On November 10, Rethink Priorities hosted a workshop with five expert panellists on the subject of "Cognition, welfare, and the problem of interspecies comparisons".

Recording | Speaker Bios

These are my rough notes from the workshop. I am not an expert in this area but I hope these notes may be useful for readers who would like to learn more about this topic.

I am grateful to the panellists for sharing their insights on this topic and to Jason Schukraft, Gavin Taylor, Rachael Harrison, and Bob Fischer for providing feedback on these notes.

The panellists

Lynne Sneddon (fish)

• Evidence of pain is fundamental to which animals are protected
• Less focus on positive experiences (but there is some)
• Measures include: selective attention, behavioural measures (pain incurring situations vs not), avoidance learning/ fear responses, motivational/ preferences (e.g. anti-predator)
• Consider evolutionary pressures and life history with experimental paradigms
• Sentient vs not - is not binary
• Need to test if the pain dominates attention

Jean-Loup Rault (social behaviour in mammals, esp. pigs)

• Focussed on indicators (e.g. oxytocin, neuroimaging, behaviour)
• Study of pro-social tendencies in farmed animals has been neglected
• Measures include: hormonal (oxytocin, EEG, neuroimaging, behaviour), attention bias
• Looking at empathy, emotional contagion and regulation

Christine Nicol (primarily chickens, also mice, pigs, horses)

• Animal preferences e.g. space, transport
• Interested in designing the tests
• Measures like conditional place preferences, memory tests, pain and analgesics (e.g. fractured keel bone - show preference for analgesic vs neutral - no preference shown)
• Not just preference for space, but do they have preference for previous experiences?
• How are the birds making choices? Consistent? Rational?
• Animal cognition
• We’d invest millions in communicating with an alien species, so why don’t we do the same with species on earth? (see her 4-part podcast series “Would you eat an alien?”
• Chickens learn from dominant birds (social learning), excellent spatial memory, remember things that go out of sight (object permanence), can anticipate the future (for a few seconds) (self control), example of Nicol and Pope (1996) Animal Behaviour paper showing mother hens respond to chicks showing incorrect behaviours with more teaching/ warning behaviours
• Integration of processes/approaches
• Either preference tests or welfare indicators (i.e. health, behaviours, physiology), but there were no papers that link the two approaches
• Looked at the correlation between the two (with 100 indicators), e.g. Paul et al., 2022 Animal Behaviour
• Triangulation
• Look at preferences, judgment biases and health
• If these align we can take it as welfare relevant, if not it could be a bad choice of indicators or welfare is not relevant for that animal/situation

Bob Elwood (crustaceans)

• Mainstream view was that crustacean pain response is purely by nociceptive reflex
• i.e. they have no awareness or feeling related to the stimulus
• But he found that hermit crabs can make motivational trade-offs between shocks and shells
• E.g. taking into account shell quality in making a decision to stay or leave shell when receiving a small shock - i.e. decision vs reflex.
• This suggests that it can’t simply be nociceptive reflex
• The memory of the aversive state lasted 24 hours.
• Prawns tend to wounded areas as well
• Crayfish that receive shocks become more risk-averse due to increased serotonin. Fossat et al., 2014 Science - link anxiety to shock with space use
• You can overcome that behaviour by giving them Librium
• Rapid avoidance learning - responses to pain have an evolutionary basis (i.e. avoiding damage to skin/ body)
• “Awareness” of the site of the injury
• Cannot prove pain (though consistent with pain criteria), but accumulating evidence

Lars Chittka (bees)

• Charles Henry Turner observed ‘excitement’ in digger wasps
• “Backyard Brains” - control a cockroach
• Reflex automatons or philosophical zombies?
• There is basic nociception - because they try to get away from a noxious stimulus
• Bees will spread a scent (alarm pheromone) to communicate a threat and this spreads an endogenous painkiller in their bodies.
• Bees are great at problem-solving (e.g. puzzles), social learning and sharing solutions with other bees, can count, object manipulate
• Long-lasting behaviour change (more cautious) in response to spider attacks
• Comprehensive measures, taken together, may indicate consciousness
• Observe indicators that correlate with human emotional states

Is it possible to compare the valenced experience of two species? If it was possible, what would it look like, how would you do it?

Bob

• If animals are not prepared to give something up, then the pain is trivial. Need to look at trade-offs to avoid the stimulus.
• If I found a neural correlate in an animal experiencing a painful stimulus, it won’t tell me anything about what the animal is feeling. We need the subject to communicate the nature of their experience.
• Crabs have spatial awareness of the size of a shell and whether it will fit through a gap. They’ll switch to a low-quality, but smaller shell to escape an enclosure.
• Mantis example: should the pain of being eaten prevent them from mating? If they’re attacked before mating begins they try to get away, but when mating has started there is something that dampens the flight response.

Lynne

• Could provide noxious stimulus and observe change in behaviour and how long that lasts for (but doesn’t really answer the question)
• Can’t measure human pain objectively so how can we compare across species?
• Humans have cognitive control over pain, but simple animals without a developed cortex don’t have the level of cognition to make trade-offs
• Does that mean the pain is much worse for them?
• Lars was sceptical of using brain regions to categorise
• Christine gives that idea more weight
• Tattoos are an interesting case study, why do people go through the pain?
• What cost is an animal willing to pay to access pain relief? Can you compare the ‘cost’ across species?

Lars

• Quantifying pain is impossible (dependent on experience, perception etc), but can try to clarify if pain is present (binary, yes/no) and how long it lasts.
• You can regulate pain with your mind and with endogenous pain relief
• The opposite case is the builder reacting to the nail in his foot (which actually went between his toes).
• You can test if there is SOME pain or NOT pain, but not a scale
• You can also see how long pain responses last
• Correlation is not causation so we need to be careful about using neural correlates to determine if other species are conscious.
• Insects experience relief of pain as a reward
• The field is plagued by publication bias (things animals can do, rather than things they can’t do)
• Bees can learn from observing each other, but we don’t know anything about their mental states when they are doing this.

Jean-Loup

• It is subjective and difficult to compare
• Pain is simpler than comparing other mental states
• It may be easier to study social animals as they have been selected to be communicative to others? (e.g. social bees vs solitary bees) - socio-ecological drivers?

Christine

• Pain is not the ONLY valenced experience, it can be regulated by social and environmental factors
• There are a range of dimensions for valenced experience
• It’s not clear what the conditions are for having sentience
• There are neural correlates with awareness (inc. dreaming) but not with deep sleep or a vegetative state.
• What would we look for in invertebrates to determine consciousness?
• In a population of chickens, different chickens develop preferences for different environments and they make consistent and rational choices (e.g. A>B, B>C, so A>C)
• Mammals (e.g. rats) can react based on past experience much better than birds (e.g. chickens) (this is called contrast effects) - but note lack of data, publishing bias for positive results
• Has never felt a social connection with a chicken, but instantly did with keas (as well as mammals)

Adam Shriver

• Are there neurological processes behind pain?
• A pinprick is less painful than breaking a leg so comparisons are possible

Trading off between species e.g. a pig to a shrimp

Christine

• Even if the level of pain was the same, would give much more weight to the pig than to a crustacean or insect. Pigs are more complex (i.e. multiple and sometimes related factors) and it can impact their social relationships too. There’s a lot more to think about. “Broader complexity”
• Accepts that shrimp can feel pain, but if there is a trade-off she would choose the pig because they have more interests and more dimensions of welfare.
• Some of those extra dimensions can be used to compensate for pain

Lars

• Even human babies didn’t receive analgesics for circumcision (despite showing behavioural indicators for pain but pre-verbal so cannot express pain in words). Dismissing it as a reflex is too convenient and that excuse can be applied to other animals too.
• Insects do not qualify as animals under UK law so he can use any invasive methods he wants without any restriction.
• The quantity is important too. The nutritional value of 1 cow would require ~millions of shrimp.
• Some animals may experience less pain and suffering than others.
• We celebrate animals with senses that are different to humans (e.g. UV, ultrasound), but when it comes to cognition, we value animals that have similar brains to ours.

Lynne

• All animals should be treated equally, otherwise, you’re being speciesist.
• Donald Broom has developed a list of criteria for sentience
• Ability to form relationships
• Ability to communicate
• Reciprocity between individuals
• Make decisions based on cost-risk
• Positive or negative feelings
• Some level of awareness
• Should we direct more funding to research on species where we know very little about their welfare, rather than towards pigs which we understand well?

Bob

• In Switzerland, there has been legal change about how to treat lobsters
• The UK is considering decapods and cephalopods
• It’s a question of empathy - people don’t care about shrimp, but they might object to chickens being harmed
• We don’t know which ones suffer more, but we can compare their responses
• Humans place more weight on animals that are more like us but there are no ‘higher’ animals. All animals have had the same duration of evolutionary adaptation and have come to their own solutions for their environment.
• However, he would still choose the pig because he feels greater empathy and there is legal precedent.

Jean-Loup

• Do we “think” a certain species is sentient, or do we “know”?
• How much evidence is sufficient for decision making?
• We have no “smoking gun” criteria for sentience (in animals or machines)

Q&A session

Can we tie welfare to a behavioural matrix rather than subjective feelings?

• Lars - nociception is sensation and can be mediated without subjective unpleasant experience e.g. withdrawal reflexes. Pain is the subjective experience and requires an element of consciousness.
• Jean-Loup - we use indirect indicators as correlates, the conscious element is central to the welfare concern
• Christine - if we wait until we have an answer on consciousness then we’ll be waiting a long time, but we can still make progress by assuming consciousness. She supports the matrix concept and thinks welfare is more than one thing. Marianne Dawkins says that welfare consists of health and desires. Health indicators that don’t influence desires can be ignored.
• Bob - nociception enables a reflex response to prevent tissue damage, but if it has no awareness then it can keep getting damaged, pain provides long term protection which conveys an evolutionary advantage. Prefers “awareness” rather than “sentience” or “consciousness”.

What evidence would convince you that interspecies comparisons are impossible?

• Lars - strong credence that it’s not possible, it’s hard enough to compare between humans. Perhaps the absence of pain in plants as they don’t have the necessary receptors. Perhaps we can only say it’s present or not present rather than putting it on a scale.

Would it be possible to measure brain activity during trade-offs & choices?

• Bob - there is no “pain” to be perceived, it is an interpretation of a nociceptive stimulus (like sound is an interpretation of vibrations)
• Christine - if an animal chooses to accept pain, it indicates that there is something more important than that pain e.g. women accept pain to have a baby.

How is cognition related to welfare?

• Jean-Loup - we interpret welfare from cognition. Animals can remember and predict things
• Lars - cognition is not indicative of the capacity to suffer or feel pain. Our current tests require a certain level of intelligence which results in a bias. We may be missing species that are suffering but cannot learn to respond to a stimulus. Cognition/ intelligence, emotion, and pain are separate - don’t need one for the other necessarily.

When should we apply the precautionary principle? What level of evidence is sufficient?

• Lynne - stress response, behavioural changes that are prolonged and complex (not just reflex), show that analgesics reduce the behaviour changes, changes decision making (learns to avoid the stimulus or alters its preferences), if animals don’t have a negative experience then they won’t learn

Is the difficulty in welfare comparison conceptual or empirical?

• Jean-Loup - all the animals we’ve discussed have sentience so interspecies comparison is not important as we should reduce pain for them all
• Adam - but how do we make trade-offs?
• Bob - what cost would an animal incur to avoid pain? That’s easy to determine in a human experiment. If a crab pays a large cost (in terms of fitness) to escape a stimulus you could conclude that the pain is worse for the crab.
• Christine - if the rest of the life was impacted (e.g. a depressed state), you could try making some comparison on that basis, we’re starting from a point where we’re already making interspecies comparisons all the time but if we get it wrong by overthinking it then we might end up making things worse, but her prior is that we’ll get better at it rather than worse.

How can we strengthen public understanding to influence society and policy?

• Lynne - engaging with the public and the media, promote research on Twitter, lots of interviews with journalists and documentaries, involved in government bodies and reports, engage with animal charities and campaigning groups, increase public perception and awareness because that drives the improvements e.g. willing to pay more for higher welfare standards
• Bob - a different approach, declined to join campaign groups, he believes in what they’re doing but believes that a scientist should remain neutral and not defend a position, always willing to talk to the media but they will always twist things and ignore the nuance, UK government has been suppressing the report on decapod and cephalopods, he suspects there’s something in there that the government doesn’t like (probably related to the fishing industry)

How can we assess welfare in the wild? Should we improve their environment?

• Lynne - when humans interfere we have an effect on welfare, we should try to understand that and mitigate it, invasive species have a negative effect on other species but they were put there by humans and their welfare matters just as much so they should be managed ethically, humans put them there so we’re responsible for them
• Jean-Loup - there were reports of PTSD for animals that faced trauma in the wild, if we apply the same indicators we use on domestic species it will have similar implications for wild animals too.

What animal welfare legislation would you change and why?

• Christine - focus on genetic robustness of farm animals, legislation focuses on their environment, but many animals are being bred that are not fit or robust e.g. broiler chicken skeletons are frail. Lots of groups use terrible methods to kill rats and mice but this is not allowed for researchers
• Bob - ban dismemberment of live animals e.g. lobster processing, in some fisheries they pull off the crab claws and throw the body back
• Lynne - humane stunning and slaughter of wild fishes, add stunning equipment to large fishing boats so the fish are not gutted while they are alive, we should include crustaceans in the European ethics requirements
• Lars - stop subsidies for meat farmers, insects deserve greater consideration, there is a good case for researchers limiting their invasive procedures
• Jean-Loup - include animal welfare in the school curriculum, a lot of harm comes from ignorance, people don’t realise that animals experience pain and this hinders progress, children seem to learn speciesism from adults and start life with more affinity with other animals.

My immediate takeaways

• It’s clear that many animals are aware of noxious stimuli otherwise they would not be able to change their behaviour as a result.
• However, we don’t yet have a way to determine if this awareness is experienced with a negative valence (other than identifying indicators that correlate with negative valence in humans).
• We need to make decisions in the real world right now. With our current level of understanding, I think it is justifiable to use indicators as predictors of suffering, even though we lack an objective measure for suffering.
• We already follow this approach for humans that are unable to communicate such as a fetus or a person in a vegetative state.
• In conclusion, it seems like the best method available to us right now is to use preference studies and indicators that correlate with subjective well-being in humans and apply these to non-human animals to determine their welfare objectively.

Reflections from Rachael Harrison

• Christine’s point about the need for multiple and often correlating measures is an important one, and one I agree with and have been emphasising in relation to the proxy list in discussions with Bob.
• Most speakers were reluctant to make species comparisons directly (or indirectly), except Christine, or what the species can’t do (as opposed to can, though I understand the various issues with negative results). Perhaps a comparative cognition person might be useful at the upcoming workshop who may be more willing to do this. Also, perhaps someone that doesn’t believe that insects/ fish etc. show pain or emotion etc. to have that side of the argument presented too.
• The adaptive value of pain from an evolutionary sense is important - seems applicable to some degree to capacity for welfare too.
• Little practical example that came to mind - when I was a zookeeper back in the day, we fed dead fish and mammals as zoo animal food, but live insects. I always wondered about this ethically, perhaps a useful anecdote for illustrating the impact of legislation in practice (alongside research ethics etc).
• Should sociality be a proxy for CFW
• Very pain-focussed, perhaps more cognition or emotion/ affect focussed viewpoints next time would be useful too. May also be worth having someone from RP present at the start of the next workshop about what the aim of the CFW project is, showing why these decisions (inter-species comparisons) are needed, and your aim for this to be empirically grounded.