Philosophers and activists argue that fish and shrimp experience enormous amounts of agonizing pain, but the science behind their claims is far from convincing.
I have a bunch of things to say about the subject, but let me note just a few:
1) You should be highly uncertain about the subject. When lots of smart people disagree on a subject that you're not an expert on, even if you've done a decent amount of reading about it, you should basically never be more than 80% confident, especially if your view is the minority view. But the case for shrimp welfare goes through even if you're 95% sure shrimp aren't conscious, just because of the insane degree of the carnage and effectiveness.
2) I think we're largely in the dark about what ingredients are needed for consciousness. If this is right, then Key's speculation by analogy about the ingredients needed not being present in fish is highly improbable. Behavior is all we have to go off of. And shrimp behave in lots of ways that we'd expect them to if they felt pain (see my piece).
3) The sorts of things slime molds do are very rudimentary compared to what shrimp do--relevantly like a computer or mouse trap with many settings rather than a conscious being.
4) I do think you misrepresented me a bit. I'm now super confident in shrimp pain--I'd put it in the low 60s.
5) I couldn't find a source for the claim that shrimp eat their own body parts. I couldn't even find people using the word autophagy to mean eating one's own body parts https://en.wikipedia.org/wiki/Autophagy
6) The Key view implies octopi aren't conscious, which I take to be a decisive counterexample.
To 1), I'll restate what I said to Glenn above. The principal disagreements in this debate are not over complicated technical questions in neuroscience that you would need a PhD in the area to evaluate. They're fundamentally philosophical disagreements over questions like: Are observations of behavior or neurobiology the best way to determine whether an animal is conscious? And I believe a reasonably informed observer can pass judgment on these questions.
2) I fundamentally disagree with the idea that consciousness is still very mysterious and that we haven't (and possibly can't ever) make major progress in understanding consciousness. Ultimately, it's going to be explained by simpler biological processes and Key has provided an enormous body of empirical evidence supporting his theory that signal amplification (for example) is a critical process supporting consciousness. (His ideas also make intuitive sense - your body has billions of sense receptors and for any one of those signals to be felt as pain, the signal must be amplified at some point in your brain.)
I don't see anyone even trying to dispute this - most of the disagreement is, as I said, appeals to mystery (maybe there's some unknown part of the fish brain that can do this, like the optic tectum or brain stem) or appeals to behavior (fish act like they're conscious so they must be conscious and we just haven't figured out how their consciousness works yet.)
3) I'm not sure the relevance of this. Yes, slime molds have very simple behaviors relative to shrimp. Shrimp have very simple behaviors relative to humans. None of this tells us where we can fairly draw the line with respect to consciousness.
4) if you can point me to a line you think I misrepresented you on, I'll correct it.
and search for "autophagy", also just google "autophagy shrimp" and studies will come up.
6) I don't see how that is the case. Do we know octopi lack the neurobiology to do the tasks Key describes? My understanding is that the octopi neurology is considerably more complex than that of fish.
Beyond that, this seems to assume that octopi behavior is just so complex it can't be unconscious, but as other skeptics have pointed out, even highly complex human behavior can be undertaken unconsciously. We need more than that to determine whether something is conscious.
It's simply the case that consciousness is still at least somewhat mysterious, though. Even the easy problems, and the neural correlates, are the subject of significant disagreement amongst experts. And BB didn't claim that we "haven't made major progress", let alone that we can't ever, so that seems like a bit of a straw man and/or motte and bailey. The question isn't whether we have made progress, but whether we *know* that consciousness requires a particular neurobiological feature. And we simply do not know that, even if Key makes a convincing argument (I haven't read it in full yet); if only for outside view reasons, we cannot ever be too confident that we as laymen are correct and a majority of academic experts are incorrect (even if you do identify a plausible way academics could become misguided on this, in terms of social pressure etc.). No matter how convincing we find the argument, it's just not rational to be certain that we know better than most domain experts; we have to retain at least some level of scepticism.
Your post has caused me to lower my confidence in fish feeling pain, for sure, which I did think was basically settled. But it raises alarm, and prevents me from updating as much as I might otherwise, that you are expressing such clearly unjustified overconfidence about the causal mechanisms underlying consciousness. In much the same way that you rightly suggest we should lower our credence in response to advocates eg ignoring failures to replicate (and thank you for pointing this out- I am very disappointed to have been misled on that front), I can't help but adjust mine in your argument when you seem to be circumventing a key objection with apparently indefensible certainty about a pivotal crux of disagreement.
>> And BB didn't claim that we "haven't made major progress", let alone that we can't ever, so that seems like a bit of a straw man and/or motte and bailey. The question isn't whether we have made progress, but whether we *know* that consciousness requires a particular neurobiological feature.
What progress, though? Is it that consciousness is in some way related to brains? Even the Ancient Greeks were aware of that relation. Modern neuroscience should be able to surpass that, and Brian Key here is simply presenting the best working theory for what underlying neurological processes are necessary (not sufficient) for consciousness. He then simply evaluates animal species by whether they have the neurobiology to support those processes. I sympathize with his frustration, to be honest, because it seems like there is extreme reluctance to even grant that modern science has made *that* degree of progress in understanding what more basic processes underlie consciousness.
> Brian Key here is simply presenting the best working theory for what underlying neurological processes are necessary (not sufficient) for consciousness. He then simply evaluates animal species by whether they have the neurobiology to support those processes.
He's presenting a plausible theory, to my understanding. But it's one of many, and the progress that we have made nonetheless does not allow for a high degree of confidence on such questions- especially when there are other competing theories, with more support from relevant experts (outside view should always remain in our sight).
And given it largely conflicts with the behavioural evidence, which is overall more consistent with fish feeling pain (although again, you were convincing here that that evidence is weaker than commonly understood), you have to have a very high degree of confidence indeed that fish lack the neural correlates in order to conclude that fish probably don't feel pain. And I just think that overconfidence is clearly unjustified, given the uncertainty that basically every expert but Brian Key professes in this area (itself a signal to trust them over him in my view).
Interesting read. I’m going to look more into whether ending eyestalk ablation actually causes the industry to farm more shrimp and catch more fish, as that’s a massive risk I don’t want to end up supporting. (If that is true, it would probably be better to focus shrimp advocacy on reducing consumption…) But I’m not convinced of the main argument.
I had also come across the Key paper and my impression is that it’s a small minority view in the field. If I was more knowledgeable about neuroscience and philosophy of mind (and hence if all the research was a lot more comprehensible to me), I’d probably be able to make a judgment based more directly on the evidence… but otherwise I’m going to base my priors on what most people in the field are saying. That seems to be very strong in the direction of “fish experience things that can be better or worse for them” and also more in favor of shrimp feeling than not feeling.
If you can make a stronger case that this is because of cancel culture, then I’d update a lot in the direction of no pain. Would like to hear more about this.
You’d have to show me very strong evidence of that, however, since the number of fish and shrimp used for food is so huge that it would be worth it to “dump money into the ocean” as I had put it) even if you assumed P(sentience) is, say, 0.01 or lower.
If there's one article I'd recommend you read to get a sense of the "skeptical" view, it would be this one. It contains a lot of references to risk of eyestalk ablation and other potential harms of the policies suggested to reduce pain. It also has a good description of some of the cancel culture aspect.
Otherwise, I understand your concern about deferring to consensus - I'm not an expert in neuroscience either. But I have been following the debate for some time, and I do believe there's serious social and political pressure to affirm fish/invertebrate pain and play to anthropomorphic sympathies in the public. And, as I mentioned in my piece, there are specific prominent examples not replicating, which should lower our confidence in arguments that are based on them, as well as communities that rapidly, uncritically embrace them.
But also, I'd say many of the disagreements with skeptics are not the result of complex technical questions about, say, Brian Key's work, but are about fairly straightforward philosophical differences that are more fundamental. Whether to privilege observations of behavior or neurobiological plausibility is a basic disagreement between skeptics and realists about fish pain here. I just think the conceptual arguments for taking a neurobiological approach are far more convincing. This by Brian Key sums up this well:
> I do believe there's serious social and political pressure to affirm fish/invertebrate pain and play to anthropomorphic sympathies in the public
The last part strikes me as somewhat implausible; pressure to play to public sympathy? For fish?! The enormous majority of people demonstrably could not care less about fish. If anything, I would expect there to be pressure in the other direction, because the enormous majority of people strongly want to believe fish suffering doesn't matter.
That's not to say that you're wrong, mind you; it's just surprising enough that I think it requires fuller justification.
Thanks for putting this together: very clear and helpful.
I think it is pretty clear that while in the broader public there is a bias against believing invertebrates feel pain, there’s strong self-selection at work among pain researchers and animal rights activists that creates presumption in favour of animal suffering. This is part of why I find Glenn and BB’s appeal to the scientific consensus on this topic quite unpersuasive. It is not actually that hard for educated non-experts to come to reliable conclusions about the state of the scientific evidence in many scientific fields. This, I think, is one such field.
Thanks! It's a fairly niche debate, but I've been quite consumed with it recently.
As far as I can tell, the pro-pain side tends to rely heavily on two claims: that, as you say, "most" researchers disagree with the skeptics, and that fish/invertebrate behavior provides sufficiently convincing evidence that they feel pain, regardless of whether we understand how that pain could be produced.
I ultimately don't know if the first claim is actually true or not - there are certainly a lot of replies to Key's paper, but it's unclear to me if we can say a majority of researchers really disagree with him or not. Regardless, the debate appears emotionally and politically charged enough that I'm skeptical we can rely on "the consensus" to be a reliable indicator of what we should believe. (I feel similarly about things like the efficacy of child gender transition - I'm not sure consensus is necessarily a reliable indicator of what an educated non-expert person "should" think about the matter.) So I don't find that convincing in this case.
The second claim reflects a more fundamental philosophical difference I think. As another skeptic pointed out, it's possible to program robots that fulfill many of the behavioral indicators of sentience. Since robots are (presumably) not sentient, it seems to me that behavior is not a reliable indicator for consciousness. But so many people I encounter refuse to accept this. They see certain behaviors in animals as strong evidence of sentience, regardless of the neurological plausibility, and I think that's a major sticking point that tends to frustrate the skeptics.
Interesting and hopefully true! Do you feel comfortable expressing your degree of credence in this conclusion in terms of a percentage? How would you respond to someone who thinks shrimp are only 1% likely to be sentient but still thinks shrimp stunners are a good EV bet because of the large numbers of shrimp affected?
I think before I could offer a percentage credence, I would have to first have a working theory for how fish etc are supposed to feel pain, as opposed to mysterianism about how certain parts of the fish brain may, in some unknown way, support fish pain. Until someone offers that, I just don't see how it's productive to talk about credences in the idea itself, especially if those credences are supposed to guide our actions in the world.
I really don't think it's a defensible position that you can't offer a probability estimate until you have a working theory of how fish are "supposed" to feel pain. Whatever the physical mechanism, there is obviously some evidence that they do, even if ultimately you think it is weak. Your credence certainly shouldn't be 100%, as you almost seem to imply! So what is it roughly? How confident are you that they don't? Because that's quite important in this context, where the scale of suffering, if it exists, is so vast.
Well, I'll ask you: what is your credence in the idea that the grass in your backyard feels pain? Some scientists have argued that plants can feel pain based on plant "behavior", such as reactions to stress.
Seems like a bit of a deflection- and if you're implying any equivalence here, I think that's very misguided. There's vastly more behavioural evidence of subjective experience of suffering in fish than grass, and vastly less mechanistic reason to doubt it, given that grass literally does not have a brain, nervous system etc.- things that seem almost certainly required for subjective experience of pain- whereas fish merely don't have certain particular structures which might plausibly be correlates (although as I say I did find this a fairly convincing argument, pending further investigation of e.g Key's work, and has reduced my credence significantly in fish pain even if it's still well north of 50%). These are *very* different propositions, even if both are primarily justified in terms of behavioural observations.
But in order to avoid hypocrisy, I'll bite: I'd give it somewhere in the region of 2%. Mechanistically, it seems almost impossible, but ultimately consciousness is still mysterious, and it's totally possible that we are completely wrong about how it arises. I have to have some scepticism of my conclusions in an area of such uncertainty, so the minimum credence I can really give to any living organism experiencing pain is probably about 2%.
I have a bunch of things to say about the subject, but let me note just a few:
1) You should be highly uncertain about the subject. When lots of smart people disagree on a subject that you're not an expert on, even if you've done a decent amount of reading about it, you should basically never be more than 80% confident, especially if your view is the minority view. But the case for shrimp welfare goes through even if you're 95% sure shrimp aren't conscious, just because of the insane degree of the carnage and effectiveness.
2) I think we're largely in the dark about what ingredients are needed for consciousness. If this is right, then Key's speculation by analogy about the ingredients needed not being present in fish is highly improbable. Behavior is all we have to go off of. And shrimp behave in lots of ways that we'd expect them to if they felt pain (see my piece).
3) The sorts of things slime molds do are very rudimentary compared to what shrimp do--relevantly like a computer or mouse trap with many settings rather than a conscious being.
4) I do think you misrepresented me a bit. I'm now super confident in shrimp pain--I'd put it in the low 60s.
5) I couldn't find a source for the claim that shrimp eat their own body parts. I couldn't even find people using the word autophagy to mean eating one's own body parts https://en.wikipedia.org/wiki/Autophagy
6) The Key view implies octopi aren't conscious, which I take to be a decisive counterexample.
To 1), I'll restate what I said to Glenn above. The principal disagreements in this debate are not over complicated technical questions in neuroscience that you would need a PhD in the area to evaluate. They're fundamentally philosophical disagreements over questions like: Are observations of behavior or neurobiology the best way to determine whether an animal is conscious? And I believe a reasonably informed observer can pass judgment on these questions.
2) I fundamentally disagree with the idea that consciousness is still very mysterious and that we haven't (and possibly can't ever) make major progress in understanding consciousness. Ultimately, it's going to be explained by simpler biological processes and Key has provided an enormous body of empirical evidence supporting his theory that signal amplification (for example) is a critical process supporting consciousness. (His ideas also make intuitive sense - your body has billions of sense receptors and for any one of those signals to be felt as pain, the signal must be amplified at some point in your brain.)
I don't see anyone even trying to dispute this - most of the disagreement is, as I said, appeals to mystery (maybe there's some unknown part of the fish brain that can do this, like the optic tectum or brain stem) or appeals to behavior (fish act like they're conscious so they must be conscious and we just haven't figured out how their consciousness works yet.)
3) I'm not sure the relevance of this. Yes, slime molds have very simple behaviors relative to shrimp. Shrimp have very simple behaviors relative to humans. None of this tells us where we can fairly draw the line with respect to consciousness.
4) if you can point me to a line you think I misrepresented you on, I'll correct it.
5) See
https://www.tandfonline.com/doi/full/10.1080/23308249.2023.2257802#d1e1741
and search for "autophagy", also just google "autophagy shrimp" and studies will come up.
6) I don't see how that is the case. Do we know octopi lack the neurobiology to do the tasks Key describes? My understanding is that the octopi neurology is considerably more complex than that of fish.
Beyond that, this seems to assume that octopi behavior is just so complex it can't be unconscious, but as other skeptics have pointed out, even highly complex human behavior can be undertaken unconsciously. We need more than that to determine whether something is conscious.
It's simply the case that consciousness is still at least somewhat mysterious, though. Even the easy problems, and the neural correlates, are the subject of significant disagreement amongst experts. And BB didn't claim that we "haven't made major progress", let alone that we can't ever, so that seems like a bit of a straw man and/or motte and bailey. The question isn't whether we have made progress, but whether we *know* that consciousness requires a particular neurobiological feature. And we simply do not know that, even if Key makes a convincing argument (I haven't read it in full yet); if only for outside view reasons, we cannot ever be too confident that we as laymen are correct and a majority of academic experts are incorrect (even if you do identify a plausible way academics could become misguided on this, in terms of social pressure etc.). No matter how convincing we find the argument, it's just not rational to be certain that we know better than most domain experts; we have to retain at least some level of scepticism.
Your post has caused me to lower my confidence in fish feeling pain, for sure, which I did think was basically settled. But it raises alarm, and prevents me from updating as much as I might otherwise, that you are expressing such clearly unjustified overconfidence about the causal mechanisms underlying consciousness. In much the same way that you rightly suggest we should lower our credence in response to advocates eg ignoring failures to replicate (and thank you for pointing this out- I am very disappointed to have been misled on that front), I can't help but adjust mine in your argument when you seem to be circumventing a key objection with apparently indefensible certainty about a pivotal crux of disagreement.
>> And BB didn't claim that we "haven't made major progress", let alone that we can't ever, so that seems like a bit of a straw man and/or motte and bailey. The question isn't whether we have made progress, but whether we *know* that consciousness requires a particular neurobiological feature.
What progress, though? Is it that consciousness is in some way related to brains? Even the Ancient Greeks were aware of that relation. Modern neuroscience should be able to surpass that, and Brian Key here is simply presenting the best working theory for what underlying neurological processes are necessary (not sufficient) for consciousness. He then simply evaluates animal species by whether they have the neurobiology to support those processes. I sympathize with his frustration, to be honest, because it seems like there is extreme reluctance to even grant that modern science has made *that* degree of progress in understanding what more basic processes underlie consciousness.
> Brian Key here is simply presenting the best working theory for what underlying neurological processes are necessary (not sufficient) for consciousness. He then simply evaluates animal species by whether they have the neurobiology to support those processes.
He's presenting a plausible theory, to my understanding. But it's one of many, and the progress that we have made nonetheless does not allow for a high degree of confidence on such questions- especially when there are other competing theories, with more support from relevant experts (outside view should always remain in our sight).
And given it largely conflicts with the behavioural evidence, which is overall more consistent with fish feeling pain (although again, you were convincing here that that evidence is weaker than commonly understood), you have to have a very high degree of confidence indeed that fish lack the neural correlates in order to conclude that fish probably don't feel pain. And I just think that overconfidence is clearly unjustified, given the uncertainty that basically every expert but Brian Key professes in this area (itself a signal to trust them over him in my view).
Interesting read. I’m going to look more into whether ending eyestalk ablation actually causes the industry to farm more shrimp and catch more fish, as that’s a massive risk I don’t want to end up supporting. (If that is true, it would probably be better to focus shrimp advocacy on reducing consumption…) But I’m not convinced of the main argument.
I had also come across the Key paper and my impression is that it’s a small minority view in the field. If I was more knowledgeable about neuroscience and philosophy of mind (and hence if all the research was a lot more comprehensible to me), I’d probably be able to make a judgment based more directly on the evidence… but otherwise I’m going to base my priors on what most people in the field are saying. That seems to be very strong in the direction of “fish experience things that can be better or worse for them” and also more in favor of shrimp feeling than not feeling.
If you can make a stronger case that this is because of cancel culture, then I’d update a lot in the direction of no pain. Would like to hear more about this.
You’d have to show me very strong evidence of that, however, since the number of fish and shrimp used for food is so huge that it would be worth it to “dump money into the ocean” as I had put it) even if you assumed P(sentience) is, say, 0.01 or lower.
If there's one article I'd recommend you read to get a sense of the "skeptical" view, it would be this one. It contains a lot of references to risk of eyestalk ablation and other potential harms of the policies suggested to reduce pain. It also has a good description of some of the cancel culture aspect.
https://www.tandfonline.com/doi/full/10.1080/23308249.2023.2257802#
Otherwise, I understand your concern about deferring to consensus - I'm not an expert in neuroscience either. But I have been following the debate for some time, and I do believe there's serious social and political pressure to affirm fish/invertebrate pain and play to anthropomorphic sympathies in the public. And, as I mentioned in my piece, there are specific prominent examples not replicating, which should lower our confidence in arguments that are based on them, as well as communities that rapidly, uncritically embrace them.
But also, I'd say many of the disagreements with skeptics are not the result of complex technical questions about, say, Brian Key's work, but are about fairly straightforward philosophical differences that are more fundamental. Whether to privilege observations of behavior or neurobiological plausibility is a basic disagreement between skeptics and realists about fish pain here. I just think the conceptual arguments for taking a neurobiological approach are far more convincing. This by Brian Key sums up this well:
https://www.wellbeingintlstudiesrepository.org/animsent/vol1/iss3/44/
> I do believe there's serious social and political pressure to affirm fish/invertebrate pain and play to anthropomorphic sympathies in the public
The last part strikes me as somewhat implausible; pressure to play to public sympathy? For fish?! The enormous majority of people demonstrably could not care less about fish. If anything, I would expect there to be pressure in the other direction, because the enormous majority of people strongly want to believe fish suffering doesn't matter.
That's not to say that you're wrong, mind you; it's just surprising enough that I think it requires fuller justification.
Thanks for putting this together: very clear and helpful.
I think it is pretty clear that while in the broader public there is a bias against believing invertebrates feel pain, there’s strong self-selection at work among pain researchers and animal rights activists that creates presumption in favour of animal suffering. This is part of why I find Glenn and BB’s appeal to the scientific consensus on this topic quite unpersuasive. It is not actually that hard for educated non-experts to come to reliable conclusions about the state of the scientific evidence in many scientific fields. This, I think, is one such field.
Thanks! It's a fairly niche debate, but I've been quite consumed with it recently.
As far as I can tell, the pro-pain side tends to rely heavily on two claims: that, as you say, "most" researchers disagree with the skeptics, and that fish/invertebrate behavior provides sufficiently convincing evidence that they feel pain, regardless of whether we understand how that pain could be produced.
I ultimately don't know if the first claim is actually true or not - there are certainly a lot of replies to Key's paper, but it's unclear to me if we can say a majority of researchers really disagree with him or not. Regardless, the debate appears emotionally and politically charged enough that I'm skeptical we can rely on "the consensus" to be a reliable indicator of what we should believe. (I feel similarly about things like the efficacy of child gender transition - I'm not sure consensus is necessarily a reliable indicator of what an educated non-expert person "should" think about the matter.) So I don't find that convincing in this case.
The second claim reflects a more fundamental philosophical difference I think. As another skeptic pointed out, it's possible to program robots that fulfill many of the behavioral indicators of sentience. Since robots are (presumably) not sentient, it seems to me that behavior is not a reliable indicator for consciousness. But so many people I encounter refuse to accept this. They see certain behaviors in animals as strong evidence of sentience, regardless of the neurological plausibility, and I think that's a major sticking point that tends to frustrate the skeptics.
Interesting and hopefully true! Do you feel comfortable expressing your degree of credence in this conclusion in terms of a percentage? How would you respond to someone who thinks shrimp are only 1% likely to be sentient but still thinks shrimp stunners are a good EV bet because of the large numbers of shrimp affected?
Thank you!
I think before I could offer a percentage credence, I would have to first have a working theory for how fish etc are supposed to feel pain, as opposed to mysterianism about how certain parts of the fish brain may, in some unknown way, support fish pain. Until someone offers that, I just don't see how it's productive to talk about credences in the idea itself, especially if those credences are supposed to guide our actions in the world.
https://slatestarcodex.com/2013/05/02/if-its-worth-doing-its-worth-doing-with-made-up-statistics/
I really don't think it's a defensible position that you can't offer a probability estimate until you have a working theory of how fish are "supposed" to feel pain. Whatever the physical mechanism, there is obviously some evidence that they do, even if ultimately you think it is weak. Your credence certainly shouldn't be 100%, as you almost seem to imply! So what is it roughly? How confident are you that they don't? Because that's quite important in this context, where the scale of suffering, if it exists, is so vast.
Well, I'll ask you: what is your credence in the idea that the grass in your backyard feels pain? Some scientists have argued that plants can feel pain based on plant "behavior", such as reactions to stress.
Seems like a bit of a deflection- and if you're implying any equivalence here, I think that's very misguided. There's vastly more behavioural evidence of subjective experience of suffering in fish than grass, and vastly less mechanistic reason to doubt it, given that grass literally does not have a brain, nervous system etc.- things that seem almost certainly required for subjective experience of pain- whereas fish merely don't have certain particular structures which might plausibly be correlates (although as I say I did find this a fairly convincing argument, pending further investigation of e.g Key's work, and has reduced my credence significantly in fish pain even if it's still well north of 50%). These are *very* different propositions, even if both are primarily justified in terms of behavioural observations.
But in order to avoid hypocrisy, I'll bite: I'd give it somewhere in the region of 2%. Mechanistically, it seems almost impossible, but ultimately consciousness is still mysterious, and it's totally possible that we are completely wrong about how it arises. I have to have some scepticism of my conclusions in an area of such uncertainty, so the minimum credence I can really give to any living organism experiencing pain is probably about 2%.