Our Initial Work on a Deep Investigation of Animal Sentience

Introduction

Many people think that whether or not we have moral obligations to animals hinges upon whether animals are sentient. In this short post, we use the term “sentience” to refer to the capacity for “conscious experiences” including “pleasure” and “suffering.”¹ For several years, ACE’s long-term research agenda has included a project investigating and cataloguing evidence on animal sentience. We made some progress on this in 2016, thanks to making our third hire in the research department which freed up some staff time.

In Fall 2016, we found out that Sentience Politics, an antispeciesist think tank run by the Effective Altruism Foundation, was working on a similar project. We spoke with them about our respective plans and progress, and we decided it would be best for ACE to pause its research in the area for the time being and follow Sentience Politics’ research progress. Sentience Politics plans to put substantial work into this project, probably more than we would, and it seems like having one quality research project on this topic is nearly as good as having two. In this post, we describe our perspective on the importance and implications of this research area, then briefly touch on the philosophical questions regarding animal sentience and next steps for research in this area.

Implications of animal sentience research

Many people find it obvious that all farmed animals have rich mental lives that make their experiences morally important. However, there could be interesting and useful insights gained from investigating this topic more thoroughly. Animal sentience seems like an especially important research area when we consider the extremely large numbers of small creatures in the world.

For example, there are possibly 10¹⁸ to 10¹⁹ (1-10 quintillion) “bugs” (e.g. insects, earthworms) on earth. Many of us intuitively would say these animals only count for a fraction, if any, of the moral value of a human or other mammal, perhaps because we think bugs are less sentient, if at all.²^,³ So what if we think a bug counts for only 10^-6 (one out of a million) the moral value of a standard farmed animal like a chicken, fish, pig, or cow? This seems like a pretty small number that would fit some people’s intuitions. There are plausible indications of sentience for bees and other insects, although there’s little agreement on what the criteria for identifying sentience are:

Even this apparently modest estimate that a bug has 10^-6 of the moral value of a farmed animal would suggest that all the bugs of the world matter approximately 42 to 416 times more than all of the chickens, turkeys, cows, and pigs on farms.⁵ But what if our figure is more like 10^-9 (one out of a billion)? In that case, land-dwelling farmed animals would count for 2.4 to 24 times the amount that bugs matter.

Refining our estimates could have important implications in other areas as well. For example, while farmed animals outnumber humans approximately 10:1,⁶ if we think farmed animals were less than 1% likely to be as sentient as humans, then we should value the human population more than the population of farmed animals.⁷ While we would still need to consider other factors for cause prioritization like the suffering per individual and neglectedness and tractability of each area, getting more precise estimates seems potentially quite useful.

Developing a better understanding of sentience may also help us decide how to prioritize helping fish and chickens, such as when developing new plant-based or cultured foods or when campaigning for welfare reforms. Farmed fish make up approximately 43% of all days spent on farms, and chickens make up 51.3% (29.7% meat chickens, 21.6% egg-laying hens).⁸ This means that if the welfare of each group is similar and other factors are also similar on average, then a 10 times difference in sentience could make one population much more of a priority. If we consider the approximately 1-3 trillion wild fish caught globally each year, and think the welfare and lifespan factors are similar between wild and farmed fish, then fish impact is likely a more important consideration if we think chickens and fish are similarly sentient. Investigating animal sentience could plausibly indicate that chickens are an order of magnitude more sentient (or more likely to be sentient), suggesting that work on helping chickens is more important. Research could also indicate a lack of a meaningful distinction between the sentience of chickens and fish, or between another group like pigs and chickens, each of which would have their own important implications.

Philosophical perspective

There are many philosophical questions that might help us decide which beings we care about. We discussed some of those problems this year, but found that we were unable to come to much agreement. We found that ACE staff often disagreed about the best ways to approach them. Here are some examples of the problems we discussed:

What criteria should we use for determining whether a being deserves moral consideration? For example, should moral consideration be afforded to anyone who is “conscious”? Anyone who is “sentient”? Anyone with the capacity for “pleasure” and “suffering”? Anyone who experiences “emotions” or has “interests”?⁹ And is there even a right answer to this question?
What are the most plausible theories of what sentience consists of?
What constitutes evidence of sentience? For example, is the display of certain behaviors evidence of sentience? Is the possession of certain neuroanatomical structures evidence of sentience?

ACE staff did agree that regardless of which of the most promising theories of consciousness is correct (if any of them are correct), there are good reasons to think that at least commonly farmed animals (fish, chickens, pigs, etc.) are sentient and deserve moral consideration. Regardless of our theory of sentience, we can use behavioral and neuroscientific evidence to inform our assessments of animal sentience, although it is unclear how to weigh them and produce a final judgment. Further work studying this evidence could produce tentative qualitative conclusions, or even quantitative estimates based on likelihood and/or degree of sentience.

Next steps

We think the most-needed research in this area is collecting and cataloguing the empirical evidence for sentience, such as neuroscientific and animal behavior studies sorted by feature (e.g. theory of mind, response to painkillers, sensory integration, play behavior) and population (e.g. chickens, insects). Such a collection doesn’t seem to currently exist, but we believe Sentience Politics will produce one within a year or so. We sent them the resources we had collected, as well as our views on the philosophical issues discussed above. It’s possible we’ll revisit this in the future, either to provide our take on Sentience Politics’ work, or to do further research of our own.

We use quotes here to clarify that these terms are not well-defined without further explanation. The definition we use here is not specific enough for rigorous analysis.
For an argument that different beings might deserve different degrees of moral consideration, see David DeGrazia (2008).
It’s controversial whether sentience is a binary feature or whether it comes in degrees. One person who has argued that sentience comes in degrees is Daniel Dennett (1995).
This means learning to avoid or seek out certain stimuli, such as a bitter liquid, due to association with a punishment or reward.
Number of land-dwelling farmed animals taken from this page as 3.3*1010 (33 billion).
Human population is estimated at 7.4 billion and farmed animals include both the 33 billion land-dwelling animals mentioned above and many more fish. The number of farmed fish alive at any one time is unclear, but around 37-120 billion were killed from farms in 2010.
To be clear, we are considering the possibility that humans are 100 times as sentient as animals in order to illustrate a point. We don’t necessarily believe it to be true.
These calculations are made using the information in ACE’s Impact Calculator: average lifespans and average animal equivalents consumed per person. While these figures are surely imperfect, they are probably one of the best approximations readily available.
We use quotes here to clarify that these terms are not technically rigorous without further explanation.