(Cross-posted from my new substack The Ethical Economist: a blog covering Economics, Ethics and Effective Altruism.)
Consider a life which, on bringing it into existence, neither increases nor decreases social welfare because it is perfectly neutral in quality. What is such a life like?
This is a hugely important question. For Effective Altruists, the answer has implications for the value of the far future and the relative importance of saving lives versus reducing suffering. Even those not looking to improve the world should be interested - the answer also has implications for the ethics of having children and even whether or not you were wronged by being brought into existence. Despite this, the question has been given surprisingly little attention in both EA and non-EA circles.
But what have those who have looked into this question had to say? Well let me take you through the literature…
Indifference between life and death
The great philosopher Henry Sidgwick’s view was that a person is at the zero-level of wellbeing when they are indifferent between suicide and continuing to live. This seems pretty intuitive - if you choose to keep living, your life is probably a good one, and vice versa. Under this definition, the vast majority of people live positive lives. After all, most people don’t kill themselves.
Not so fast says renowned economist Partha Dasgupta. The idea that someone who wants to continue living must be living a life of positive welfare ignores the badness of death. It is certainly possible for someone to be living a negative life, but be reluctant to end it because the subjective badness of killing oneself is greater than the badness of continuing to live. Death is indeed highly unpleasant, for various reasons. Dasgupta writes:
Religious prohibition, fear of the process of dying (the possibility of suffering pain, the feeling of isolation), the thought that one would be betraying family and friends, and the deep resistance to the idea of taking one’s own life that has been built into us through selection pressure would cause someone even in deep misery to balk. It may even be that no matter what life throws at us we adjust to it, if only to make it possible to carry on.
I think Daspugta is correct - someone who is indifferent between living on and taking their own life is clearly below the zero-level. This is also intuitive - we would usually say that someone on the brink of suicide is living a miserable life.
The creation test
Dasgupta’s criticism of Sidgwick is compelling, but does he have anything to say himself on what the zero-level might be?
Yes - Dasgupta’s proposal is that instead of thinking about the zero-level in terms of one’s preference to continue to exist or not, we should think about what kind of life we would comfortably create. When solely considering the welfare of a new person, we wouldn’t create them if they were to live a life of negative welfare. Therefore the absolute worst life we would be willing to create would be one of zero welfare.
The creation test is an interesting proposal, but how useful is it? Is there a good way to determine which lives we should be willing to create? As philosopher Christopher Cowie points out, the natural approach would be to ask whether the potential person’s life would, in terms of its own welfare, be net good or bad. The problem with this approach is that it would render the creation test redundant. If we want to say that being willing to create someone for their own sake is the way to figure out if that person has positive welfare, we can’t then use positive welfare as the starting point to decide if we are willing to create someone - this would be circular. If the creation test is to be useful then, we would have to determine if it is acceptable to create someone for their own sake, without thinking about their quality of life. This seems not only impossible to me, but even slightly ridiculous. I agree with Cowie therefore that the creation test doesn’t seem useful.
Simply asking people if their life is good…
Why don’t we just ask people about the quality of their life? We could ask people various questions about the characteristics of their life and then if they think they are living a positive, negative, or neutral life. This might help us start to paint a picture of a neutral life. Note that this approach wouldn’t run into the problem Sidgwick did - in theory at least, people who are on the brink of suicide could honestly answer that they are living a negative life, and that they would kill themselves if it weren’t for the badness of death itself.
Unfortunately in practice I suspect that there are strong social pressures for people not to report that they are living a negative life, even if that is how they genuinely feel. Social desirability bias is a well-known phenomenon in social science research. For this reason alone I don’t think simply asking people if they are living a positive, negative or neutral life is a promising approach.
There are other potential issues. In his book, Better Never to Have Been, David Benatar raises various psychological tendencies which he argues tend to make us overrate our lives. This includes a phenomenon of “Pollyannaism” whereby people are inclined towards optimism in judgements of their life, with a tendency to recall positive experiences with greater frequency and reliability than negative experiences. Indeed Benatar notes that “most people believe that they are better off than most others or than the average person”. Benatar also argues that, given enough time, we adapt to bad situations such as being confined to a wheelchair, and then report similar attitudes towards life as before they occurred. If this is the case, we may not be able to trust self-judgements. I don’t find this too hard to accept given the well-known phenomenon of Stockholm Syndrome.
Benatar’s arguments have come under criticism, but I think are likely to have at least a grain of truth. In any case, I think that the problem of social desirability bias renders ‘simply asking people if their life is good’ a highly flawed approach.
Hedonism and experience sampling
We need a framework in which to think about quality of life. The simplest one is hedonism, whereby the quality of life is entirely determined by the quality of the hedonic states it contains. These states can either be good (‘pleasures’) or bad (‘pains’). The greater the duration or intensity of a good state, the better, and vice versa. Under hedonism, a zero-level life is one where goods and bads are perfectly balanced, or in which there are no goods or bads at all.
Finally we might be getting somewhere. A life in which there are no goods and bads at all isn’t too hard to imagine - it could be one in which someone is born in a coma and remains in that state for their whole life. I would feel comfortable calling this a zero-level life. The problem is that this doesn’t really help us when it comes to determining the quality of lives that currently exist, or that we may produce. After all, a never-ending coma isn’t exactly a common way of life…
What about a life in which goods and bads exactly cancel each other out? Such a life could take many different forms. In a graph with quality on the vertical axis and time on the horizontal axis, a neutral life would be any one in which the area below the horizontal axis exactly equals the area above it. For example:
In the above example, there are periods that are neutral. This may be when there is no experience whatsoever, such as a dreamless sleep, or it may be when experience simply has a neutral quality, such as if someone were lost in thought and daydreaming about something entirely banal. There are also periods of minor pleasure, such as when one eats a nice snack, and minor pain, such as when one hits their knee. Finally, there are periods of high pleasure, such as when one has sex or feels love towards their partner, and high pain, such as when one loses a loved one or suffers from depression.
In the above diagram, the goods and bads exactly cancel out, but would this be the case for a typical life? Certainly most lives will have periods of neutrality, minor pains and pleasures, and major pains and pleasures. But can we start to quantify these?
Ideally we would have Edgeworth’s hedonimeter: a machine that could continually register the strength of one’s pleasures and pains over time. Of course we don’t have such a machine, but interesting steps have been made in this direction. In Experience Sampling Method (ESM) studies, participants are notified a number of times over a given period—say, eight times a day, for seven days—and are asked about aspects of their mental state, where they are, what they are doing, and who they are with. People can log such data through their smartphones. Such studies have reported that people (unsurprisingly) are happiest in natural environments and when socialising with friends but, to my knowledge, haven’t been used to try to determine if people are actually living net positive lives, or what a zero-level of wellbeing life might look like. I think this is an important avenue for further research.
An ESM approach has clear differences to ‘simply asking people about how good their life is’. For starters, you are asking people how good the moment they were just living is, not their whole life. This judgement will undoubtedly be less prone to social desirability bias - most people are happy to admit that they are sometimes in a negative mood. One also doesn’t have to think back over a long period which, as Benatar argued, may fall prey to Pollyannaism.
There may of course still be biases associated with ESM, but it seems to me that ESM may be the most promising approach to constructing a realistic picture of a perfectly neutral life, in which the goods exactly cancel out the bads. I would love to see more ESM studies, and would happily participate in one myself - even if it means potentially finding out I’m living a far worse life than I thought I was…
How would you aggregate within a life under hedonism? Can we take for granted people's intuitive judgements of felt hedonistic intensity? I lean towards no, and I think there's actually no (at least precise) fact of the matter. I can imagine it being the case that cardinal hedonistic intensity assessments are created by parts of the brain that aren't responsible for the hedonistic component of experience, rather than "read off", and judgements would differ between people who differ only in the parts of the brain resposponsible just for the cardinal assessments.
Maybe one promising approach is a mixed experience test to match intensities of positive and negative valence, although this wouldn't give us a cardinal scale, only a scale where we can say some pains are about as bad as some pleasures are good. What I have in mind is having someone experience both negative and positive components at the same time, and checking which dominates their attention, or how the person compares the overall experience to a hedonistically neutral one. However, I still have doubts about whether this reflects what we want. Mixed states could be too different from purely positive or purely negative states (e.g. attention may work in complicated ways), so judgements may not reliably generalize to them from this test.
If you want to just use people's preferences over hedonistic components of experience, then this is less paternalistic, and judgments will vary widely. For example, I don't think it's true that more intense pleasure is necessarily better, all else equal. That being said, I lean more towards a preference view anyway.
Would love to see more research on this!
This is an incredibly important question. It is also an incredibly dangerous one. There are many real EA whose views on this topic constitute either an X-risk or an S-risk to EAs with only subtly different assessments: people who, given a truly aligned omnipotent AGI would either wipe out the majority of humans or create many lives others view as unhappy. Historically, well-intentioned eugenicists have killed many people who self-identify as having worthwhile lives.
I also think there is a miscalibration in the creation test; many humans instinctively view people similar to them as competition, and either like or dislike the idea of clones of themselves for other reasons. The advantage of the suicide test is that you are centering your judgement on a real person who can express their actual preferences in the moment, rather than a hypothetical case. That seems worth a lot of offset error to me.
Another reason the hedonic sampling and other r self reported treasure will be too positive: We rarely will sample people in the last months of their lives, or who are deeply ill and suffering but incapacitated. The end of life pain seems like a major share of the downside of a life. This also seems unlikely to be picked up in the suicide metric, as the pain may be unexpected and because euthanasia is difficult.
(Same would apply to babies and small children, but I’m not sure what direction that goes in. Babies cry a lot but also seem super happy at times, so much so that I am slightly jealous of them.)
I'd like us to do this more! We could also do small children. It won't work for babies of course.
Relevant: happiness - How happy are people relative to neutral (as measured by experience sampling)? - Psychology & Neuroscience Stack Exchange
For two objections to global desire theory, a theory which would basically take for granted answers to questions very similar to the indifference and creation tests to compare a life to neutral, see Life Satisfaction and its Discontents by Michael Plant:
Of course, if automaximization is a problem, then so would be directly manipulating brains. You can still have them make cognitive judgements weighing up good and bad in their lives, but you get to choose the weights for them and perhaps what they consider good or bad. Basically wireheading.
Don't buy that this is important? Will MacAskill raises it as an important research question in his EA Global Fireside chat at the 31-minute mark. Can't say I think his proposal of asking people how good their lives are is the best approach though...
I don't buy that what a neutral life is like is an important question.
I listened to a few minutes of the timestamp you linked but unless I missed something, Will is talking about his interest in finding out what proportion of people have lives above and below zero, not what a neutral life is like.
I don't see any tight connections between the value of finding out more about neutral lives and what implications that might have for efforts to reduce existential risk or other longtermist efforts. It's more related to the important question of saving lives vs reducing suffering, but I don't see any clear implications here either. If you spell out what connections you see I might be more convinced.
It seems to me that the ethics of having children and the question of antinatalism are swamped by many considerations besides what a neutral life is like. Again, if you spell out the connections you see here I might be more interested.
I hope this is useful feedback!
Thanks this is useful pushback. I didn't want to go into this detail in the blog post to stop it being too long, but perhaps a separate post could go into this as it is important!
My main response is that understanding the neutral level might be useful to determine how many people currently do / will live above/below the neutral level. For example this paper sets a critical (neutral) level in terms of per capita yearly consumption and then uses this level to judge if global social welfare is increasing, by understanding if additional people have been/are living positive or negative lives. If the critical level is too high, additional lives are negative and we may want to reduce suffering rather than save lives (from a shorttermist perspective).
Similarly, understanding the critical level can help us understand if future additional people are likely to live positive or negative lives, by comparing what we expect their lives to be to this critical level. If future beings are likely to be near/at/below the neutral level, we may find that improving the future conditional on it existing (e.g. reducing s-risk) is far more promising than ensuring the future does exist (e.g. reducing extinction risk). Some writers, like David Benatar, argue that the neutral level is likely to be very "high" (compared to what we naturally think it is) - if he's right approaches to reducing extinction risk may be misguided.
Otherwise, understanding the neutral level may actually be directly relevant with regards to the repugnant conclusion in population ethics. Essentially some have argued that if a neutral level is much better than we might naturally think, the repugnant conclusion loses its repugnance. I won't go into that in too much more detail now but this is a relevant book that covers this question.
I would note that even if understanding the neutral level isn't important to understanding how many people currently do / will live above/below the neutral level, I'd hope my post is still useful as pretty much everything I'm saying is still relevant to that latter consideration. For example, the experience sampling approach, asking people if their lives are good, observing how many people commit suicide, and thinking who we would happily create can all in theory be used to understand how many people are above/below neutral (although I specify in my post which I think would be the most promising approach).