Sarah Perry, contributing editor of Ribbonfarm, and killer risotto maker (no joke, check the book), wrote a book about suicide and antinatalism some years ago. It hasn't got much attention, which makes it an excellent choice for one of my blogposts, following my policy of not writing about what others are writing about.

The book I'm about to review deals with suicide and antinatalism, the philosophical view that bringing more people to the world is immoral. In my review I will only cover the first part. Altough I've read the whole thing, I want to write an antinatalism post later on, and I will cover that part there. In terms of chapters, this post covers chapter 1-8, plus the Appendix.

A summary of the first part of the book in one or two lines would be that suicide is a good choice for some people, and that biases of various sorts make it difficult for many people to see this.

First, you can find previous reviews here, here,here , and here. I'll try not to repeat things already mentioned there, so read those first.

(I) Suicide as vanilla ice cream

Every Cradle is a Grave opens with one of the best captatio benevolentiae that I've seen. This is when the authors asks the readers to take the content of the book seriously, discussing why they might decide not to do it, and give them reasons to do it. This is important for a book that discusses a controversial topic such as suicide, and more so when the conclusions are on an even higher order of controversy. You can get an idea from some of the first paragraphs,

This is a book about ethics. People don’t often change their minds about ethics. When they do, it is generally for social reasons, not because they are exposed to reasoned argument. Reasoned arguments more often allow people to cement their existing opinions. Ethical beliefs are, in any case, extremely limited in their ability to influence actions. I will advocate several ethical positions that are counterintuitive, and that some people would describe as evil. These ethical positions include the view that life—not just human life, but all life capable of having experiences—is very bad. It is very immoral, I will argue, to have babies or to otherwise create aware beings. I will also argue that suicide is not wrong or a product of mental illness, but an ethically privileged, rational response to the badness of life. [...]

I suspect that I have made more converts to the cause of questioning life’s value simply by being an adorable house- wife who makes a killer chanterelle risotto than by any particular argument I’ve constructed. Since I can’t make you risotto, I have tried to present my arguments in a calm and reasoned manner, with abiding respect for the human- ity that we all share. Perhaps I will come across as the sort of cartoon villain you should accept as an epistemic peer. But whether or not you allow me to influence you with my dangerous ideas, I hope you will believe me when I tell you that I am very much on your side. You are, after all, an aware being having experiences. This is true whether or not you have had or will have children, and this is true whether you want to live or want to die. Thank you for reading my book.

My own approach to suicide, which is worth presenting here, is that suicide is like vanilla ice cream. Some people like vanilla ice cream, others do not. It is good that people who never have eaten vanilla ice cream at least try it to ensure that their rejection makes sense and they are not missing out something they could like. And it would be wrong to force-feed someone vanilla ice cream, of course.

Likewise, with suicide, some people like living, and other people don't. It is good that people who don't like living make sure to think twice that they don't really like living, plus that they consider if they will like living in the future. If someone has gone through that thought process, it would be wrong to stop someone from committing suicide.

This assumes that there is no argument in favour of being alive vs being dead beyond one's own preference. Utilitarian reasoning could support something like that: Perhaps, even if you really dislike your life, you should keep on living if that increases net aggregate utility. But I'm not an utilitarian, and it's hard to argue that one should be a miserable slave just because the world would be better with you in it. Note that one can also devote oneself to others, and even suffer when doing it, and yet think that such a thing is worth doing out of supererogatory considerations.

Or maybe living is good in itself, and one has to change one's way of seeing one's life to adapt it to the requirement that you must go on living. I confess I would like to have an argument for that, but I currently don't, so while I would want to say that you should change your perception and preferences to accommodate living, I cannot rationally endorse it. Yet, at least.

Crucially, unlike with vanilla ice cream, suicide is irreversible. Which is why one would be justified in stopping someone trying to commit suicide out of a whim. The justification here is hypothetical consent, the same that is used to do surgery on someone not conscious that needs said surgery to stay alive: Had you calmly thought about the situation, you would consent to that. How do we know? Because most people who have been stopped from committing suicide are glad that it happened, because they didn't really want to die.

This vision radically contrasts with popular views of suicide, where it is something terrible, regardless of the life the person was living. One must live, whatever it happens, even if that means incarcerating that person in a mental institution until they want to live. Suicide is seen as a cowardly act that must simply not be allowed to happen.

(II) Assorted comments

What follows is a collection of key paragraphs from the book, with some commentary of my own.

But people do not exist as individual units separate from human relationships and groups. A great deal of the cost of committing suicide faced by a person wanting to die is social and empathetic: it is resonant in the loneliness and grief that his death will cause, or at least hasten, among parents,  children, siblings, a spouse, or friends. As social creatures, we begin forming bonds at least as soon as we are born; these bonds, while often no more voluntarily chosen than our own births, are powerful motivations. Those with whom we have formed social bonds….

To me, this seemed a bit odd: If I were considering suicide, I would be thinking on my own future, weighting present misery against future possible improvements. But the vision of suicide that is explained in the book is all about others, about groups. It is strong social bonds, in her view, what keep suicides - or perhaps, all of us? - from actually committing suicide.

Simple misery does not seem to reliably cause suicide, the way it would in Bryan Caplan’s naive model; rather, people

seem to commit suicide when they are freed from, or perhaps rather deprived of, the social bonds that were keeping them alive.

On this view, life would be a prisoner dilemma of sorts: It is because of others that we would be alive. But if we could just let go of those social bonds... then we wouldn't make other people suffer when we commit suicide, and then we would do so. (?) This doesn't seem right to me. But could it be true of other people?

In that paragraph, Perry refers to Bryan Caplan's free disposalmodel of suicide. If you don't like living, you can just commit suicide! You can always see no to the unasked-for gift of life.

But not so fast, says Perry: Suicide is not that easy! Society is far from being a place where suicide is as trivial as undergoing surgery. She illustrates this with the interesting idea of a free disposal society: a society that doesn't view suicide as something evil, but a choice one can make. People would be taught from a young age to see their life is theirs, and that they can end if so they want. One could go to a doctor, ask for an appointment, and after some counselling and discussing, one would press a button to safely and effortlessly inject oneself with some drug that causes instant and painless death. In this society, life would really be a free disposal good. And we are far from living in that society, argues Perry. Committing suicide is hard, even if we want to, because of our biased and irrational ways. For example, if one wants to jump from a bridge, one will start to think of all the pain that it can cause, and the probability that one might not die (Higher than you think!) and thus live the rest of your days in a bed, doing nothing.

Chapter two starts by mentioning that life is something that needs justification. That there are tons of suffering and bad things out there, and that we need a good reason to keep the show going. That is: Why live?. My answer is: but the world is great, and there is not that much suffering. Why not live?

Sarah Perry points that what means is not happiness or pleasure, but meaning. And I agree. But then she makes the move (throughout the book) of saying that meaning-perception is an illusion. Meaning is not really there. And thus if what matters is not even there...

The belief that life is meaningful tends to take the form of a strong feeling rather than a reasoned conclusion; indeed, one of the functions of meaning is to shield a person from the harmful effects of reasoning by providing a value that is justified for its own sake, a foundational rock for cognition below which no whys need be answered

My reply to that: So would be the belief that suffering is bad, for all reasoning begins from premises that seem true to the reasoner, and for which no whys are provided.

The heavy modern self has a hard task: it must do for itself what human religion and community did in the past. It must provide itself with meaning.

This is another truth. Until recently, one could seek meaning in God or the Nation, or the Community. Now, one must answer 'Why am I doing this?' and the only one answering is oneself. And perhaps some find difficulties answering the question. If there is no answer, another question arises: 'Should I continue doing this?'. And if the answer is no, then we have a candidate for a potential suicide.

Chapter three analyses suicide using Jonathan Haidt's Moral Foundations Theory: How people use the different foundations (Harm, Care, Purity, Authority, Loyalty...) to reason about suicide.

Sacredness is necessary for the coordination of human action, for politics, for orderly human life. It is essentially a valuable illusion created cooperatively by the social unit, often over a long time at great cost, and then maintained and defended against mutations and competing sacredness structures. Sacredness secretly informs all of our judgments, even those that seem to be purely related to harm or fairness— indeed, even those that don’t seem to have a moral dimension.

This is one of the paragraphs where we can see themes that Sarah is currently talking about in ribbonfarm (rituals, community, etc). I wouldn't say that sacredness is that important, though, and that many of the rules in place that we have are down to self-interest. Not all: I do think that some altruism is needed for a modern society to work well. But that small core of other-regarding preferences (Here I talk about altruism beyond kin or reciprocal altruism) wouldn't count as sacredness. One could of course make the simile that morality is like religion, and thus moral belief are like sacred beliefs, but philosophy doesn't work like religion, and moral beliefs are subject to critique and discussion. Likewise, a belief that the Law is sacred, that one ought to comply with the Law because it is the Law doesn't seem necessary. A belief that the Law is a tool to be used for coordination, coupled with penalties for breaching the Law seem to suffice.

Chapter four

One of the most salient features of the things that make life seem meaningful is that they frequently rely on illusion: the illusion of unchanging permanence, of a future state of happiness, of one’s ability to affect the world. It is my view that the sense of meaningfulness is itself an illusion, a cognitive phenomenon that is very adaptive for individuals and groups 

[...] The feeling that life is meaningful is a pre-rational sensory perception that is widely shared. However, the specific meanings that people find satisfying and convincing are disparate and often contradictory. These underlying realities should make us question whether the sense that life is meaningful—or that some specific meaning can be found in life—is a true observation, or merely an illusion. […]

The things that we find to be meaningful are, in fact, miniature Experience Machines. They rely on illusion and filter the information that reaches us so that we may continue to feel that life is meaningful, or continue to search for meaning in life if it is missing. They are very useful; they help us organize our behavior, coordinate with others, and manage our emotions. In a practical sense they often make the suffering of life bearable; but, once they are recognized to be illusions, they cannot justify suffering in an abstract sense any more than pleasure can.

While one can attack meaningfulness with the spear of evolutionary theory and argue it is a bias, so we can do the same with suffering avoidance. In fact, suffering avoidance is a more _meaningful _candidate for an evolutionary bias of that sort than having a sense of meaningfulness. But plausibility is not a definitive argument. Evolutionary psychology is supposed to be more than the just-so stories that its opponents so much criticise.

Next in this chapter, the idea of a reverse experience machine is presented. The regular experience machine, a thought experiment first found in Robert Nozick's Anarchy, State, and Utopia, asks the reader to imagine the existence of a machine that you can plug yourself into. If you do, you will fall in a dreamlike state where you will go on living experiencing the best life you can experience in your own terms. Within the machine, you will forget that you are in the dream. Nozick concludes that there are reasons not to plug into the machine.

The reverse experience machine is finding out we are inside the machine. Would we go out?

The Reverse Experience Machine experiment is close to the situation we find ourselves in—if we found out we were in an Experience Machine already, would we choose to leave it for the real world? Institutions, religions, social communities, and even individual people function as Experience Machines, creating and maintaining illusions that help us feel that life is worthwhile. A meaning realist would reject the Experience Machine, but to be consistent he must also reject those aspects of life that use illusion or information filters to provide meaning. A meaning subjectivist has little ground to reject the Experience Machine. This has implications for the justification of life’s misery based on meaningfulness.

This is a section that I didn't fully get, but if my understanding is correct, then I see a bit of an issue: that it is assumed that things that are used as building blocks for meaningfulness (or meaningness ?) are really generators of a fake reality, like the experience machine. But there are differences between the experience machine and the things she mentions: You don't know where you really are while in the experience machine, and you have false beliefs about that which surrounds you. (e.g. you would be interacting with unconscious robots rather than with people, and you would be learning about the simulation, not the actual universe). With institutions and people, we do know what they are, and what they are doing, for real.

Another question is if one should plug or not into or out of an experience machine. It depends on how much you value knowing true beliefs, I'd say. Some people would not unplug from the machine, because they are used to living there. For some people, fake parents, fake friends, and fake life is better than being thrown into a completely unknown world. But not for everyone.

Chapter five discusses whether suicide is something impulsive or something rationally pre-planned. Most people clearly side with the first view, and Sarah gives the reader some evidence, included peer-reviewed research, to the contrary.

One problem with this theory is that suicide does not, in fact, appear to be an impulsive act,32but generally involves a plan. And given the existing barriers to suicide, a suicide that appears “impulsive” may actually reflect the genuine rational desires of the suicide. The person who rationally prefers to die may be unfairly prevented from doing so by legal and practical barriers; he may need an “impulse” to push him over the edge and enact his rational desire.

So what is it? Suicide: fast or slow? I don't know much about the topic myself, so I outsource my opinion to psychiatrist Scott Alexander (One of my trusted authorities on a range of matters), who defends the view that suicide is mostly impulsive, and that people don't really want to do it. I even sent him a recent literature review that argues for the rational suicide thesis, and he wasn't still convinced. You can find his reply here.

Chapter six deals with, among other things, IQ and suicide. You'd think that IQ is negatively correlated with suicide, as most of the good things are positively correlated with IQ, and suicide argueable is a bad thing (And people with higher IQ would be more succesful, hence have less reasons to commit suicide). Sarah Perry agrees, with caveats: There is one study by Lewis Terman on a group with very high IQ ("Terman Termites", N>1500), and in that group, suicidality rates were four times as high as in the average population. The weird thing is that IQ reduce chances of suicide in a fairly predictive matter,



But things are not that easy: If we plot suicide rates with average national IQ, we doget a positive relationship!. One thing, though, is that those studies do not attempt to engage with the literature on individual differences, and just assume that what is valid at the nation level will be valid at the individual level, and we just saw that it is not the case.


If we plot the actual variables, we see that a linear fit doesn't make much sense, and it also seems that suicide rates are first low, then they peak, and then they decrease, with IQ.

Sarah does not decompose the correlations to this level (no one else has, so it's not much of an issue), and says that high IQ societies are more complex, and thus easier to 'fall through the cracks', be alone, lose contact with social groups, and end up committing suicide. Perhaps we could try to see why the relation is the way it is. Let's go for one usual suspect: Post-communist countries, still recovering from the fall of the Soviet Union. (IQ data is from Lynn and Vanhanen (2002), suicide data is from Schmidtke (1999)


The trend remains, and if anything becomes more clear.

One other idea is to try to use data from a different year, to see if the relation is still there. This is using WHO data from 2012,


Now the correlation becomes much much smaller to the point of not really being there. WHO uses age-standardised suicide rates, trying to correct for the fact that suicidality changes with population age.

The countries at the top are, in order, South Korea (106), Sri Lanka (81), Lithuania (97), Suriname (89), Kazakhstan (93), and Turkmenistan (87). The dot at IQ 65 is Zimbabwe. The countries at the bottom are Kuwait (83), Jamaica (72), Azerbaijan (87), Egypt (83), Bahamas (78), and Barbados (78). The three dots at the right end of the chart are Korea, Hong Kong, and Japan.

If you desperately need an R version of these plots and/or you are a fan of Loess smoothing, here you have it :)


Doesn't look very nice, so I tried to run a couple of multivariate regressions using more factors. The results are in the Appendix of this post. The correlation doesn't disappear, but the assumptions that we need for the models (like normally distributed errors) don't hold.

Even with those caveats,  it looks like we have to accept -provisionally, and with low confidence- that there is a -positive- correlation between IQ and suicide rate at a national level, that doesn't exist at an individual level, which is weird, and requires explanation. Furthermore, look at what happens if we remove the three observations to the right (red line). Then you get an increase in suicide, and then a decrease.

More could be said about this (Martin Voracek has many papers about it, and no one else seems to be doing work on this), but the IQ-suicide relationship is something that merits further investigation. Reconciling individual and aggregate-level findings is something that I haven't seen done.

We now turn to the Appendix of the book

But stories are not real. They are constructs that we apply to the universe, but there is no story out in the universe. There is no “gist” or “point” to the universe, as stories have gists and points. We construct meaning to serve our evolutionarily-determined ends, and this is, I think, the most central of all the cognitive biases. [...]

Living outside of any story—living without hope for the future, without the belief that one is part of a narrative—is confusing. It’s hard to get anything done when nothing has a point. For any not-immediately-pleasurable action (or inaction) I contemplate—getting up in the morning, vacuuming, answering the phone—there is no readily-available answer to the ever-present question in my mind: “why?” At least, there is no long-term “why.” [...]

There are no stories in heaven; heaven is all epilogue. It functions as a bookend on our stories; we may even call it the “hereafter,” as in “happily ever after.” There can be no conflict in heaven, so there can be no stories, either.

Why are drugs, prostitution, gambling and suicide illegal, when they clearly give so much relief to suffering people? I think it is because, at a societal level, we are deluded into thinking that happiness is possible, maybe even easy or likely, without these things. I have called this “cheery social policy.”

The fundamental problem with this sort of cheeriness is the assumption that a good life—a pleasant life—is relatively easy to achieve. Cheery people are able to hold such a belief because they are able to ignore—and perhaps can’t even conceive of—the suffering of a significant minority of the population. A good life is not easily achieved for many of us [...]

This is the view from hell. Hell is not the state of experiencing a great deal of suffering with no pleasure to “balance it out.” Hell is popping out of the notion of meaning altogether. And this Hell is the meta-condition that we are all in, whether we perceive it or not.

At the root of my quibbles with the book is what default should we choose for judging life: Why live? or Why not live? As we lack an argument for why one ought to live  if one doesn't want to, this is important. If one looks for an answer to the first question in some sort of cosmic or universal meaning, the answer won't be there, and thus there will be no reason to live. As a Spanish saying goes, don't ask the elm for pears. Don't set impossibly high standards or you'll be disappointed.

Thus if we want to live, we live, and if we don't, rationality dictates that we should commit suicide, even though suicide in practise is difficult. This is what Gary Becker and Eric Posner argue in their 2004 paper 'Suicide: An Economic Approach'. Their model also explains why suicidal people tend to engage in risky activities (smoking or heavy drinking, for example): They have little to lose, and much to gain.

Suicide, then, is one thing we can do with our lives, and it can be a rational decision. I am not convinced that it generally is; I still think it is more of an impulsive thing. A question for future posts would be to see if it is possible to produce an argument in favour of existing vs not existing.


# suicide library(gvlma) suicide<-read.table("suicide.csv",sep="\t",header=TRUE,dec=",") row.names(suicide)=suicide$X suicide$X<-NULL fit1<-lm(Suicide~IQ,data=suicide) fit1
## ## Call: ## lm(formula = Suicide ~ IQ, data = suicide) ## ## Coefficients: ## (Intercept) IQ ## -7.7576 0.1979
gvmodel<-gvlma(fit1) summary(gvmodel)
## ## Call: ## lm(formula = Suicide ~ IQ, data = suicide) ## ## Residuals: ## Min 1Q Median 3Q Max ## -7.7647 -4.4519 -0.9562 2.1717 20.5310 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) -7.75762 8.17359 -0.949 0.3455 ## IQ 0.19786 0.08807 2.247 0.0275 * ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 6.253 on 79 degrees of freedom ## Multiple R-squared: 0.06005, Adjusted R-squared: 0.04815 ## F-statistic: 5.047 on 1 and 79 DF, p-value: 0.02746 ## ## ## ASSESSMENT OF THE LINEAR MODEL ASSUMPTIONS ## USING THE GLOBAL TEST ON 4 DEGREES-OF-FREEDOM: ## Level of Significance = 0.05 ## ## Call: ## gvlma(x = fit1) ## ## Value p-value Decision ## Global Stat 35.488 3.687e-07 Assumptions NOT satisfied! ## Skewness 22.385 2.231e-06 Assumptions NOT satisfied! ## Kurtosis 8.648 3.274e-03 Assumptions NOT satisfied! ## Link Function 2.327 1.271e-01 Assumptions acceptable. ## Heteroscedasticity 2.128 1.446e-01 Assumptions acceptable.
# diagnostic plotslayout(matrix(c(1,2,3,4),2,2)) # optional 4 graphs/pageplot(fit1)


## Analysis of Variance Table ## ## Response: Suicide ## Df Sum Sq Mean Sq F value Pr(>F) ## IQ 1 197.31 197.314 5.047 0.02746 * ## Residuals 79 3088.56 39.096 ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
fit2<-lm(Suicide~IQ+Asian+Nordic+PostCom+European,data=suicide) fit2
## ## Call: ## lm(formula = Suicide ~ IQ + Asian + Nordic + PostCom + European, ## data = suicide) ## ## Coefficients: ## (Intercept) IQ Asian Nordic PostCom ## -8.1010 0.1952 2.7421 0.4542 3.2541 ## European ## -2.2825
## (Intercept) IQ Asian Nordic PostCom European ## -8.1009584 0.1951910 2.7420597 0.4541673 3.2541055 -2.2825034
gvmodel<-gvlma(fit2) summary(gvmodel)
## ## Call: ## lm(formula = Suicide ~ IQ + Asian + Nordic + PostCom + European, ## data = suicide) ## ## Residuals: ## Min 1Q Median 3Q Max ## -10.4348 -3.6348 -0.6308 2.5181 21.0905 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) -8.1010 10.6618 -0.760 0.450 ## IQ 0.1952 0.1225 1.594 0.115 ## Asian 2.7421 3.6195 0.758 0.451 ## Nordic 0.4542 3.2484 0.140 0.889 ## PostCom 3.2541 1.8706 1.740 0.086 . ## European -2.2825 2.3320 -0.979 0.331 ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 6.08 on 75 degrees of freedom ## Multiple R-squared: 0.1563, Adjusted R-squared: 0.1 ## F-statistic: 2.778 on 5 and 75 DF, p-value: 0.02345 ## ## ## ASSESSMENT OF THE LINEAR MODEL ASSUMPTIONS ## USING THE GLOBAL TEST ON 4 DEGREES-OF-FREEDOM: ## Level of Significance = 0.05 ## ## Call: ## gvlma(x = fit2) ## ## Value p-value Decision ## Global Stat 52.384 1.146e-10 Assumptions NOT satisfied! ## Skewness 18.926 1.359e-05 Assumptions NOT satisfied! ## Kurtosis 16.931 3.877e-05 Assumptions NOT satisfied! ## Link Function 9.991 1.573e-03 Assumptions NOT satisfied! ## Heteroscedasticity 6.535 1.057e-02 Assumptions NOT satisfied!
## Analysis of Variance Table ## ## Response: Suicide ## Df Sum Sq Mean Sq F value Pr(>F) ## IQ 1 197.31 197.314 5.3377 0.02362 * ## Asian 1 28.05 28.046 0.7587 0.38651 ## Nordic 1 0.21 0.214 0.0058 0.93962 ## PostCom 1 252.44 252.443 6.8291 0.01083 * ## European 1 35.41 35.413 0.9580 0.33084 ## Residuals 75 2772.44 36.966 ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# diagnostic plotslayout(matrix(c(1,2,3,4),2,2)) # optional 4 graphs/pageplot(fit2)


## Analysis of Variance Table ## ## Response: Suicide ## Df Sum Sq Mean Sq F value Pr(>F) ## IQ 1 197.31 197.314 5.3377 0.02362 * ## Asian 1 28.05 28.046 0.7587 0.38651 ## Nordic 1 0.21 0.214 0.0058 0.93962 ## PostCom 1 252.44 252.443 6.8291 0.01083 * ## European 1 35.41 35.413 0.9580 0.33084 ## Residuals 75 2772.44 36.966 ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# Ggplotting suicide2<-suicide suicide2<-subset(suicide2,IQ<104) rownames(suicide2) <- c() #ggplot(data=suicide,aes(x=IQ,y=Suicide))+geom_smooth(span=0.5)+labs(y="Normalised suicide rate",x="IQ")+geom_point()+geom_smooth(span=0.5,data=suicide2,color='red')

Comments from WordPress