If the pain continues, it also signals a damaged state which needs to be taken into account in planning future moves.
So then pain is information, information communicated to the reasoning parts of the brain, and serves as part of the motivation to learn or engage in certain types of planning.
People often dislike the conclusion that pain, or any other mental quality, is information. It seems like it should be something more. This dislike is often bundled with an overall notion that consciousness can’t be just information processing. What’s needed, say people like John Searle and Christof Koch, are the brain’s causal powers.
But I think this reaction comes from an unproductive conception of information.
I’ve often resisted defining “information” here on the blog. Like “energy”, it’s a very useful concept that is devilishly hard to define in a manner that addresses all the ways we use it.
Many people reach for the definition from Claude Shannon’s information theory: information is reduction in uncertainty. That definition is powerful when the focus is on the transmission of information. (Which of course, is what Shannon was interested in.) But when I think about something like DNA, I wonder what uncertainty is being reduced for the proteins that translate it into RNA? Or the ones that replicate it during cell division?
Historically, when pressed for my own definition, I’ve offered something like: patterns that, due to their causal history, can have effects in a system. While serviceable, it’s a bit awkward and not something I was ever thrilled with.
Not that long ago, in a conversation about information in the brain, philosopher Eric Schwitzgebel argued simply that information is causation. The more I think about this statement, the more I like it. It seems to effectively capture a lot of the above in a very simple statement. It also seems to capture the way the word is used from physics, to Shannon information, to complex IT systems.
Information is causation.
This actually fits with something many neuroscientists often say: that information is a difference that makes a difference.
This means an information processing system is effectively a system of concentrated causality, a causal nexus. The brain in particular could be thought of as a system designed to concentrate causal forces for the benefit of the organism. It also means that saying it’s the causal powers that matter rather than information, is a distinction without a difference.
The nice thing about this definition is, instead of saying pain is information, we can say that pain is causation. Maybe that’s easier to swallow?
What do you think? Is there something I’m missing that distinguishes pain from information? Or information from causation? If so, what?
73 thoughts on “Pain is information, but what is information?”
Is pain a metabolic process? After all, you couldn’t have pain without your brain consuming glucose. Yet saying that pain is a metabolic process might not be very explanatory. That’s the worry I have about saying that pain is information.
Now to a different aspect of the subject.
Judea Pearl’s book Causality has what I think may be the best treatment of how we model and know about causality. In Pearl’s definitions, causal diagrams can have “directed” edges, drawn with an arrow pointing from parent node to child node. And they can have bidirectional arrows, in which case the nodes so related are called spouses. These edges are investigated via “interventions”, which set the value of a node to a particular state in a way that bypasses the normal influence of the other variables in the model. Both spouses and children of the intervened-upon node will feel a “difference that makes a difference”.
But note, in Pearl’s terminology (which is pretty common-sense), only the children (and grandchildren, etc.) are related to the source node by “cause-and-effect”. Spouse nodes are neither causes nor effects – but they are still related in a “difference that makes a difference” way. So information, in the sense you want, is broader than causality.
LikeLiked by 2 people
I wouldn’t say pain is primarily a metabolic process although like anything in the body it has metabolic aspects. But I would say it’s primarily information. Of course, technically everything is physical information, but for most things, the differentiation part is relatively low compared to the energy magnitudes. But with pain, the information is the primary part.
I haven’t read Pearl, so I can’t speak to his models. But based on what you describe here, even if the spouse nodes have no causal history in common, they have causal effects that converge on the children. So to those children, they’re both causal. If not, then I don’t see how we can say they’re spouses, or related in any way. Crucially, for anything that can be affected by the pattern they form, that effect will be causal. (Unless I’m missing something?)
One of the things I cut from the post (I’m trying to keep these things shorter) was a discussion about the fact that information often has to form “coalitions” of a sort to produce effects. It’s kind of like the inputs to an AND gate. There’s nothing saying they have any shared causal history, but they have causal effects that converge on the gate, resulting in the gate’s output.
You choose to read a blog post, or a book, or take in some other kind of information, largely based on information you already have in your head. If you choose to take action, it will be the result of causal forces from all those sources. If you don’t, it might be the result of patterns from long ago causal sources (i.e. memory).
LikeLiked by 2 people
I’m not saying I disagree with the statement “with pain, the information is the primary part.” But I think you functionalists go way too fast to get to that conclusion. My fool’s gold post touches on this. Most things aren’t “primarily” information, even though they’re not “primarily” raw energy either.
On spouses, yes spouses make a difference to further events and not just each other. But there’s no reason in principle why it couldn’t be spouses all the way down. Or almost paradoxically: it can be cause-and-effect at the macroscopic level, while being spouse-and-spouses at the microscopic level. In fact, on the most elegant theories we’ve got, that’s the way it is.
LikeLiked by 1 person
On going way too fast to a conclusion, all I can say is I’m always open to revisiting it on sufficient evidence or logic.
It seems like if it were spouses all the way down, then they wouldn’t be spouses, just completely independent lines. I can’t see how information could ever be shared between them.
By temporarily ignoring parts of the network of differences that make a difference, you can make a model of the remaining pieces. Then you extend your model if there is a simple and straightforward way of doing so. For example, you can assume that the electrons and quarks you studied are similar to the other electrons and quarks.
So you’re saying information from one line could be used to deduce what’s in the other line? But then the question becomes, what leads to the commonalities that make that possible? If the two lines are in the same universe, the the laws of physics in that universe seem like the ultimate commonality.
If we only have access to one line, how would we even know there was another line?
Let’s take a concrete example from economics. The amounts of government spending, GDP growth, interest rates, unemployment, and inflation are all spouses in a Pearl-style model. Each typically influences each of the others.
But a major war causes government spending to skyrocket independently of its usual economic causes. This is an “intervention”. If this happens in the middle of a recession, unemployment rates typically decline and GDP rebounds. If war happens during a boom time, typically inflation and/or interest rates rise in response. By making these observations, we get a picture of how the govt spending variable influences the others, depending on the current state of these others.
Of course, calling war spending an “intervention” depends on our studying economics, not military history. If you try to study the entire universe all at once at every scale, you might not get very far. But that’s not how we do it. So we’re good.
For microscopic quantum mechanical objects and their spousal relations across time, see Lev Vaidman’s paper on the Two State Vector Formalism. “Spousal” is my word (following Pearl), not Vaidman’s.
The concrete example seems to show the limitations of these types of modeling. Things may be spouses within the scope of what the model is currently looking at, but if we take a wide enough view, eventually they’re siblings. One of the central equations in macroeconomics: MV=PQ, causally ties many of those indicators together.
Thanks for the paper link! That looks interesting.
[Okay then … buckle up. This is where my head has been for the last several months. I’m trying to work out how to explain what follows clearly and succinctly, so any feedback from anyone is much appreciated. I’m still in the weeds, but I can pull you down here with me. Lucky you.]
I like almost none of what you, Mike, just wrote. 🙂 Well, almost none may be harsh. I like that the main reason we feel pain is that we can associate it with other things and use those associations in the future.
But pain is not information. Pain is the recognition of damage. Yes, this is an information process, but pain is the process, not the information.
And information is not causation. Information is very frequently *about* causation, or *due to* causation, but they’re not the same.
So what is information? When most people talk about information, that information is about something, so, semantic information, as opposed to just a quantity of information, which is usually referred to as Shannon information. Semantic information is best understood using the information theory concept of mutual information. From Info. Theory, two variables share mutual information if knowing one tells you something about the other (with some probability above chance). (I’ll get back to this below).
What is causation? Causation is patterns in how physical systems change. When we say this causes that, we’re saying the pattern described by “this” leads, according to the laws of physics, to the pattern described by “that”. I prefer to describe this in my Input->[mechanism]->Output framework. I say that the mechanism “causes” the Output when presented with the Input. I like this description because it lines up nicely with Aristotle’s causes.
So how is information related to causation? Causal processes generate physical systems (Outputs) with mutual information relative to the Inputs and/or mechanisms. Consider, fuel->[fire]->smoke. The smoke has mutual information with respect to the fire (and fuel). (Note that mutual information goes both ways. Seeing smoke means there’s probably a fire, and seeing a fire means there will probably be smoke.)
So I said pain is an information process. What’s an information process? I’m going to answer that by saying an information process is a computation. This will be a very broad concept of computation, but I think using the term is apt. Information theory promises me (according to Seth Lloyd on YouTube) that all computations are a combination of 4 operations: COPY, AND, OR, and NOT. [Actually, I think you can combine AND and OR, because you can replace one with the other, but whatever.] So again, what is an information process? It’s a computation, and the information being processed is the mutual information of the input. In a COPY operation, the mutual information is preserved (minus noise), such that the output has the same mutual information (approx.) as the input. In a NOT operation, the mutual information is inverted, so a correlation becomes an anti-correlation. In an AND or OR, the mutual information is combined.
So what is pain? Pain is a process whereby, thru a long series of computational operations (AND’s , OR’s , NOT’s, COPY’s), a representation vehicle (“sign vehicle” according to C.S.Peirce, “text” according to David Haig) is created that bears mutual information with respect to another physical system (damaged tissue, etc.), and (note, this is a very important “and”, which would be capitalized except that might be confused with the computational AND) that vehicle is the input to a response which gains its value from the fact of the mutual information. Again, the response is a valuable response with respect to the mutual information in the vehicle. Without the response, there is no representation, and no pain.
LikeLiked by 4 people
James of S,
It’s okay if you don’t like much of it. You give your reasons, which is excellent!
With pain, I very much agree that a process is involved, a very complex mulilayered process in fact, beginning in the brainstem, up through various subcortical regions, and into the anterior cingulate. But as you describe with your input->mechanism->output framework, the output can be considered as a thing in itself. (Admittedly this is oversimplifying an enormously complex system.) And it’s the outputs that get broadcast, that is, has causal effects throughout the thalamocortical network. The casual effects of the process are through its outputs.
I think this distinction is important because we can have most of the process in place, such that a person feels the somatosensory or interoceptive sensation, has the low level physiological effects, but due to a lesion in the right brain region, feels no distress about it. In other words, we can have most of the process without the experience itself if the final stages are malfunctioning.
On information being about something, I agree. But the aboutness isn’t inherent to the patterns themselves. They exist in relation to the other entity. That relation comes about due to some shared causal history. (Assuming it isn’t a coincidence, and so not real information.) And the usefulness of that pattern for telling us about the other entity is a causal effect.
But crucially, the information can be used by processes that understand nothing about its aboutness. I mentioned DNA in the post. The proteins that transcribe it don’t know that the DNA in a gene is about protein structures. They just react to what’s there. It’s only us who come in after the fact and recognize the relationship. But it seems perverse to say DNA was not information until we learned about it. Likewise, most of the processing the device you’re using to read this happens without that understanding. Viewing information as causation, concentrated causation, allows us to account for how it’s used by processes that, as Dennett says, are competent without comprehension.
I think we agree on information processing and computation. Neural processing doesn’t provide clean delineations between the primal transformations you list, but they’re all in the mix.
So maybe you can tell me what I’m missing?
[in my best Ricardo Montalban : “He tasks me.”]
Re:pain, you said “we can have most of the process without the experience itself if the final stages are malfunctioning.“ Not sure what you mean. Are you saying this person does not experience pain, or that what they experience is not awful? To me, we don’t have “most of the process”. Instead we have multiple processes, all of the outputs of the signal, whether or not some of them are “awful”. If only the “pain” ones are awful, then we don’t have pain. Again, some of these may result from a broadcast, and some may result from a more direct, local output. I know you are only interested in the ones that result from the broadcast, which is fine. But the broadcasting is not the experience. Doing something in response to the broadcast is the experience. If the response is not awful, it’s not pain. But all of the responses are experiences. Or we can lump the responses together and call that “the” experience. Eye of the beholder.
On information and causality, you’re missing my point, which is the role of mutual information. Mutual information is not inherent in a single physical system, but it is inherent in the relation of one physical system to another. That relation *almost* always comes about due to shared causal history, but not necessarily. The relation is still there in the Boltzmann brain, or in swampman. And it’s important that it’s there, because that’s how natural selection works. A mechanism that responds to, say, a moving dark spot in the visual field, has mutual information relative to flies. But the first such mechanism was generated randomly, by mutation. Nature selected that mechanism, because it could tie a valuable (tongue snatch) response to it. The mutual information was an affordance.
The problem with equating information with causation comes when you use phrases like “concentrated causation”. What the heck does that mean? I can understand concentrated mutual information (created via a series of computations in the sense I outlined), and I can follow the causal (computational) history that gets you there, but that phrase just collides with my (and I suspect most people’s) understanding of causation. And I think viewing information as mutual information allows for competence without comprehension just fine.
[I task you? Considering Kahn was quoting Captain Ahab’s remarks about Moby Dick, I’m wondering if I’m about to get speared. ]
When you talk about the response, are you talking about the pre-conscious responses and assessments? I consider those to be part of the processes that produce the information that enters consciousness as pain. Or are you talking about the possible conscious response? If so, then it seems strange to say pain isn’t pain until we consciously respond to it. Or am I missing your point entirely?
On mutual information, I don’t think I missed your point. I just think there will always be a causal connection.
Your examples of non-causal mutual information are Boltzmann brains and swampman. Along those lines, we could also throw in our doppelgangers in an infinite universe. I have to say that if we have to go to these kinds of examples for counters to the information-causation link, it’s not making that link look weak.
These would be coincidences, profoundly improbable ones, that would require infinite, or near infinite space (doppelgangers), space and time (Bolzmann brains), or just that an infinitesimal probability comes true (swapman). In those cases, the shared causal history amounts to the laws of physics (and whatever produced them). You might call BS on this, but if you’re going to invoke things requiring infinities and profoundly infinitesimal probabilities, I think it’s fair game for me to invoke the ultimate causal commonality. The coincidences can only come about due to the finite constraints imposed by physical laws.
Mutations are chance occurrences, but selection isn’t, and is based on causal factors.
When I said “concentrated causality” I meant something like the idea that the organism, through its distance senses, is able to respond to things in the environment far beyond its immediate somatosensory environment. It’s equivalent to when I say in the consciousness hierarchy that distance senses increase the scope of what the reflexes are reacting to.
I don’t reject your account of mutual information. I just disagree that it’s incompatible with information as causation.
“ If so, then it seems strange to say pain isn’t pain until we consciously respond to it. ”
I’m saying there is no experienced pain until there is a response to the broadcast signal. There will be multiple responses, because broadcast, but if there are no responses, no pain. One of the responses will be the attribution of awfulness. Whether you want to call responses which aren’t “awful” pain is your call. The broadcast signal is not the pain. The broadcast signal carries mutual information with respect to the damage (or whatever), and this mutual information is an affordance to respond. The information is not the pain.
As for “information as causation”, like Wyrd says, it’s up to you. But I think you risk causing more confusion than understanding when using phrases like “concentrated causation”.
[hmmm. My spear seems to have a soft rubber point … ]
Okay, I see the point you were trying to make on the response. Maybe it’ll help if I note that pain seems to involve at least two broadcasts, the sensory sensation, and the assessment. I’ve seen some people say that the sensory part sans the assessment is pain, just not pain the person is distressed over. But I think pain as we commonly understand it requires both.
But your overall point applies to any conscious experience. Under GWT (or global theories in general) it’s not an experience until it has causal effects (“fame in the brain”) throughout the specialty systems. In that sense, the whole thing is a process, but pain is still an important informational input, the same as any sensory / affect combination.
On information and causation, I’m open to having my mind changed. The question is, are there examples of disassociation between the two? The only possibility I’ve thought of so far are the singularities in black holes, and that depends on which theory of black hole information turns out to be right.
But, but, but, I gave an example. The mechanism before it is selected. There it is, with its mutual information with respect to flies, without any significant past causal relation to flies. It might not get selected, in which case it will just go away.
I responded to that example, but admittedly it was cursory. Sorry.
So, a moving dark spot is informative because of its causal relation to the actual fly. The frog’s nervous system didn’t just randomly started showing dark spots that happened to coincide with the passage of a fly. Photons bounce off the fly, then hit photoreceptors on the frog’s retina, leading to an electrochemical cascade to the frog’s optic tectum.
What was random was the initial mutation that caused the frog to snap its tongue at it. There have probably been innumerable mutations causing the frog to snap its tongue at all kinds of things, but only some of them were adaptive. The current reaction was selected for because it was adaptive, and what makes it adaptive are the causal relationships. More than likely, there was a long line of mutations building on each other for the frog’s current capability, but all had to be selected for, and there were probably far more non-functional variations that were neutral or selected against.
So, I’m not seeing the disassociation, but maybe I’m missing something?
The dissociation exists before there is an association. Selection creates the association.
New tack: would you say time is causation? There is no causation without time, and no time without causation.
Hmmm. The only way we have to measure time is by observing the change in very consistent systems (the movement of the sun, Earth’s rotation and orbit, changes in the energy levels of atoms, etc). Those changes seem like they have a very causal relation with time.
Relativity complicates the picture. The rate of time can be affected by energy levels, which seems to mean it not only is a cause, but can in turn be causally affected.
So, I haven’t given this a lot of thought, but on first consideration, it certainly does seem causal.
[warning: if information is causation, and causation is time, information is time. Think that might be harder to justify.]
That can happen when fundamentally different concepts are equated. The semantics that differentiate things gets erased. The hope is that seeing two things as equal leads to insight.
Actually, the more I think about this, the more comfortable I am with the idea that time is information. If we lived in a purely Newtonian universe, I might be leery, thinking that maybe time and space are just a stage for events to unfurl.
But remember that spacetime is a unified concept in general relativity. And we can now detect gravitational waves, literally ripples in spacetime, and use their direction and intensity to deduce where they came from and what it says about the source. It’s a new source of information in astronomy.
I might still see issues with the idea, but right now it doesn’t look obviously wrong. But maybe I’m overlooking some difficulties?
That is important clarification! Info is not just causation, it is the right kind of causation or pertinent causation. It is about some kind of things but not about many others, so you now call it “concentrated info,” a “shared causal history.” Not all diffs make a diff.
Then also, any output can be taken as a thing in itself. That is huge, it can be considered a culmination, an accretion, with its own point of view. So that it’s history is there for it as it’s trail of information, that, yes, becomes info only after the output (us, in our case) happens/is created. That is the relativity of info. DNA was info all along to the proteins it was building. That was competence without comprehension. We only recently figured out its informational role to us. Competence with comprehension, we now know its role and actively manipulate it.
Thanks, that was helpful. That does seem to be Dennett’s position.
LikeLiked by 1 person
Thanks! But one clarification. I do think all causation is information, at least in the physics sense of information, but it’s not all usable information for particular purposes.
Put me down for (information > pain) and (information ≠ causation). I think we probably all agree on the former? For the latter, I have to agree with JoS that they’re disjoint sets. In my view, causality is essentially just the rules of physics + time.
It seems clear information is involved in pain and causality, but as you point out, it’s a very large concept covering a lot of territory.
I do agree pain is an ancient system for teaching us: Don’t do that again!
LikeLiked by 2 people
I’ll admit the equivalence between information and causation, at first glance, seems too simple, too reductionist. But I’ve been mulling this for a while, and I’m having a hard time thinking of examples of one that aren’t an example of the other. The only possibility that occurs to me is one we’ve discussed before, black holes, but even that depends on which theory we favor.
But I’m open to the possibility that this is a failure of imagination on my part. Maybe there is an example of causation that is not informative, or information that isn’t causal, and I just haven’t found it yet.
I agree any causation has information; I’m not sure all information directly involves causation. My sense is that information can be static, whereas I see causation as always dynamic. I can see how one could view both in terms of the other, but that does seem too reductive to me. (I’ll have to ponder the idea of static information; see if I can come up with some good examples.)
Maybe the real question is: What is the value of equating them? What does it buy us?
The value for me is a nice warm glow inside that I’ve dealt with a vexatious hole in my understanding of reality. Given the importance of information in so many domains, it’s always bothered me that I couldn’t find a meaningful definition.
That said, if that definition doesn’t work, I’d rather know about it than live in my warm glow in ignorance.
If it works for you, great!
For me, I suspect that “information” is both too fundamental and too diverse to ever truly define. (All I can do is describe enough cases to build a gestalt.) As with many such fundamental concepts, there aren’t more fundamental concepts I can use when “information” is a concept that applies to so many domains and situations.
(It’s an example of how, in Computer Science, the old joke is that, for any CS question, the answer always begins, “It depends (on what you mean)…”)
You might be right. It is notable that I didn’t really reduce information so much as identify it as equivalent to another fundamental concept.
That joke also exists in philosophy. I wonder if it’s an academy wide meme.
LikeLiked by 1 person
Yeah, I wouldn’t be surprised if it applies to a lot of domains.
LikeLiked by 1 person
“ For me, I suspect that “information” is both too fundamental and too diverse to ever truly define.”
Or, you could use a mathematically established definition, like mutual information, and see where that takes you.
The issue I see is that a mathematically rigorous definition of mutual information depends on a rigorous definition of information. As you mentioned, they are rigorously defined in information theory, which is great if we take this down to that level.
(FWIW: It is, in fact, how I define information. Not the entropy part of it (which, as Mike says, applies mostly to information storage and transmission) so much as the core idea that all information ultimately reduces to bits.)
I’m afraid I don’t see how mutual information opposes the information = causation idea. Your first two comments say almost exactly the same thing if the word “mutual” is entirely removed. I’m not sure you disagree with Mike as much as you at first thought. 🙂
(Again, FWIW: My disagreement rests entirely on the difference between runner and running. One’s a noun; one’s a verb. I just can’t equate them. 😮 )
“ The issue I see is that a mathematically rigorous definition of mutual information depends on a rigorous definition of information.”
This, I think, is not accurate. Mathematically, mutual information is a measure of correlation. How correlated with Y is X? Measuring X tells you something about measuring Y. Mutual information is defined in terms of probabilities.
So again, mutual information is correlation, and apparently paradigmatically, correlation is not causation.
I quite agree. All I’m saying is that we still require a rigorous definition of what’s being correlated. As you said, information theory provides that definition.
Hey Mike! Sorry I haven’t commented as much as I had hoped to at this point.
I like this “why” discussion of subjective consciousness; it’s easier to wrap my head around than the “how” of it! Your thought about pain motivating future action makes sense to me from a self preservation standpoint. I watched a video of Bret Weinstein and Heather Heying discussing their thoughts on the relationship between culture and consciousness. (https://jmp.princeton.edu/heyingweinstein —-> at around 20 minutes they start discussing their definitions). They define consciousness as “The fraction of cognition that is packaged for exchange between individuals”. If I understood the talk correctly, they think that the purpose for subjective consciousness (pain being one such subjective feeling) is to be able to empathize with how someone else feels in order to facilitate social problem solving (not sure how well I summarized their hypothesis…). Their talk also makes sense to me. As social organisms who also have a sense of self preservation, I think your hypothesis and their hypothesis can fit on both sides of the same coin (or maybe two sides of a multi sided dice). 🙂
LikeLiked by 1 person
Good hearing from you! No worries on commenting. A lot of people here only jump in on posts they find interesting.
Consciousness is a hazy and amorphous concept, with lots of competing definitions floating around. I haven’t been able to find any fact of the matter on which is the one true right one, which largely makes it something that exists in the eye of the beholder.
Weinstein and Heying’s version sounds similar to social theories of consciousness. Consciousness is a theory of mind, one that can be turned inward to understand ourselves, as well as outward to understand others. In its broader conception, it applies too all social species, which includes many mammal and bird species. In its narrower versions, it requires language and only applies to humans. The overall view resonates with Michael Graziano’s attention schema theory.
I think it’s worth noting that many non-social species also have self preservation instincts. (Indeed, the self preservation impulses seem to predate consciousness.)
But in the end, I think what most captures most people’s conception of consciousness, is how much like us a system is. Social animals are much more like us than non-social animals, but any perceiving animals is more like us than plants, bacteria, etc.
You may be surprised to find I largely agree with you.
One role of conscious pain might be learning which as some have suggested may be related to consciousness. Reflexes covered the reaction to this touching but it would be better for the organism not to touch it at all. Motivation for the future.
I hope to have my post done on The Sensitive Soul soon and I talk some on causation. There is chapter in The Sensitive Soul on it and they seem to take position similar to mine. Obviously we can always talk about a chain of causation and end up in infinite loops where everything causes everything else. As far as matter goes, I guess we could say the Big Bang is the cause of everything. End of story. For practical purposes, I think it is better to abstract out systems with boundaries and internal dynamics. I think consciousness is such a system.
LikeLiked by 1 person
I thought you might agree on the learning part. Definitely, simply saying “causation” by itself, just as saying “information” by itself, isn’t an explanation of consciousness. Much more is needed, and I definitely didn’t intent this post to be that.
Looking forward to your post!
You might also be surprised to learn that I agree with your assessment that pain is information and that information is causation, but not for the same reasons you see it. It’s the underlying metaphysics that gives “meaning” to a concept, not the idea itself, right? We discussed the importance of hierarchy in our previous conversations, where a correct hierarchy is essential because it gives rise to meaning. Is not meaning the objective of explanatory power, meaning that can be used to make accurate predictions, like; “if I tough that hot stove again I will get burned”?
Now, if one is able to grok JoS’ assessment, pain is a representation such as physical damage, even his explication of the process he actually assigns to the definition of pain is also a representation. But here’s the deal breaker: representations are not physical, which once again begs the question: is information a physical phenomenon or is pain quailia? Now we have reached the intersection of Keith Frankish-land where, any “thing” that is not physical must therefore be is an illusion. Does this mean that illusions have causal power?
I enjoyed this post Mike, but I knew that right out of the gate, it was an intellectual set-up that would quickly go south…
LikeLiked by 1 person
Frankish’s conclusion is that qualia, as traditionally understood, are non-physical, and therefore the entire concept is illusory. Here I agree with Eric Schwitzgebel that that only applies to a theory laden conception of qualia that many philosophers take as inevitable. Along with him, I disagree that those theories are inevitable. If we take a pre-theoretical conception of qualia, then I think they exist, in a manner that doesn’t imply non-physicality.
Along the same lines, I think information is physical. A lot of people seem to think it isn’t. I could see saying it isn’t within a platonic type framework, in which concepts are abstract and inert, but I’m not a platonist. Even if we take platonism as true, actual instantiations of information in this world are always physical.
On things going south, well, it occurred to me that it might be controversial. But so far, the controversy has been surprisingly mild.
What I meant about things going south is that there would be no consensus.
I think its blatantly obvious that material is the carrier of information, information that is limited by our understanding of special relativity. But keeping hierarchy in mind, the greater question is whether material is the information or whether material is merely the carrier of information? That’s a really compelling question.
LikeLiked by 1 person
I liked almost all of what James of Seattle commented 🙂 . I may be out of my depth here, but I believe pain to be closer to “data” than to “information”. Information seems to me to be in the mind of the beholder. It’s what some people are able to make of the data while others can’t or don’t. Both data and information can have causal effects: data can cause reflex behaviors at the very least and information may cause complex abstract behaviors. Data seem to me to be closer to our perceptual senses while information seems closer to our higher-order cognitive processes. Sometimes information is just information; that is, it doesn’t necessarily cause us to act; for example, we may decide. based on the information, that there is no need to act. Just my own humble opinion.
LikeLiked by 1 person
Mike, your comment reminds me of something we had to learn in the ITIL standards a few years ago, called the DIKW pyramid. The idea is that at the bottom of the pyramid, you have data, then higher up, information, then knowledge, then at the top, wisdom. So your idea that information is something more than just the data is a common one. Part of the problem is that our use of these words aren’t always consistent. Physicists often use the word “information” in a primal manner that’s probably closer to the DIKW’s meaning of “data”.
Certainly we may not decide to act on certain information. But I don’t think that means the information isn’t causal. It just means that, by itself, it’s not sufficient to cause action. If we had certain information in our heads already, it might have changed the balance and led to us acting.
Above, I used the example of a logic gate, such as an AND gate. For an AND gate to output a 1 or true value, both of its inputs must be 1 or true. If either is 0 or false, the gate will output 0 / false. Arguably, both inputs are information (you might argue data), and I think both are causal forces, but they can only drive a particular result if they are both on. That’s the way I see externally received information in relation to information (knowledge) already in our head. It may take both to be in certain states to require us to act, but if only one is present, it was still a causal force, just not a sufficient one.
LikeLiked by 1 person
Thanks, Mike. I liked your answer. Yes, I know about AND, OR, and NOT gates. So, as I understand you, all information (and cognitive?) processes are essentially computational. Although our higher-level software assemblers, compilers, and interpreters may have customized instructions and/or operations for different applications or apps, they all run on biological AND, OR, and NOT gates. All information is causal because it is input, processed (or switched through as-is), and output. Did I over-simplify or did I get the gist of it?
LikeLiked by 1 person
Thanks Mike. You got the gist of it. My view is computationalist. Before I get into too much trouble, let me hasten to add that neural computation is very different from the digital computation in our devices. Neurons operate according to the same principles as logic gates, but they’re more akin to several thousand logic gates, and have continuous rather than discrete dynamics, which adds a stochastic nature to their operations, so we have to be careful with comparisons.
So which use of “information” is closest to yours – that of physicists, DIKW standards, Shannon information theorists, or some other well known standard?
Oops, hadn’t yet read your response to Philosopher Eric. So, physical information. That fits well.
LikeLiked by 1 person
Pain is a signal, like you say, so in that sense it is communication that is not noise, i.e. meaningful information in flight. What is also interesting is that the experience of this signal can be controlled somewhat, blending it back into noise. So the experience is optional, with the right medication for example.
Weird as it may seem today, not so many decades ago corporal punishment was considered as a vital element of education, as a way to make powerful impressions in the empty table minds of young people, without visibly damaging their bodies. This is not in fashion anymore, modern world expects everyone to carry their own carrots and sticks with them, to be self-motivated in other words.
I am not saying this is the evolutionary cause of pain, but certainly many social animals have been observed inflicting pain for reasons that relate to the group. Even if the origin of the signal is in the sensory feedback, it can be adapted for social feedback as well, with some theory of mind.
LikeLiked by 1 person
Good points. A lot of traditional societal remedies are painful, and we can see their evolutionary origins in the way troops of monkey’s treat anti-social behavior. But they evolved because they work, albeit often not fairly.
I’m reading Penrose’s Cycles of Time and it prompted two questions for you: What do you see as the information-causality link regarding the information contained in the CMB? What is your view of information outside our causal light cone? (Not making any points here; just curious about the view equating information and causality.)
Hmmm. I don’t perceive that the CMB challenges the association. We’ve used it to deduce things about the early universe, and we can do so because it’s a relatively pure causal remnant of that early universe.
The light cone question is interesting. Due to the expansion of the universe, we can receive information from outside of our current light cone, although our ability to have causal effects are limited to that cone. The difference is more striking than I realized when I looked into this a while back. Sans some kind of FTL technology, 97% of the observable universe is forever beyond our reach.
Back before we found out that the BICEP2 results were dust (literally), it seemed like, due to cosmic inflation, we might be able to use that information to deduce things about vast stretches of the universe beyond our current horizon.
But it seems like in all cases, our ability to acquire information is at the same scope as our ability to be causally affected. Unless I’m wrong. 🙂
The CMB wasn’t so much a challenge as a probe into whether the view focuses on information as causally produced or that information causes something. With the CMB I take you to be pointing to that information causes something. (That information is caused seems trivially true.)
Your light cone response seems to confirm that? That information has causal power?
I can certainly see the basis for equating the two, but I’ve also figured out what I don’t care for it myself. I see it as conflating what I perceive as fundamentally a noun and a verb. It’s like conflating the runner with the running. There is definitely a linkage; can’t have one without the other. I just don’t see that as an equality though. 🤷🏼♂️
It strikes me as being one of those matters of interpretation and definition, so it’s totally fair game.
I can see that view. That distinction was probably what kept the complexity in my previous definition.
The language doesn’t help. You have “cause”, which can be used as a noun or a verb. What then does “causation” refer to exactly? And you have “inform”, a verb, and “information”, which seems like a noun referring to something that does what the verb signifies. A big muddle.
LikeLiked by 1 person
Ha! For me the verb/noun distinction removes the ambiguity. “Cause” and “causation” both refer to a process. We have to go back to the Latin stem before language equates “inform” and “information” because there is that strong verb/noun distinction.
That said, they do stem from the same concept, and we can agree they go hand-in-hand. (Nuf sed! 😉 )
LikeLiked by 1 person
Nice one Mike. Agreed on the planning value of pain.
I’d also say “Pain is a request for a second opinion.” That is, the first “opinion” is a reflexive one, an open-and-shut case summarily dismissed. Pain moves beyond the reflex loop and brings state information up through a hierarchy of cognitive processing, where it can be processed in a broader context; e.g. “Should I continue running even though my leg is hurting?” The answer to that question depends upon context (“Am I in danger here?” or “How important is it to finish this marathon?”), a context which can only be provided by the full power of our cognitive system.
LikeLiked by 1 person
That’s a good way of describing it. I actually think a good view of consciousness overall is as providing a second opinion, one more considered and reasoned than the initial first opinion. To do this, the higher order systems have refined object recognition, episodic memory, top down attention, learning, and a lot of the stuff we typically label as cognitive. It’s a system that takes more time, providing Kahneman’s System 2 thinking. And it decides what in the System 1 reactions to allow on inhibit.
It seems to me that as the term is commonly used, “information” can both be provided to a conscious entity, as well as animate the function of a given non-conscious machine. I’ll quickly display each, as well as suggest that the two shouldn’t be conflated.
If I feel a strong pain associated with the need to urinate, this will tend to “inform” me about the circumstance. Though I wouldn’t quite say that pain “exists as” information (given the conflation issue that I’m about to address), I certainly do say that I’m informed by pain.
Beyond information provided to the conscious entity, or the term’s traditional form, today it’s commonly also used in reference to non-conscious machines. A television picture may be animated by means of television signals, for example, and we’ve come to refer to such signals themselves as “information”. Note that typing provides information to your computer, and it may process that information to then provide a different set of information which helps animate a computer screen.
Regarding the conflation of the two, given that we’ve come to refer to stuff which animates our machines to exist as “information itself”, and that things like pain will inform the conscious entity in a traditional sense, it’s understandable that some today have decided that qualia probably exist as information beyond any specific mechanism — only the pattern matters here.
Beyond conflating the conscious and non-conscious forms of the word, consider a lesson that exists even from the non-conscious variety. I don’t know of anything produced by our machines which is said to exist as information alone. In all cases that I know of, specific mechanisms must be animated in order for information to help produce any output function.
Apparently some prominent scientists and philosophers have become heavily invested in qualia existing without mechanism based instantiation, however. (I’d forgotten that professor Schwitzgebel referred to information as “causality itself”, which seems like a tautological mess.) If the thus “non problem” of qualia is explained away by dismissing any need for specific mechanical instantiation, note that this frees theorists in all sorts of otherwise sci-fi ways. Few seem to have acknowledge there to be a naturalistic downside to this convention however. Wouldn’t this mean that the right set of information laden paper which is converted into another, should create what we experience when our thumbs get whacked? Strange! So it could be that causal dynamics “of this world” instead depend upon information animating the right kinds of mechanisms.
Should humanity ever grasp why any of the four forces exist? Even if further progress does happen to be made, such speculation should always end in “…because”. The naturalist however will presume that they’re mechanism based, like all else. And what mechanism might the processed information associated with brain information be animating to produce qualia? My money is on certain electromagnetic radiation produced by neurons. (And it could be that some of the popular “information only” qualia theories in neuroscience would survive such a paradigm shift anyway. Some may effectively describe what it takes to animate such mechanisms.)
LikeLiked by 1 person
So, on the two types of information, let’s give each of them a name. The more primal one, the one I equate with causation, represents information in the sense typically used by physicists, so let’s call it physical information, or p-info. The more restricted version involving usage by conscious minds, let’s call that conscious information, or c-info.
So, the question is, why do we use “information” to refer to both of these? I think the answer is that when we discover p-info and learn about its role, it becomes c-info. We refer to DNA as information because we recognize its role in transmission of the recipe for replicating biological organisms. But before we discovered it, it operated for billions of years as p-info. Once we discovered it, it also became c-info.
All c-info is also p-info. (If you can identify cases where this isn’t true, I’m very interested.) As far as I can tell, all p-info has the potential to be c-info, at least in principle, though there may be cases where it will never be practical.
So, while we can talk about these as different types of information, the use of the word “information” for both strikes me as both rational and coherent.
On the qualia argument, aren’t you the one always arguing that we should accept the other person’s definitions? If so, what objection remains for qualia being p-info, or at least processing of p-info? Wouldn’t any mechanism inevitably involve p-info? And once we understand those mechanisms, wouldn’t they become c-info? If not, why not?
I think the argument for electromagnetic fields playing a major role is weak, but even if they did, it seems it would be just another information processing mechanism.
LikeLiked by 1 person
I’ve posted a response. https://physicalethics.wordpress.com/2020/05/30/qualia-as-information-processing-and-philosophical-thought-experiments/
LikeLiked by 1 person
If ‘pain’ is information then it is linguistic information, and only that.
LikeLiked by 1 person
Not sure if I follow you here. Are you saying that referring to it as information is only a linguistic move? Or that pain is only linguistic information and no other kind?
Hi, I’m suggesting that we can only refer to PAIN through language, specifically the word ‘pain’. There is something physical going on underneath: signals sent via the nervous system to the brain and neurons firing. But this is all physical.
Our conscious experience of this is through language, specifically the word ‘pain’. We can try and explain it and define it using others words but we just get more language. At the end of the day, all our conscious experiences are in language. There is no other way to bring them to mind.
I agree that pain is physical, but so is information, so I don’t see those statements as necessarily incompatible.
But if you see consciousness as dependent on language (many do) then I can see where you’re coming from. I think language and its supporting capabilities are a major part of human consciousness. But I think there is a pre-linguistic consciousness.
That said, I’m also aware of Helen Keller’s story and her judgment about her pre-language existence, but it’s tangled up with her sensory disabilities. Ultimately, consciousness is in the eye of the beholder.
Total aside because I know y’all are interested in brain stuff: Read an article in Psychology Today (IIRC) about using fMRIs to train a DLNN. The fMRIs were of people they’d trained to tie seven (or was it five?) different knots. Those folks were both tying and thinking about tying those knots.
Apparently, on new unlabelled inputs, the NN had 100% accuracy in determining which knot people were either tying or thinking about tying. The article mentions this seems to support high-level theories due to the identification of specific groups of neurons involved in each knot.
LikeLiked by 1 person
Thanks. Sounds like something I definitely need to dig up. I know neural networks are being used to analyze that kind of data, but didn’t realize they were getting that good.
LikeLiked by 1 person
What happens when you bring in a new person and fMRI them, does the NN get them right too? I remember some similar studies of (I think it was) visual cortex activity that resulted in person-specific models.
I can’t recall the article’s language specifically enough to answer your question for sure, but the excitement involved the identification of groups of neurons involved in tying specific knots, so there certainly could be individual specificity.
That said, the whole deal with NNs is they aggregate lots of data from different sources into a configuration space that allows highly successful identification of new inputs as in or out of that space. It’s possible this experiment applied to new individuals who’d been trained on the knots, scanned, and those scans presented as unknown inputs. If so, that is pretty exciting.
An always interesting question is whether a “seeming” level of pain sensation reliably correlates with a verifiable level of physical trauma. I suppose not. A tooth-ache often seems to bother me much more than it should. Dennett is good on this. Even more so, Norton Nelkin’s 1986 article in, “The Journal of Philosophy”, entitled, “Pain and Pain Sensations”.
LikeLiked by 1 person
Obviously pain evolved to let us know about actual damage, but like anything, it can malfunction. We can have more distress than maybe the injury warrants, or outright phantom pain. But there are people who, either through injury or pathology, have lost the ability to feel pain, or who can feel the sensation but without distress. That might sound like a good thing, but their life expectancy is reportedly not good.
Two things that jump out at me.
People with Congenital Insensitivity to Pain don’t avoid damage. They often walk around with broken bones and tongues bitten off. So pain isn’t only about the future. Wouldn’t your proposed focus on the future rule out the pain of simpler animals (phylogenetically or ontogenetically) who arguably have no future selves?
Second, is there not information which causes nothing? If my philosophizing never convinces anyone of anything is it no longer information? That’s a joke example, you might say, because it caused something in me, but surely there is information that doesn’t cause anything. What does the information that you and I are 45th cousins 32 times removed actually cause in the world? I had to make those numbers up because we don’t and possibly can’t know the reality, but the information exists. Even more inconsequential information exists, like Dan Dennett’s “current average location of all his missing socks.” What does that cause? (Other than a chuckle when I describe the information without providing it.)
LikeLiked by 1 person
I think we have to make a distinction between the experience of pain, and nociception. If someone has nociception, the transmission of signals through the peripheral nervous system to the spinal cord and lower level brain networks, then they’ll have many automatic and habitual responses, such as stopping a bite if it’s cutting into the tongue. Unfortunately, there are people whose nociception is dysfunctional, which may cause them to keep biting into their tongue. And a lot of what we call “pain” in simple animals is actually just nociception.
That said, I don’t know if you noticed, but I did leave myself some wiggle room in the language for some near real time processing to happen. Some might say I was being weaselly, but this is biology and I think we have to acknowledge that things are rarely cut and dry. And the future behavior being affected might only be a second or two from the present. For example, I would consider walking around with broken bones to be a case of their future behavior not being sufficiently modified.
On information which causes nothing, consider an AND logic gate. It has two inputs. If both are signalling 1, the gate will signal 1, but if either is signalling 0 instead, the gate will signal 0. Each input is information. Each is also a causal force (made more evident if we looked at this at the transistor level). But neither by themselves are a sufficient causal force to cause the output to be 1. They both must be signalling 1 for that to happen. This is a very simplified version of what happens in a neuron, which has thousands of inputs, each signifying various concepts. Individually, most won’t have enough causal power to cause the neuron to fire, but they’re still a causal force.
Backing this up to the real world, if I see a headline of some pop star ODing, it probably won’t have much of an effect on my mental state, but it might on a teenage girl’s. Likewise, your philosophy is much more likely to have an effect on me and my future behavior than someone utterly uninterested in philosophy. The external information by itself is insufficient to induce change, but coupled with internal information, it can be possible.
In another conversation above, I made distinction between physical information and conscious information. We use the word “information” for both because physical information can always, in principle, become conscious information. But in principle is not always in practice. 45th cousins and average location of missing socks is physical information that can have minor causal effects, but not ones we could ever measure in any practical sense.