Great topic. I would enjoy seeing this as a top level post.
So what is the answer that would satisfy the child?
For an adult, saying “that was a crow, and crows sometimes caw” seems like a fairly complete answer because we already possess a lot of contextual information about animals and why they make noises.
A lot of the lower-level ‘explanation’ you described above was really about how the crow cawed (lungs, vibrations, etc.) With this information, you could build a mechanical crow that did in fact crow—but the mechanism for how the mechanical crow caws would have nothing to do with why the organic crow caws.
The real explanation for the crow’s caw is in the biology of the crow. The crow caws to communicate something to other crows, because it is part of a social group.
As reductionists, we all accept that biology would be reducible to interacting layers of complicated physics, but this example about the crow really gives us a concrete example to see that it may not be immediately straight-forward how reductionism is supposed to work. No amount of detail regarding how the crow caws is going to get at why it does, because we can build the mechanical crow that caws for no reason. On the other hand, we can make a little mechanical gadget that beeps for the same reason that the crow caws—to let other gadgets know it’s around or needs something. The gadget and the crow don’t have to have much of anything in common material-wise in order to have the same ‘why’. What they do have in common is something more abstract, a type of relational identity as a member of a group that exchanges information.
Later edit: I think we could inflate ‘physics’ to include this type of information, because physics has mathematics (and algebra). So we’ll be able to define things that really depend on relationships and interactions, rather than actual material properties, but I wonder to some extent if this is what was envisioned as going to be eliminated by reductionism?
For an adult, saying “that was a crow, and crows sometimes caw” seems like a fairly complete answer because we already possess a lot of contextual information about animals and why they make noises.
Ok, but what is the form that contextual information takes? I’m skeptical that most adults actually have a well-formed set of beliefs about the causes or biophysics of animal behavior. I think my mind includes a function that tells me what sorts of things are acceptable hypotheses about animal behavior and I have on hand a few particular facts about why particular animals do particular things. But I don’t think anyone can actually proceed along the different levels of abstraction I outlined. I’m worried that what actually makes us think “there is a crow and crows caw” is that it just connects the observation to something we’re familiar with. We’re used to animals and the things that they do and probably have a few norms that guide our expectations with animals. But at rarely if ever are we reducing or explaining things in terms of concepts we already already fully understand. Rather, we we just render familiar the unfamiliar. Think of explanations as translations, for example. If I say “a plude is what a voom does” no one will have any idea what I mean. But if I tell you that a plude is a mind and a voom is a brain suddenly people will think they understand what I mean even if they don’t actually know what a mind is or anything about a brain.
A lot of the lower-level ‘explanation’ you described above was really about how the crow cawed (lungs, vibrations, etc.) With this information, you could build a mechanical crow that did in fact crow—but the mechanism for how the mechanical crow caws would have nothing to do with why the organic crow caws.
I wonder if a taxonomy of explanations would be worth while. We have physical reduction, token history explanation (how that crow got outside my window and what lead it to caw at that moment), type historical explanation (the evolutionary explanation for why crows caw)… I’m sure we can come up with more. Note though that any historical explanation is going to be incomplete in the way explain above because it will appeal to concepts and entities that need to be reduced. Anyway, the “explanation” in my above comment was never for “why do crows caw” but why do I hear cawing. And I posited that there was a crow and that crows caw. These assumptions are sufficient to predict the possibility of cawing. But even though they are predictively sufficient they don’t actually explain what happened. It is also true that the assumptions themselves are unexplained (as you say, we need biology to tell us why crows caw) but even if we knew all this we still wouldn’t have explained the cawing.
This example has the unfortunate quality of being true. But say we posit that astrological configurations control the movement of cows. Obviously this isn’t true, but pretend that it is. Pretend the movement of the planets could predict, with 100% confidence, where and how cows moved in a given field. Say I prove this to everyone tomorrow. Would anyone here be satisfied with “Planets control cows” as a fundamental law of nature? Would this explain anything? I take it a lot of people would set about trying to figure out how and why the planets affect cows and we’d look at magnetic and gravitational fields and find which bovine organs were most sensitive to such fields etc. But these explanations would be similarly unsatisfying and we would look deeper.
The problem is every individual explanation is just like the “Planets control cows” explanation. So we have to explain what is special about fundamental laws or why this doesn’t matter.
The real explanation for the crow’s caw is in the biology of the crow.
Why is this the privileged “real explanation”? For example, the real explanation is that evolution produces complex social assemblies in need of signaling mechanisms. Or the real explanation is that an asteroid or comet disrupted the previous biological configuration, allowing crow-like birds to evolve. Etc...
As reductionists, we all accept that biology would be reducible to interacting layers of complicated physics, but this example about the crow really gives us a concrete example to see that it may not be immediately straight-forward how reductionism is supposed to work.
We don’t expect that reductionist approaches have the magical potential to successfully answer all possible questions. It’s possible (and necessary) for information to be irretrievably lost. So how it’s supposed to work is actually straight-forward: seek evidence that distinguishes between different causal hypotheses for the crow’s caw.
Depending on what you’re looking for, there may be no meaningful explanation, as in the case of chaotic systems. For example, the most concise explanation for a particular cloud being of a particular shape may just be the entire mountain of data comprising the positions and velocities of the air and water particles involved.
No amount of detail regarding how the crow caws is going to get at why it does
I think this is an overstatement. The only hard upper bound we have on how much information might be contained in a crow’s vocal system is the number of possible states in the physical system comprising it, which is huge. It even seems conceivable that significant portions of a crow’s DNA might be reconstructible from a detailed enough understanding of its vocal system.
because we can build the mechanical crow that caws for no reason
I’d say the mechanical crow caws because it was built that way. Then you’re faced with the question of how something can possibly be built for no reason.
Later edit: I think we could inflate ‘physics’ to include this type of information, because physics has mathematics (and algebra).
But “this type of information” is stated in terms of physically observable phenomena. We can reason logically and mathematically about the things we observe without new physics, as long as the observations themselves have believable reductions to known physics. I don’t see what you’re looking for that isn’t captured by a reductionist model of the crows, their communication mechanisms, their brains, and their evolutionary history.
No amount of detail regarding how the crow caws is going to get at why it does
I think this is an overstatement. The only hard upper bound we have on how much information might be contained in a crow’s vocal system is the number of possible states in the physical system comprising it, which is huge. It even seems conceivable that significant portions of a crow’s DNA might be reconstructible from a detailed enough understanding of its vocal system.
I’m certain there is not enough information in how the crow caws. For example, there is not enough information even in the DNA; if the base pairs could be deduced from it’s caw (which I would guess is impossible), because the full explanation will involve it’s environment and the other crows. (For example, if you cloned a horse in a sterile laboratory, you wouldn’t know why it swished it’s tail without also cloning a fly.)
We know there is enough information in the whole universe. The crow, its environment and its entire evolution history do explain everything about it.
So our different answers to the ‘why’ a crow caws are different ways and angles of summarixing the limited story that we know about the whole universe. I agree with Jack that it would be useful to have a ‘good’ classification of these answers. However, it’s not a project I would be interested in following, generally, because the quality of the outcome is too subjective. I would enjoy reading the classification of someone who thought the way I did, and would find it frustrating to read that of someone who thought differently, with no tools to distinguish ‘quality’ beyond this feeling of accord or frustration.
Why is this the privileged “real explanation”?
Exactly. At least, we may agree that the crow did caw.
I’m certain there is not enough information in how the crow caws. For example, there is not enough information even in the DNA; if the base pairs could be deduced from it’s caw (which I would guess is impossible), because the full explanation will involve it’s environment and the other crows.
I’d agree that there probably isn’t enough information, but I think your certainty is misplaced. I’m guessing the crow’s DNA contains quite a lot of information about its environment and social habits.
For example, if you cloned a horse in a sterile laboratory, you wouldn’t know why it swished it’s tail without also cloning a fly.
I have yet to be convinced that a Bayesian superintelligence couldn’t infer the existence of fly-like organisms from a horse’s DNA.
I’d agree that there probably isn’t enough information, but I think your certainty is misplaced. I’m guessing the crow’s DNA contains quite a lot of information about its environment and social habits.
Actually, it seems we agree. I’d agree that there could be enough information in the horse DNA to deduce many salient features about the fly. In fact, I might even put a higher probability on the information being in there somewhere than you would. But I thought we were trying to determine where such information is coded … in other words, how large a swathe of information would you need to guarantee that you have enough?
But I see the conversation has drifted over time.
What I was saying at the beginning, which I believe you disagreed with, was that the answer was mathematical in some way (algebraic, actually, because my favored answer to the ‘why’ was about relationships among the crows rather than about the materials the crow is made of) while you were pressing it should still be answered in the physicality of the universe:
I don’t see what you’re looking for that isn’t captured by a reductionist model of the crows, their communication mechanisms, their brains, and their evolutionary history.
So by now I’ve now changed my view. I agree with you that all the answers do ultimately lie in the materials: the crows and their material environment. At the time of my first post, I had preferred to answer that the crow had a “purpose” (to speak with other crows) but of course this is a story which would actually reduce to a bunch of statistics over time that crows had better fitness when they communicated in effective ways.
I have yet to be convinced that a Bayesian superintelligence couldn’t infer the existence of fly-like organisms from a horse’s DNA.
Well, sure. A Bayesian superintelligence would probably guess that a crow caws to communicate with other crows even without the crow’s DNA. There’s a lot of similarity and pattern in the universe, and you can infer much by analogy. What we’re debating, however, isn’t what a superpower might be able to infer but where the information is coded for why the crow caws.
Perhaps the universe is deterministic and everything can be deduced by a superintelligence from the periodic table of the elements and the number of pigeons born in Maine on Sunday. Only in this sense would the DNA of the crow contain information about the caw and the DNA of the horse contain information about the fly.
This is why I am so confident: The DNA base pairs of life are random, except for the fact that they need to code information that leads to better fitness. Yet coding information provided by the environment itself would be redundant information-wise. So while the information could be there, by accident, there’s no reason that it would be there necessarily.
I imagine that if efficient fly-swatting leads to some genetic advantage, then one might deduce the size and weight of a fly from the length and motion dynamics of the tail. That would be neat. But unlikely, because what are the chances that the tail is so tuned? Why should the information necessarily be there?
Great topic. I would enjoy seeing this as a top level post.
So what is the answer that would satisfy the child?
For an adult, saying “that was a crow, and crows sometimes caw” seems like a fairly complete answer because we already possess a lot of contextual information about animals and why they make noises.
A lot of the lower-level ‘explanation’ you described above was really about how the crow cawed (lungs, vibrations, etc.) With this information, you could build a mechanical crow that did in fact crow—but the mechanism for how the mechanical crow caws would have nothing to do with why the organic crow caws.
The real explanation for the crow’s caw is in the biology of the crow. The crow caws to communicate something to other crows, because it is part of a social group.
As reductionists, we all accept that biology would be reducible to interacting layers of complicated physics, but this example about the crow really gives us a concrete example to see that it may not be immediately straight-forward how reductionism is supposed to work. No amount of detail regarding how the crow caws is going to get at why it does, because we can build the mechanical crow that caws for no reason. On the other hand, we can make a little mechanical gadget that beeps for the same reason that the crow caws—to let other gadgets know it’s around or needs something. The gadget and the crow don’t have to have much of anything in common material-wise in order to have the same ‘why’. What they do have in common is something more abstract, a type of relational identity as a member of a group that exchanges information.
Later edit: I think we could inflate ‘physics’ to include this type of information, because physics has mathematics (and algebra). So we’ll be able to define things that really depend on relationships and interactions, rather than actual material properties, but I wonder to some extent if this is what was envisioned as going to be eliminated by reductionism?
Ok, but what is the form that contextual information takes? I’m skeptical that most adults actually have a well-formed set of beliefs about the causes or biophysics of animal behavior. I think my mind includes a function that tells me what sorts of things are acceptable hypotheses about animal behavior and I have on hand a few particular facts about why particular animals do particular things. But I don’t think anyone can actually proceed along the different levels of abstraction I outlined. I’m worried that what actually makes us think “there is a crow and crows caw” is that it just connects the observation to something we’re familiar with. We’re used to animals and the things that they do and probably have a few norms that guide our expectations with animals. But at rarely if ever are we reducing or explaining things in terms of concepts we already already fully understand. Rather, we we just render familiar the unfamiliar. Think of explanations as translations, for example. If I say “a plude is what a voom does” no one will have any idea what I mean. But if I tell you that a plude is a mind and a voom is a brain suddenly people will think they understand what I mean even if they don’t actually know what a mind is or anything about a brain.
I wonder if a taxonomy of explanations would be worth while. We have physical reduction, token history explanation (how that crow got outside my window and what lead it to caw at that moment), type historical explanation (the evolutionary explanation for why crows caw)… I’m sure we can come up with more. Note though that any historical explanation is going to be incomplete in the way explain above because it will appeal to concepts and entities that need to be reduced. Anyway, the “explanation” in my above comment was never for “why do crows caw” but why do I hear cawing. And I posited that there was a crow and that crows caw. These assumptions are sufficient to predict the possibility of cawing. But even though they are predictively sufficient they don’t actually explain what happened. It is also true that the assumptions themselves are unexplained (as you say, we need biology to tell us why crows caw) but even if we knew all this we still wouldn’t have explained the cawing.
This example has the unfortunate quality of being true. But say we posit that astrological configurations control the movement of cows. Obviously this isn’t true, but pretend that it is. Pretend the movement of the planets could predict, with 100% confidence, where and how cows moved in a given field. Say I prove this to everyone tomorrow. Would anyone here be satisfied with “Planets control cows” as a fundamental law of nature? Would this explain anything? I take it a lot of people would set about trying to figure out how and why the planets affect cows and we’d look at magnetic and gravitational fields and find which bovine organs were most sensitive to such fields etc. But these explanations would be similarly unsatisfying and we would look deeper.
The problem is every individual explanation is just like the “Planets control cows” explanation. So we have to explain what is special about fundamental laws or why this doesn’t matter.
Why is this the privileged “real explanation”? For example, the real explanation is that evolution produces complex social assemblies in need of signaling mechanisms. Or the real explanation is that an asteroid or comet disrupted the previous biological configuration, allowing crow-like birds to evolve. Etc...
We don’t expect that reductionist approaches have the magical potential to successfully answer all possible questions. It’s possible (and necessary) for information to be irretrievably lost. So how it’s supposed to work is actually straight-forward: seek evidence that distinguishes between different causal hypotheses for the crow’s caw.
Depending on what you’re looking for, there may be no meaningful explanation, as in the case of chaotic systems. For example, the most concise explanation for a particular cloud being of a particular shape may just be the entire mountain of data comprising the positions and velocities of the air and water particles involved.
I think this is an overstatement. The only hard upper bound we have on how much information might be contained in a crow’s vocal system is the number of possible states in the physical system comprising it, which is huge. It even seems conceivable that significant portions of a crow’s DNA might be reconstructible from a detailed enough understanding of its vocal system.
I’d say the mechanical crow caws because it was built that way. Then you’re faced with the question of how something can possibly be built for no reason.
But “this type of information” is stated in terms of physically observable phenomena. We can reason logically and mathematically about the things we observe without new physics, as long as the observations themselves have believable reductions to known physics. I don’t see what you’re looking for that isn’t captured by a reductionist model of the crows, their communication mechanisms, their brains, and their evolutionary history.
I’m certain there is not enough information in how the crow caws. For example, there is not enough information even in the DNA; if the base pairs could be deduced from it’s caw (which I would guess is impossible), because the full explanation will involve it’s environment and the other crows. (For example, if you cloned a horse in a sterile laboratory, you wouldn’t know why it swished it’s tail without also cloning a fly.)
We know there is enough information in the whole universe. The crow, its environment and its entire evolution history do explain everything about it.
So our different answers to the ‘why’ a crow caws are different ways and angles of summarixing the limited story that we know about the whole universe. I agree with Jack that it would be useful to have a ‘good’ classification of these answers. However, it’s not a project I would be interested in following, generally, because the quality of the outcome is too subjective. I would enjoy reading the classification of someone who thought the way I did, and would find it frustrating to read that of someone who thought differently, with no tools to distinguish ‘quality’ beyond this feeling of accord or frustration.
Exactly. At least, we may agree that the crow did caw.
I’d agree that there probably isn’t enough information, but I think your certainty is misplaced. I’m guessing the crow’s DNA contains quite a lot of information about its environment and social habits.
I have yet to be convinced that a Bayesian superintelligence couldn’t infer the existence of fly-like organisms from a horse’s DNA.
Actually, it seems we agree. I’d agree that there could be enough information in the horse DNA to deduce many salient features about the fly. In fact, I might even put a higher probability on the information being in there somewhere than you would. But I thought we were trying to determine where such information is coded … in other words, how large a swathe of information would you need to guarantee that you have enough?
But I see the conversation has drifted over time.
What I was saying at the beginning, which I believe you disagreed with, was that the answer was mathematical in some way (algebraic, actually, because my favored answer to the ‘why’ was about relationships among the crows rather than about the materials the crow is made of) while you were pressing it should still be answered in the physicality of the universe:
So by now I’ve now changed my view. I agree with you that all the answers do ultimately lie in the materials: the crows and their material environment. At the time of my first post, I had preferred to answer that the crow had a “purpose” (to speak with other crows) but of course this is a story which would actually reduce to a bunch of statistics over time that crows had better fitness when they communicated in effective ways.
Well, sure. A Bayesian superintelligence would probably guess that a crow caws to communicate with other crows even without the crow’s DNA. There’s a lot of similarity and pattern in the universe, and you can infer much by analogy. What we’re debating, however, isn’t what a superpower might be able to infer but where the information is coded for why the crow caws.
Perhaps the universe is deterministic and everything can be deduced by a superintelligence from the periodic table of the elements and the number of pigeons born in Maine on Sunday. Only in this sense would the DNA of the crow contain information about the caw and the DNA of the horse contain information about the fly.
This is why I am so confident: The DNA base pairs of life are random, except for the fact that they need to code information that leads to better fitness. Yet coding information provided by the environment itself would be redundant information-wise. So while the information could be there, by accident, there’s no reason that it would be there necessarily.
I imagine that if efficient fly-swatting leads to some genetic advantage, then one might deduce the size and weight of a fly from the length and motion dynamics of the tail. That would be neat. But unlikely, because what are the chances that the tail is so tuned? Why should the information necessarily be there?