I agree it would be interesting to gather specific data on such questions by asking people. A lot depends on the actual life of someone who evaluates their life satisfaction at 2/10. Do they experience a great deal of suffering, though not extreme/ unbearable? Or are they reasonably happy much of the time, but see no hope of improving their material wellbeing, especially when they see the kinds of lives people have in richer countries? 2/10 may mean different things to different people. The life satisfaction measure is just a momentary cognitive assessment, whereas strong negative affect, experienced over long durations, is in a sense more real. Trying to eliminate the very worst, literally unbearable experiences – in humans and in animals – seems to me very reasonable.

You could also ask the person with 2/10 life satisfaction whether they would agree to live in a rich country but have excruciating attacks for several hours a day for at least 2-3 months/year. And for the preferences elicited with such questions to really be meaningful, the respondent would have to know what it's really like, and to be able to compare stretches of both experiences. Brian Tomasik has referred to different utility functions of people imagining and even consenting to experience extreme suffering, and people actually experiencing it. I doubt many people would choose cluster headaches after experiencing them, regardless of the rest of their life circumstances.

In practice, of course, it isn't a binary choice anyways, and we can devote resources to improving access to effective treatment for cluster headache while trying to alleviate global poverty.

Thanks for your answers and links, @David_Moss. The key point, as mentioned in the post and by David, is the distinction between a hedonic experience (affect) and an evaluation. But as to why people would still rate their life satisfaction at ~5/10 when they are regularly experiencing extreme suffering, I would expect that those who have learned to stay positive and look on the bright side of their life are better able to cope – to handle the pain when it happens, and to function well and experience moments of joy the rest of the time. As I mentioned, some people with cluster headaches have a strong support network within the patient community. Someone suffering alone for long periods with a severe mental illness may view their life very differently, and tell themselves a different story, than someone who is supported and encouraged by a community of fellow "warriors", even if the second person still experiences the suffering as extreme.

In terms of people also caring about other things besides their hedonic states, I would argue that people ultimately care about things because they think these will make them happy, or give their lives meaning and thereby lead to happiness. They value the happy moments they experience and tell themselves that these contribute to their life satisfaction. This doesn't negate the fact that the suffering may be literally unbearable at times.

A couple of specific responses:
-I personally don't think that "a life worth living" is all that useful as an objective concept. Most importantly, it could suggest that it's still worthwhile bringing future people and animals into existence who will experience unbearable suffering - who pass involuntarily through bottlenecks of unbearableness. The notion itself is a reflection of our biases.
-While I don't think the Cantril ladder is necessarily perceived as strictly linear, I think you're right about the distinction in that the midway point may reflect a perceived tipping point between net-positive and net-negative. Again, that's just the perception of a person evaluating their life, with their inherent biases, not a determination that filling the world with people who experience unbearable pain part of the time is a good thing.

I think the most important takeaway is not to reduce our consideration for suffering because people's life satisfaction scores could be worse, but the contrary: to treat extreme/unbearable suffering as a matter of urgency, even if people are doing their best to maintain a positive perspective.

Thanks, Marius. Some responses:

Re: numbers vs categories: I largely agree with you. Numerical scales such as the one we used for suffering can be used intuitively to represent intensity, and the numbers can be stand-ins for categories. But problems can arise when the numbers are misinterpreted as being points on a linear scale (which it is not), or that they represent (even non-linearly) quantities of a defined unit that can simply be aggregated in various ways to yield a single value. Having separate categories of severe and extreme suffering is a way of avoiding these problems. I very much agree with the qualitative distinctness of suffering so extreme that giving up one's life is seen as preferable in that moment. For prioritisation purposes, there doesn't seem to be any need to try to distinguish between levels of unbearable suffering (as you imply) - they all deserve very high prioritisation. In principle one would never want such experiences to occur.

Re: correlating suffering and EQ-5D-5L: Yes, I agree it would be good to compare existing measures with the suffering score. EQ-5D-5L has scores for pain/discomfort and for anxiety/depression, so we would expect moderate-to-strong correlations of suffering with both, if data were collected in the same way. But for health economics, EQ-5D-5L is used especially to quantify overall utility, whereas we are arguing for separate one-dimensional suffering metrics that directly measure the key ethical parameter: experiencing an unpleasant state and wanting to be free of it. Our approach is also to try to capture extreme suffering that wouldn't necessarily be captured using momentary measures of wellbeing, such as with EQ-5D-5L. So there's a distinction not just in the scales themselves, but how they are potentially used.

Re: how they bear this suffering: We did, in fact, include such a question: "Is there anything that worked well to alleviate your suffering from this condition or situation, that you think other people should know about?" But I didn't include responses in this post, as it would have made it too long.

Re: the comparative approach: Agreed, it wouldn't be equivalent to the DALY for the reason you gave (no tradeoff exercise), just allow rankings.

There was a small trial that was recently completed at Yale. The administrative hurdles are greater, including DEA approval in the US, but certainly not insurmountable. It might be easier in some other countries with more permissive laws and where psychedelics have already been legally prescribed, like Canada and Switzerland, but approval is still necessary.

I think it's great that you did this analysis. There's a strong tendency, including among many health economists and rationalists, to want to use one single metric to cover everything that matters and aggregate all the data to get a single number (utility, wellbeing, suffering...). This makes it much easier to make decisions, since you're just comparing potential outcomes based on single numbers. The problem is that not everything can be added together in a way that is both meaningful and non-arbitrary, even if you introduce different weightings. In this case, while aggregating different intensities of suffering to get a single number can simplify things, it loses critical information. So focusing on something more narrowly defined and maintaining a degree of granularity in the analysis ensures that the worst suffering isn't neglected against a background of more widespread but less intense suffering.

If you put all the existing evidence together – which is much more than just those two studies – even with the risks of bias, it really is overwhelming. See, for example, Fig. 2 in our policy paper, based on a paper by Schindler et al. https://www.preventsuffering.org/wp-content/uploads/2020/11/Legalising-Access-to-Psilocybin-for-Cluster-Headaches-Policy-Paper.pdf, and also some of the patient testimonials. The usual standards for determining the effectiveness of new medications are understandably more rigorous than for other kinds of situations, but in the case of cluster headaches, the huge number of patient reports claiming the efficacy of psychedelics (while other substances like cannabis are generally ineffective), including chronic cluster headaches that end after a few doses of psychedelics, essentially excludes the possibility that there is no causal relationship. And the extreme pain means that patients have the right to access them even if there were any doubts. We're actually preparing a case series for submission, with a Zurich-based neurologist who can legally prescribe psychedelics, on the results obtained with her patients.

There's a real need for large clinical trials. There have been a few on psilocybin and LSD as preventatives. The big obstacles are recruiting a sufficient number of patients and obtaining funding to study substances that aren't directly patentable. Demonstrating the efficacy of DMT as an abortive compared to placebo could be done on very few patients and reach high statistical significance. Demonstrating greater efficacy than Sumatriptan - a standard abortive - would be more difficult, as the latter is also fast-acting and effective in the short term. Sumatriptan is widely believed among patients to cause rebound attacks and to lose effectiveness over time, so a proper comparative study would probably need to follow patients over a period of months. I don't think that DMT would have to be legal (e.g. for personal use) to be studied as a controlled substance - just authorisation would be needed.

There is this write-up from 2020: https://www.medicusmundi.ch/de/advocacy/publikationen/mms-bulletin/palliativversorgung/advocacy-fuer-palliativmedizin-und-pflege-ein-weltweiter-querschnitt/how-a-local-champion-can-bring-the-government-on-board

More recent updates in our December 2022 newsletter: https://mailchi.mp/106e8af12947/opis-december-13469449

Yes, I  agree. Asking people who don't suffer from a condition to evaluate it is already an imprecise approach, and it just doesn't work for extreme suffering. I think self-evaluation is essential. A study on people with cluster headaches who had also experienced other sources of pain, including childbirth, kidney stones and gunshot wounds, provides a means of normalising the scale by re-setting the meaning of 10 (we reproduced the graph in our policy paper on cluster headaches). The measure was pain intensity, but the same approach could be used for suffering. Qualitative descriptors of different pain/suffering levels can help ensure that people mean the same thing in their self-evaluations. In my opinion, time tradeoffs can be helpful for ranking intensities of suffering, but I don't think they can substitute for direct intensity evaluations.

Thanks for writing this report, which is really well researched!

One point I would stress, which you also hinted at, is the inadequacy of current metrics, including the WELLBY (though a definite improvement on the DALY and QALY), to properly account for the reality of extreme suffering. While 0-10 scales are common and useful, they impose the compression of a complex phenomenon onto a superficially linear scale that then gets treated as such, as if the different points on the scale represent equivalent increases in a cardinal unit. Pain and suffering that are so severe as to prompt suicidal ideations and attempts have a qualitatively distinct aspect that isn’t captured by such scales – probably not even if we interpreted them as being logarithmic. For this reason, reducing pain or suffering from a true 10 to an 8 is probably far more significant than reducing it from a 3 to a 1 – even if the timescale of the extreme suffering is much shorter.

For example, someone with cluster headaches might find the attacks on the verge of unbearability while experiencing them, similar to torture. A temporary drop in SWB during the attacks - even if it were to zero - wouldn’t sufficiently reflect the degree of agony. Furthermore, during attack-free periods, life satisfaction might be evaluated as relatively high and not reflect the suffering experienced during attacks.

Since interventions to reduce pain and other causes of suffering are ultimately aimed at reducing the phenomenon of suffering itself, there’s a strong case for new metrics to measure it more directly than the WELLBY does. In their 2017 paper, the Lancet Commission on Palliative Care and Pain Relief proposed the SALY (Suffering-Adjusted Life-Year). In my new book The Tango of Ethics, I suggested that the SALY could be reduced to YLS (Years Lived with Suffering), and I also proposed two additional metrics that could better account for the reality of severe and extreme suffering without being diluted by aggregation with moderate suffering: “Years Lived with Severe Suffering (YLSS) could capture suffering at the level of approximately 7/10 and above. A separate metric called Days Lived with Extreme Suffering (DLES) could capture the most urgent suffering at the level of approximately 9/10 and above, and properly account for it even when experienced on short timescales.” These metrics, and especially the latter, would ensure that the most severe suffering is not overlooked in public health interventions, and would allow us to better track states that have the highest urgency.

Of course, using additional metrics would complicate cost-effectiveness comparisons of interventions that have been evaluated in other terms, such as improved life satisfaction. But this would not be a valid reason not to use them, especially as parallel measures. If we want to address suffering so extreme that it causes people to kill themselves to escape it, we need to track it more directly.

A couple of specific points regarding interventions:

-Paying for the purchase of opioids could indeed help relieve the burden to healthcare systems in low-income countries in cases where palliative care and access to morphine are already a reality, but cost remains a barrier even after other obstacles have been lowered (logistics, training, regulatory…). Note that, because they are subject to international control, opioids in these countries are usually purchased centrally through the government and not by independent healthcare institutions.

-Advocacy for the legal provision of psychedelics to treat cluster headaches and some related conditions, which we are already engaged in, might be considered more cost-effective according to a metric that attributes particular importance to extreme pain/suffering, even if we can only provide a rough estimate of the impact of such campaigns.

(One minor point: it wasn't clear to me why going from 0 to 10 on a pain scale represents an 11-point change.)