Why are human beings so interested in explaining misfortune?

(Enter our super-competition and win a mega-prize!)

Some time ago, a lady in France had the pleasure of seeing her lottery ticket win the jackpot (several million euros), only to have her dream blown to smithereens by an untoward incident. To establish that a claim is valid, the lottery is legally bound to bring together [a] the computer printout of the draw, the winning ticket and [c] the computer readout from the place where the ticket was purchased. Unfortunately, that establishment (a bureau de tabac for you connoisseurs of things French) had burned down to a pile of ashes, cash registers and computers included, the day after the poor woman bought the ticket. The claim was denied.

A blow indeed, as her life so far had not quite been a rose petal path. She was unemployed, her husband an invalid with no pension, her equally unemployed son and daughter had both turned into alcoholic vagrants. We can certainly imagine her crying, Why?, Why me?

[Note that I am not sure this story is altogether accurate – I recount from memory]

Why think about misfortune?

Why do people the world over think about misfortune, and construct elaborate theories to explain it? Here surely is one of your massive, elephant-in-the-room quasi-universals of culture, crying out for explanation, and (as usual) thoroughly neglected by standard social sciences. In all human groups, it seems, people notice and remember cases of misfortune, tally them, detect regularities – and most important, try to explain misfortune.


Also, in most human groups, explanations of misfortune center on agents, imagined (gods, spirits) or real (relatives, enemies), that brought about the untoward events.

Again, why? Why do people do that?

To us evolutionarily minded folks, these universally available accounts of misfortune are puzzling, mostly because they are false. Nor are they just slightly off target – they are downright misguided. Bad things in the world happen for a variety of reasons, but superhuman agents are not among them. There are no witches making you sick, no bad spirits that make you trip up. Why would our evolved design for a mind include the propensity to focus on and ponder at length totally useless explanations? In evolutionary terms, this is all the more puzzling as such thoughts are not just futile but also potentially harmful. The time and energy spent thinking about mystical causes are wasted for a more productive use of one’s reason.

You may tell me that this is just as true of myriad other cultural phenomena, as people fill their heads with nonsense of no possible evolutionary value – and insert your favourite example here, religious beliefs, ethnic hatred, alternative medicine, etc. Well, you may be right – the culture-as-widespread-nonsense phenomenon is much larger than the present question. But saying that there are other problems of a similar nature does not solve this one – unless you assume there should be a unique solution for all domains of culture-as-nonsense, which I do not believe for a minute.

So let me proceed to the four questions we should address if we want to have a decent model of misfortune expanations.

Question 1. Why agents?

Why are agents so frequently recruited in the explanation of misfortune? There are several ways to account for untoward occurrences. One type of explanation is your common covering-law kind of generic causal statement, whereby ordinary impersonal causal processes are involved in producing a specific outcome. The bureau de tabac burned down betcause it was full of flammable stuff, and a small flame (perhaps a cigarette butt) started a fire. Another type is a kind of karmic accounting, where bad things are the outcome of some kind of fault. The place burned down because the lady (or her ancestors) had committed some moral violations in the past. The third model is that an agent was involved. Somehow a spirit or god decided to burn down that place. This latter, agency-based account is by far the most frequent. Why is that the case?

Question 2. Why “why me?” ?

This is another universal feature of misfortune models – they explain, not a generic set of causal processes that would account for the type of event that occurred, but the particular token that is being considered. Or, if you prefer less jargon, consider the most familiar example from classical anthropology. Among the Zande, when the roof of a mud granary collapses, everyone considers this must be a case of witchcraft – bad people are involved. In case you feel superior and smugly inform those benighted Zande that roofs collapse when their pillars are thoroughly gnawed by termites – well, they know that perfectly well, only that is irrelevant – witchcraft is mentioned not to explain why roofs collapse, but why that particular one collapsed at that particular time. I know viruses cause diseases, but wy did it have to happen to me? Why me? Why now?

Why do people ask such questions? I hear you say, of course people want to know why it happened to them, of course that is universal – what could be more natural? Who cares what makes other rooftops collapse? Who cares what triggers diseases in other people? What people want to find out, of course, is the why of this particular roof collapse or disease, the one that affects them.

Now, where does all this of course stuff come from? What is so natural here? All this may seem natural to us… simply because we are human too, but that is all the more reason to try and explain it.

Question 3. Why this asymmetry between good and bad fortune?

This may be simpler to solve (indeed the solution may well be obvious) – still, this is one of the questions a good cognition and culture account should address. Most people in the world construct elaborate explanations for bad things while in many cases they are happy that good things just happen.

Question 4. Why are only some occurrences explained in agentive ways?

In the bad good old days of classical anthropology, people with a magical, primitive or prelogical mentality did everything the prelogical or magical way. They were peasants, barbarians, savages – in other words the unclubbable. But as Evans-Pritchard and many others pointed out, all these people also have causal explanations of the more sober, covering-law kind. True, witches will destroy your granary, but granaries cave in also because of termites. Indeed, in most human groups there is an explicit distinction between “simple” or “straightforward” misfortune, which requires not much explanation beyond a recognition of the generic causal processes involved, and those “special” occurrences that seem to cry out for an agentive, karmic or other explanation. During my fieldwork, I learned that Fang people in Cameroon considered some illnesses and assorted misfortune as “simple misfortune”, to be explained for instance in terms of (local models of) physiology, while others were “special”, recruiting the whole panoply of spirits and ancestors.

Why do people maintain both kinds of models? And more important, are there any recurrent differences in the kinds of events covered by these types of explanations?

My solution and our competition for a MEGA-PRIZE

I have found a marvellous solution to all these questions. Unfortunately, the space of this blog is too small to contain it. So I reserve its full publication for another occasion.

In the meantime, why not let a hundred flowers bloom and a thousand schools compete? This is why ICCI proudly opens a competition for the best evolution-compatible, human-cognition-driven, empirically testable explanatory model of these four features of human reflections on misfortune.

Competition regulations: 1. Only send contributions that would address and answer all four questions above. 2. The winner will receive a prize of US$42, offered by Pascal Boyer, in the form of a voucher for use in their favourite online bookshop. (I offer this precise sum because that’s the amount of a reviewer’s fee that I got and absolutely did not deserve). 3. Pascal Boyer is sole judge of all entries. His judgment is thoroughly subjective and may be swayed by friendship, reputation, good looks, bribes and neural misfirings. The judgment is final and unmotivated.





  • comment-avatar
    Karen Lofstrom 13 August 2011 (07:36)

    Why agents? Because human intelligence evolved in large part to solve problems of social communication and cooperation. Because “when you have a hammer, everything looks like a nail”, we tend to frame difficult problems as problems of communication and cooperation with human-like entities. Why me? Because communication assumes a self and an other or others. Also, because we fall readily into the delusion of a unitary, enduring self. Perhaps the two are connected? Hmmm. Why focus on misfortune rather than fortune? Because we have a cognitive bias towards avoiding misfortune rather than taking a chance at a reward. This is evident in every study of human processing of probabilities. We want to avoid risk, so we study it and come up with explanations that we hope will suggest useful strategies. Why only SOME occurrences? We focus on those occurrences for which we have no other coping strategy. If I feel that the situation is under control, I don’t need to invoke the gods. God/gods represent all that can’t be controlled by ordinary means.

  • comment-avatar
    Hugo Mercier 21 August 2011 (23:47)

    Thank you very much for this little puzzle!
    Here’s a very short version of the answer I’d suggest, without having
    given it quite enough thought.

    What we’re dealing with here, I think, is a mechanism of social
    cognition. It is designed to make sure that any suspicious misfortune
    that has befallen us is not due to intentional malice (or maybe to
    someone else in general). Why do we need such a mechanism? Well,
    because people are not always nice, and because more general
    mechanisms to understand bad stuff are not quite fit to figure out
    other people’s shenanigans (as opposed to nature’s tricks for
    instance). In a way, the question “Why agents?” answers itself then:
    agents are invoked because that’s what the mechanism does, it looks
    for culprits among agents. The underlying, more interesting question
    is: why do we over attribute misfortune to agents? I think we do that
    because of the relative costs of mistakes, but also because of the way
    we can figure things out. Let me expand on that.

    Many cognitive mechanisms are dedicated to understanding the bad stuff
    that happens to us. We have specific mechanisms to determine if we’ve
    eaten something bad for instance. Sometimes these mechanisms draw a
    blank. Or we have other reasons to suspect that someone could have
    been involved (we know someone is trying to harm us for instance).
    Then another mechanism quicks in. It’s goal is to find the culprit.
    It’s not a fair mechanism: it assumes that someone is guilty. If it
    assumed that no one was, it would have no point. More importantly, the
    costs of thinking that someone is guilty when she isn’t needn’t be
    that high, if we are able to revise our judgment (as opposed to the
    costs of thinking that no one is guilty when someone is, which can be
    very high). So I may start out by suspecting someone, but that person
    comes up with a good reason (an alibi, say) that she isn’t the
    culprit. My initial mistake was nearly costless, and I can turn my
    suspicions towards someone else (or give up). If I hadn’t suspected
    anyone, I would not only have been more likely to keep being tricked,
    but I would also have acquired less information and have been less
    likely to form an accurate assessment of my partners’ intentions.

    Such a mechanism can also explain the “why me?” effect. The goal of
    the mechanism is to find who did that to us. It assumes agents with a
    specific attention: Phillip wanted to burn my house! Here the
    ‘covering laws’ are that of theory of mind: we have a naive
    understanding of why people may want to harm us, how they could do so,

    Now to the asymmetry between good and bad fortune. Well, that one is
    easy within a social context. When someone harms us, the culprit
    usually doesn’t come forward to indict herself. So we’d better assume
    someone did it and then make sure we’re right. By contrast, when
    someone does something nice for us, it’s rarely a secret: she’ll quite
    spontaneously come forward and tell us. So we needn’t worry about
    trying to figure out who did nice things for us: if someone is
    responsible, we’ll know.

    Finally, why does that only apply to some occurrences? As I mentioned
    above, this mechanism only quicks in when other explanatory mechanisms
    fail, or when we have some external reasons to suspect foul play.
    Where that line is drawn will depend on the causal models that someone
    already has: things that fall out will more easily trigger the
    ‘whodunnit’ mechanism.

    A last word, about gods and spirits. One thing that may explain their
    persistence as (perceived) causes of misfortunes is that they can’t
    talk back. If they already exist, as agents, in our cognitive system,
    then if all the other potential suspects have managed to exonerate
    themselves, we can always blame the gods. At least they won’t come up
    with a good alibi. Unless they have priests to defend them…

  • comment-avatar
    Radu Umbres 22 August 2011 (12:01)

    I should be writing my thesis, but 42 bucks are 42 bucks, even in “special-purpose currency” 🙂

    More seriously, this is indeed fascinating and would like to suggest some revisions and tentative explanations.

    I would start from question 3 which I do not find simple at all. In fact, I believe this is the cornerstone of my hypothesis. I think we cannot explain just representations of “bad” fortune without explaining the mechanisms behind “good” fortune, or better said, they are so similar that if we get one right, the other should follow. There is no asymmetry from my point of view. Good things don’t just happen, they happen because we are “lucky” or “fortunate”. It is our agency or the agency of someone well-disposed to our welfare which makes it happen. “Karmic” speaking, it happens because we are virtuous. “Supra-naturally speaking”, a god or a spirit gives us an extraordinary gift, often after a generous sacrifice or donation. A short example from my fieldwork. Hailstorms hit fields of people working on a holy Sunday. Explanation: God is angry for disobeying the Sabbath. On the other hand, people often place lottery tickets near holy items such as icons, preferably Church ones, hoping for a “lucky” win. Something similar comes to mind with Taussig’s baptized money.

    I would then move to question 2. I am not sure it is just “me”. I think it is used to explain other events as well which involve significant others. In case of the Azande, it is not an individual explanation, it is socially salient. Everyone knows about falling granaries and anyone would think the same thing “it wasn’t just hungry termites behind the collapse of John’s granary”. Of course, the personal level makes it more important when it’s your granary and you will be the one taking actions to find the cause. What makes it culturally Azande is that most Azande would agree to your representation of causality and consent to procedures detecting the guilty sorcerer. But I think that the “me” factor is important in another causal mechanism. It is the aggrieved party which keeps the representation alive by recalling it when a suspicious event occurs. It is about me sharing my explanation with others, teaching it to my children, confirming that “your” case happened to “me” too and I will inform you about my theory of misfortune.

    Now, I find questions 1 and 4 related. Usually, it is only “some” occurrences because there are “some” special agents. Let me expand a little. Misfortunes have a cause, just like anything in this universe explained causally. I don’t want to discuss about what causality means, and I am just using it in terms of “A happened because of B” when A couldn’t have happened if it weren’t for B. In other words, things don’t happen “just because”. B can be defined in many ways, but it often happens that it is traced to an agent. Even in Pascal’s example with the lottery booth – why was the cigarette butt on the floor? Someone must have thrown it there. But why that special day when the poor woman needed that booth more than anything else in the world? Even a sceptical observer will look for agency of some sort, even it is not necessarily human or supernatural. I will return to this point.

    And this takes me to the type of occurrences which are explained by recourse to misfortune. They are “black swans”, events with a very small probability of occurrence with disastrous/fortunate consequences. People dying under a collapsed granary, hailstorms, lottery wins, booths burning down together with million euro tickets, etc. This is my first hypothesis which can be tested empirically. Things which are more likely to happen AND carry more important consequences will attract fewer explanations in terms of agency than others. A collapsing granary is “agentive” misfortune more than dying from poisonous snake bite when termites kill people every ten years while snakes kill them every other week. But, if similar snakes bite people all the time but only one in a thousand dies while everyone else has just a bruise, something else than the snake must be involved. Or people think it is, before I get carried away by my own magical thinking.

    I have a problem with the distinction between “karma-driven” and “god-driven” misfortune. I think there are three types. One of them is pure “karma” which resembles probabilistic thinking in balancing events. A good example would be a large lottery win followed by a dying crashed by someone committing suicide by jumping of a roof. The latter misfortune is explained by the former fortune – the universe is in equilibrium and rare luck is followed by rare misfortune and the other way around. It could be a form of divine justice, but the causal mechanisms are relationships between events and not between agents.

    Pascal’s example of moral violations is usually the “sin” explanation which is the second type. There is some sort of a covenant which has been broken with dire consequences. The explanation of misfortune is divinity settling accounts as a good accountant who never forgets and never forgives. But here I would also put witchcraft performed by another human being. As much as I remember for E-P, the witch acts unconsciously, but it is driven by a reason, there is an imbalance in the social relationship between the witch and the bewitched. That is how the Azande find the suspect and perform the inquisitive ordeal. This type is the most amenable to prevention and cure. Don’t work on Sundays and hailstorms will avoid your field (and if the neighbouring field is hit, there is your greedy workaholic-cum-sinner co-villager).

    The third type is the pure mystery – “god’s paths are unknown” and evil spirits strike randomly. In this case, there is a mechanism but one does not have access to it and no-one seems to know. But this final type suggests a relationship with other representations. The mystery is explained by recourse to other less-mysterious entities. God’s reasons might be inscrutable for my particular problem, but I know it has a larger plan and I know some of his routines (the Sabbath for example). A mean-spirited spirit (pardon the pun) might set fire to my house even if we are not bound by any contract of reciprocity. They did it for the lulz as the internet meme has it.

    This classification allows for fuzziness and overlapping. One could try and find a type 2 explanation – sin retaliation – but evidence might make type 1 or type 3 more straightforward.

    A short summary of above ideas: explanations of misfortune are folk theories aiming to find the causal origin of improbable events with significant effects for humans. From an evolutionary point of view, this seems like an over-firing of an “agent-detection” module/propensity. But I think it is adaptative since there is a high cost for a false positive but a rather low cost for a false positive because sometimes, what looks like misfortune is in fact malice and one should do something about it. Malinowski was refused companionship on some Kula travels because canoe leaders thought he brought bad luck. If something went wrong on a trip with this strange fellow but not on so many others when he was not around, it is plausible that he had something to do with the misfortune and it is reasonable to skip him on the next sail. And I would venture to say that there might be cases in which the agentive theory is right for the wrong reasons. There could be a malicious agent behind apparent misfortune and some steps taken to redress it could avoid further damage. So far, I am not aware of evidence to this effect.

    There is a second hypothesis: agentive explanations of misfortune do not have costly effects; they are fitness-neutral or moderately positive. There isn’t much effort in respecting the Sabbath or placing lottery tickets near religious items, just as there is no practical result. But when alternative explanations have fitness-enhancing results e.g. a medicine cures a disease formerly attributed to agentive misfortune, the agentive theory is either dismissed or revised to account for this new development (for example, the medicine attacks the evil spirit which is the cause of misfortune). In the absence of alternative explanation, agentive theory of misfortune

    Finally, I am not entirely certain that a single causal mechanism can account for all types of agentive misfortune theory. I have in mind a hypothetical example: if we find another village where people believe that hailstorms are provoked by malicious neighbours rather than god, we have to take into account other social mechanisms such as the appearance of monotheist religion which unifies all supernatural agencies within a single entity and here we have other causal mechanisms involving official theology and state-building.

    This seems more of a development of Pascal Boyer’s interesting proposal than an actual solution, but helas! 42 bucks deserve more work than what I put into these musings. I am very curious about what the proposer and others have to say on the topic, so looking forward to the blooming of ideas!

  • comment-avatar
    Radu Umbres 23 August 2011 (01:02)

    Reading Hugo’s interesting answer, I found some common points of argument but I would like to ask him about the asymmetry between thinking about good and bad fortune and the agents behind them. I would still argue that the two types should be covered by the same cognitive mechanism. Let me give an example from barbut, a dice game played in Romania. Players think there are jinxes but also benefactors, even if they are involuntary and their effect is unintentional. One just brings bad OR good luck to a player and should leave or respectively attend the game. What is interesting is that the reason for being such a causal agent is not really understood and neither is much attention paid to find out. It suffices that the cause is removed, whatever the mechanism behind it. This looks like a heuristic method, a rough-and-ready tool to get rid of negative influences. In this anecdotic example, the solution is simple and low-cost: just ask the person to leave/stay or refrain from playing dice if the auspices are adverse.

    However, I am prepared to accept the possibility of an empirical imbalance between occurrences of the two types of explanation. But it could be due to ecological constraints and opportunities. In an environment where low-probability events are mostly harmful (from an objective point of view), we should expect more explanations of misfortune. Conversely, when low-probability events are also (or predominantly) positive, my hypothesis would suggest a corresponding presence of (good) fortune theories. In barbut, for example, the law of conditional probabilities make clearing out everyone else or losing everything in one night as two forms of the same black swan of very improbable event. One should then better come equipped with heuristics for fortune as well as misfortune. In lotteries, as an extreme example, it is only about boons and nothing regarding misfortune – unless your ticket booth burns down, but that’s another story 🙂

  • comment-avatar
    Martin Stehberger 23 August 2011 (20:20)

    Thank you for a very interesting post on a very interesting site. First, some rather quick stabs at Questions 2 and 3. Question 4 gives me headache anyway; indeed, for my answer to Question 1, my main target, I must initially pretend not to have read Question 4 at all.

    Question 2: Why “Why me?”?

    What seems obvious is that people are usually more concerned about their own fortune/misfortune than about that of others … but admittedly this doesn’t really address the type-versus-token issue. I will address the Zande roof-collapse example below, at the end.

    Question 3: Why the asymmetry between good and bad fortune?

    Losses are more important than wins since in life everything can be lost at any time but there is no corresponding “everything” to win.

    Now to Question 1: Why agents?

    For what it’s worth, I will go in an old-fashioned direction here. When asked “why would our evolved design for a mind include the propensity to focus on and ponder at length totally useless explanations”, I would try to argue parsimoniously, for as long as possible, that these explanations are just a natural misunderstanding, to be blamed on a misleading world as opposed to on a specifically human cognitive bias. For all the following I assume people reasoning in an ancestral environment.

    Instead of “events” or “occurrences” let us consider a clearer case first: “design”. When people explain design in the natural world by way of an agent designer, this is not yet enough to postulate an evolved propensity to see agent designers, because what other explanation for design could people find anyway? (Natural selection over long periods of time? Only a super-genius would come up with that on his/her own.) And whereas agent designers (say, of tools) are well-known, there is no confirmed counterexample, of design without agent — since how would you ever “prove” it wasn’t an agent?

    But if this makes sense can’t we then argue similarly for “events” instead of “design”? Whereas many events are clearly caused by agents, there is no confirmed counterexample, of an event occurring without any agent involvement — since how would you ever prove, say for weather events, that there wasn’t some hidden or invisible agent at work? So wouldn’t it just be natural for people to assume that all events are caused by agents? I would indeed have suspected so.

    Question 4: Why are only some occurrences explained in agentive ways?

    This obviously immediately contradicts my answer to Question 1. But I would still be curious to learn of a specific clear instance of an event that is explained without any agent involvement, and instead using only covering laws, general causal processes, and the like. There can be complications here. I assume “covering law” or “generic causal process” could mean something like gravity, for example, and people may indeed recognise gravity as such a law or process, but if an object falls down the ultimate cause of the event may be (seen to be) not gravity but an agent who dropped or pushed the object.

    I believe that this view can explain the seemingly strange “double explanation” regarding the termite-damaged collapsing roofs. There are two different events (seen to be) occurring here. Firstly, the slow event of termites damaging the roof. For this event we have the termites as agents behind it. And secondly, the sudden event of roof collapse. This is not something the termites have done at that particular moment, so a different agent will be invoked to explain the event.

  • comment-avatar
    Hugo Mercier 23 August 2011 (21:39)

    Thank you Radu for your comment.

    I agree with you that the same mechanism should address both. Also, there are clearly cases in which we suspect a hidden intentionality for good fortune. Yet it seems hard to deny that people more often have an incentive to hide their actions when they hurt someone than when they benefit someone. It follows that we more often have to look for a hidden hand in the former case than in the latter. There may be other reasons, such as the one you point out, for the asymmetry, but I think that this one still holds.

  • comment-avatar
    Gabriel Gutiu 25 August 2011 (14:39)

    Interesting puzzle.

    Question 1: Why agents?

    While humans lived as hunters and gatherers they were prone to animal attacks (and probably attacks from other groups as well). So if they hear a stick break behind them or the wind move some leafs of a tree it’s safer for them to assume that it’s a potentially harmful animal/other human or maybe even an animal that they can hunt for food. If it’s “just the wind” then they just unnecessarily turned their head and their heart rate went up in anticipation of a possible attack. But if they don’t react to it then they just as well could end up dead or without a meal. So to assume that there is an agent that is responsible may very well be an evolutionary heritage from the humans hunter and gatherer times that even today has survival benefits: consider yourself walking a dark alley in some not so safe neighbourhood and you hear something behind you. What is safer: assume that it’s the wind that dropped something or a mugger?

    It doesn’t take much to argue that something like this also influences basically everything where belief, in whatever form, is a factor. So assuming that there is an agent is adaptive.

    Question 2. Why “why me?” ?

    two possiblities: #1 It’s possible to argue that because thinking about it and why it has affected “me” and believing that it’s Karma, god’s punishment for bad deeds etc. is a form of stress relief that has a positive effect on health. If people believe that there is some entity responsible for it and with good behaviour, prayer etc.. they are able to make up for what they think they did that made this happen, then this gives them a sense of control over an unfortunate situation. This acts as a (longtime) stress relief which in turn prevents longterm negative health effects.

    #2: as a means of maintaining social cohesion and minimizing (lack of a better formulation) peple at times being jerks in a group: the person who the misfortune happened to asks himself if he did something to others/broke the moral code etc that would make some entity angry; at the same time other people of his group might argue the same.. this happened to you because you behaved like a jerk then and then…

    Actually probably both arguments in combination are helpful.

    Question 3. Why this asymmetry between good and bad fortune?

    This can be explained in combination with Question 1 and 2: overdetecting negative/dangerous/”bad” agents, even if they’re not there, is less costly than underdetecting. Good fortune/luck already in itsself has positive effects (if you win the lottery, hunt some animal etc), so there is no need to alleviate some potentially negative effects by constructing a theory around it.

    Question 4. Why are only some occurrences explained in agentive ways?

    Don’t have a strong point here, but I’d argue that because the more obvious ones can be explained in realistic terms, while for the ones that require a long chain-of-effect are not possible to explain them in this way so using traditional explanations/beliefs makes it possible to explain issues that would be otherwise impossible to explain. Kinda like in science paradigmatical assumptions are needed to make theories work.

    Questions 1-3 are answered based on Sanderson (2008): Adaptation, evolution, and religion. In: Religion, vol. 38. 141-156.

    Question 4: is based on Shirokogoroff, S. M. 1999 [1935]. Psychomental complex of the Tungus. Berlin: Reinhold Schletzer Verlag.

  • comment-avatar
    Jordan Levine 28 August 2011 (05:26)

    Q1: Why agents?

    A1: As social beings who evolved to live in tight-knit clans, agent-based thinking is the most frequently rehearsed schema for determining ‘fault’. Given it’s so often rehearsed, it simply *takes the least amount of mental energy to just default to this schema* when faced with complex situations of causation. This (the degree of mental effort required) is empirically testable. (Although you’d need to test this in a society that does not intensively educate itself to think using more abstract schemas, i.e. psychology undergraduates are a write-off, here)

    Q2: Why “why me?” ?

    A2: On an individual scale, it’s a waste of time and effort to think about everyone else’s problems. All things held equal, it’s only worth our time figuring out how each of us, individually, has been wronged. Otherwise we’d have no time or energy for our own basic needs. Imagine being a full-time clinical psychologist while trying to simultaneously defend and raise a clan of hunter-gatherers. Not feasible.

    Q3: Why this asymmetry between good and bad fortune?

    A3: As with above, it’s a waste of time and energy to bother thinking about all the things that have gone ‘right’ if your primary concern is to survive in the short-run. That said, it should be noted that many religious traditions train us to counterbalance this asymmetry by being ‘thankful’ (which has its own positive psychological, and thus *long*-term survival effects), so the question’s slightly misleading.

    Q4: Why are only some occurrences explained in agentive ways?

    A4: Again, it’s simply a question of what takes more mental effort given the available evidence (particularly in the absence of an educational system that trains you to use otherwise unfamiliar or counterintuitive schemas). If there are certain misfortunes with viscerally obvious physical causes, no imagination is needed to explain them–the observable material cause is so plain as to preclude the need to waste any more mental energy or time on other, often agent-based theories. (However, it’s worth noting that if you ask people, ultimately, they may attribute the underlying *reasons* for even basic mechanical causation to the will of an unseen agent. This is an instance of scale jumping back up to a level of complexity that prompts the invocation of the exact same schema as in Question 1).

    Note 1:

    All this is empirically testable. You can either ask people directly how ‘easy’ it is to read, or think, using one explanation or another for a given misfortune (but choose the right sample!! Not W.E.I.R.D people), and/or you can set-up experiments that imply or ‘reveal’ this information by measuring, e.g., how much time it takes people to think using one schema (e.g. agent-based) or another (e.g. Newtonian mechanical).

    Note 2:

    The underlying assumption for all these answers is that we are evolved to be efficient with our mental energy. The important caveat, is that we can train ourselves to use and rehearse less familiar, at first more ‘mentally taxing,’ schemas when it’s made obvious to us that they result in more satisfying/useful answers. It’s a matter of education for pragmatism. And even our more ‘materially accurate’ schemas (e.g. Newtonian) are still only schematic simplifications of reality. So some serious humility is required here.

  • comment-avatar
    Tyler Tretsven 30 August 2011 (21:32)

    Thank you for the chance to participate, this is a interesting question. I’ve hesitated about posting this response because I have not yet completely convinced myself, but I’ve come up with an approach that differs a little from the other posters. So far, the arguments have seemed to proceed that misfortune occurs, triggering an evolved mechanism that searches for a source to that misfortune. This mechanism assumes that the culprit is a human-like agent, since it is beneficial when one can find a human source but does not have any cost if a human source cannot be found.
    This seems perfectly plausible, particularly when the misfortune is of the type that a person could perform, such as a flattened tire or a burned bureau de tabac. However, people are just as likely ask “why me?” when a hurricane wipes out their house or a when a person is diagnosed with cancer. The mechanism described above would have to have an exceptionally wide actual domain in order to permit both a flattened tire and a hurricane, the latter of which could in no way have been caused by a person.
    I would like to try to propose instead a different (though related) solution. One commonality across cultures is that the source of misfortune is usually identified somewhere, be they witches, gods, karma, or bad luck. These are all acquired concepts, each of which include that the being or force can cause misfortune indirectly for different reasons. These concepts are highly recurrent across cultures, most likely because they provide a selection advantage to the larger concept of the being or force. Belief in these concepts can also be reinforced when misfortune occurs, since the misfortune can be seen as an observable action.
    At this point, the harmer-detection mechanism described in the first paragraph can be invoked, but in a slightly different way. After the deity/karma/luck concepts are acquired, that mechanism can be used to recognize the actions of those beings when misfortune occurs. This is not an over-attribution of agency, however, since the mechanism is simply recognizing the agency of a being that is already believed to exist. What differs between these beings is that there can be different input conditions that trigger their recognition. Also, I would argue that the confusion that one feels after misfortune occurs arises from ambiguity about who to attribute the misfortune to when there are several potential targets to whom one can place blame (i.e., was it a punishment from God, bad karma, or just bad luck?) and then whether that attribution should be inhibited, rather than simply blindly throwing around an agency attribution and hoping for it to stick to a being with whom one is acquainted (i.e., I feel like someone was responsible; was it God?).
    One implication of this is that if the input conditions are not met, then the misfortune will not be attributed to that being. There is some evidence that this might be true. Imagine, if you will, the difference between the concept of God in contemporary Christianity and the Christianity of, say, the 14th century. While the (typical) contemporary conception of God is similar to that of a loving and benevolent father, the 14Th century God was a vindictive punisher. Take the example of childhood leukemia: nowadays, if a young child develops leukemia, it is not seen as a testament to the bad moral behavior of the parents (except to the most ardent fundamentalists), whereas in the 14th century (if they had known what leukemia was) it is, I think, very likely that the misfortune would be said to be a punishment from God. The difference lies only in which input conditions are set for that being to be recognized as the source of misfortune.
    A second implication is that in the absence of supernatural agent concepts, the types of misfortunes that one could not reasonably attribute to a person, such as hurricanes, would not cause one to attribute that misfortune to an agent. This is more difficult to test since just about every group has beliefs in supernatural agents, yet there are some religious traditions that do not propose supernatural agents such as Taoism and Zen Buddhism. I unfortunately do not have any specific textual examples, but my general impression from reading the Tao and about Zen Buddhism is that in these traditions, misfortune is seen as being simply part of life that one must deal with, rather than being an agentive action. However, these beliefs are probably overshadowed by other local beliefs, such as beliefs in the ancestors who can also cause misfortune, and thus these examples may not be entirely valid.
    This model, then, can answer all of the four questions posed. Misfortunes are attributed to agents because that is the pattern of supernatural beings that are most often created and are transmitted most abundantly. It was stated in the question that karmic beliefs differ from other agent beliefs. This may be true in that karma is not represented as an agentive being in the same way as, say, God; yet, I would argue that the actions of karma are nevertheless agentive, since there is the implication of a sort of intelligence behind it.
    As for the question “why me?” the supernatural agent concepts not only include that they cause misfortune, but they also include that they cause misfortune for specific reasons: karma causes misfortune in response to bad behavior, and gods (ancestors, etc.) cause misfortune either for bad behavior or in response to insufficient sacrifice. When misfortune occurs as a punishment for bad behavior, people are likely to ask “why me?” because they do not think that they deserved to be punished for anything. Several psychological studies have shown that people tend to have inflated opinions of themselves, and also tend to forget about the bad things they do while remembering the good. If misfortune occurs because of insufficient sacrifice, it may cause people to question why their sacrifice was insufficient.
    The asymmetry between good and bad fortune may exist because the supernatural agent concepts focus more on punishments than rewards. On the other hand, since people tend to remember the good things they do and forget the bad, they might feel they deserve the good, and thus not think about it all that much. Also I recall a study that showed something along the lines that punishments feel worse than rewards feel good, which might lead to the differential attention paid to them.
    Lastly, this model can account for why only some misfortunes are explained in agentive ways. This can occur because only some types of punishments or misfortunes are included in their concept as part of that agent’s repertoire. Childhood leukemia may have been in God’s repertoire in the 14th century but not in the 21st century, for example. In the Fang example, perhaps when children learn about what witches are, they are taught that they only are responsible for certain ailments and not others. Why some misfortunes should be attributed to them and not others is then just a question of how the witchcraft concept came into being and developed. Also, some misfortunes may not be serious enough for them to satisfy the input conditions of a certain agent, or they may simply not be improbable enough to fool our folk-statistical minds into thinking that they’d be impossible by chance.
    I hope I did not make too many rookie mistakes in making this argument, but I think that this model decently accounts for all of the questions posed. The only reason I hesitated to post this is that it seems counterintuitive that people only attribute misfortune to agents because they learn to, yet since virtually everyone learns supernatural concepts that they can use to explain misfortune, it is difficult to know if it really is a human universal aside from cultural acquisition. Even atheists learn these concepts, and may have their misfortune-attribution affected by them (e.g., http://johnhawks.net/weblog/site/tornado-hits-hawks-garage-2011.html ). Thanks for the time.

  • comment-avatar
    Helen De Cruz 7 September 2011 (20:52)

    Hi Pascal,

    Interesting puzzle. Here’s my stab at it. We could see this in game theoretical terms.

    1. Agents – nonagents
    Guthrie proposed in Faces in the Clouds that we have a propensity to discern the hand of agents (especially other people) in the world around us, because it is better to have a false positive than a false negative. Agents are very evolutionary relevant for us, and in particular, we can interact with them to affect our fortune/misfortune. If I suspect my neighbor, who is an avid pigeon keeper, from systematically poisoning my cats, and this happens to be true, I can alter the situation by speaking to him in a firm voice, whereas if I (wrongly in this hypothetical case) suspect that there is a cat virus going around, I have won nothing. (There is also a potential cost: supposing that the cat virus scenario is true, I may anger my neighbor by accusing him wrongfully, however, I could try to talk to others first to see if he is in the habit of poisoning cats).

    2. Why me?
    Obvious from an evolutionary point of view. Again, payoffs are higher when I think ‘why me’ and my suspicion is correct than when I think ‘why me’ about an unrelated individual. An interesting test would be whether people would be also more prone to thinking ‘Why does this happen to my child’ (or brother, or other related individual) rather than an unrelated individual.

    3. Asymmetry
    Again, a question of payoffs. If you’re on a roll, if things are going well for you, what’s the point of wondering why? By contrast, if you have systematic misfortune, your questioning why this is so leads to potential higher payoffs. E.g., suppose I live in a draughty, moldy house and wonder why I get asthma attacks, bronchitis and so on, I have a potential payoff if I correctly identify the source and move house. If things are going well, as long as they are going well, there is no added benefit.

    4. Selectivity
    Hmm, let’s see if game theory can help us here. Trying to explain all instances of misfortune is quite time and energy consuming. So we need to quickly take out those instances where, as Hugo put it, we can think that foul play is at work. This is not infallible, and in some cases systematic bad luck just happens. But those occasions are presumably outnumbered by those where we do identify foul play, e.g., the office worker who correctly notices that his colleagues do always pick him out for the practical jokes.

  • comment-avatar
    Nick Connolly 12 September 2011 (05:15)

    Why “why me?”

    To me this seems the simplest of the questions. A misfortune has occured and it is natural for the victim of this misfortune to consider why the misfortune happend to them, if only for the purpose of avoiding similar circumstances.

    I got locked out of my house. Why? Because I left my keys inside. In the future I will avoid leaving my keys inside.

    Misfortune for which our recent past actions are a simple explanation for the misfortune we suffered may be emotionally vexing but are otherwise simple. We did something (or failed to do something) and something bad happend as a consequence (or appeared to). Consequently we will avoid doing that.

    More complex are the cases where misfortune was brought about by the actions of others. Of course we still will re-evaluate our future behaviour but our choices are more complex. In the case of agression our choices might be polar extremes (e.g. be more assertive towards an agressor or to be more passive).

    A third category would then be cases of random misfortune or misfortune due to events (or agents) beyond our control. In the other two categories our urge to think through how we could have avoided the misfortune is resolved by reaching some sort of conclusion about future behaviour. However in this third category there is nothing we can do, so the matter is left unresolved and hence will be a permanent source of stress – i.e. something we will always worry about.

    Consequently we turn to more complex responses. Those responses need not be theological or supernatural. It could be stoicism or resignation to bad circumstance. However by definition they cannot be practical solutions to original misfortune.

    From this point the other questions can be answered:

    1. Why agents? Misfortune caused by agents is the next step up the hierachy from misfortune caused by our own actions (or inactions). As the purpose for considering the cause of misfortune is to make adjustments to future behaviour, when the immediate cause is a non-agent our focus is directly on our own behaviour. The next step on the hierarchy is to consider if we ourselves are not to blame (i.e. we can’t immediately change our behaviour to avoid repeating the misfortune) then who is? Normally that would be an appropriate step.

    3. Why the assymetry? There are evolutionary advantages to trying to work out the cause of a good event so we can tailor our future behaviour to make it re-occur. However, when the causes are obscure a default behaviour of simply carry on doing what one was doing before makes a lot of sense. If something worked before it makes sense to not make any changes. The engineeers maxim: if ain’t broke, don’t fix it.

    4. Why are only some causes agentive? This is harder but I would suggest that some misfortune would be understood in terms of the first step in the hierarchy – i.e. a simple non-agent based cause can be found (or appears to be found). This suggests that the right action (or inaction) can prevent the future misfortune. It is not immediately obvious, though, how we stop the termites eating out wooden house etc – but at least the nature of the problem has changed.

  • comment-avatar
    Jean-Baptiste André 20 September 2011 (17:37)

    Thank you Pascal for this great question and great game as well!

    I would like to propose a solution much along the ones proposed already in many posts (starting with Hugo’s), but adding a twist to it.

    In fact, my feeling is that Pascal did give the answer to the puzzle in his previous post on “magical thinking”! We should always distinguish intuitive and reflective beliefs, as Dan Sperber has long argued. This is true in the matter of explaining misfortune, as in other matters.

    Let me start by recalling the solution others have already very well exposed, because I endorse it as well.

    Our biological interests are always at risk of being endangered by malevolent conspecifics. This calls for a module adapted to detect possible instances of malevolent activities. This module should have at least four properties/consequences.

    First, why agents? As Hugo and others have well explained, this module is interested in agent-based explanations because it is its very job. Plus, paying too much attention to social explanations is less costly than paying insufficient attention to them. Therefore, the inferences made by this very module tend to be relevant in a wide range of contexts.

    Second, this module is interested in tokens rather than in types of misfortunes because its function is not to accumulate encyclopaedic knowledge about causalities at work in the world, but particular knowledge about the dangers of my own (or my family’s or friends’) social world.

    Third, why bad more than good fortune? When someone does something good to me, two things are possible. (i) This was intentional, and then he will let me know about his good action soon enough (as Hugo notes). (ii) This was non-intentional and came as a mere by-product of a self-serving behavior of his. In either case, it is not useful for me to invest cognitive effort in detecting his good behavior (note that the same principle is at work in the “Knobe effect”, and that this is not specific to social life but general to all instances in which a given behavior has more than one effect).

    Fourth, why only some misfortunes will do? Even though social explanations are particularly relevant in general (point 1), they need not be relevant in all instances. They must be at least plausible; competing explanations must not be far more likely; etc.

    Now, all this is part of the explanation, but not quite enough I think. Say, I am walking in the jungle and I hear a sound behind me. Even though the likelihood of it being a tiger, or another dangerous animal or human, is extremely small, it will be worth paying attention to this possibility anyway because of the cost of false negative. Therefore, I will turn over to check. This is the intuitive output of, say, my dangerous-agent-detection module and more generally of a mind geared toward relevance. In the case of misfortune, the equivalent is the intuitive feeling that there might be someone behind my misfortunes.

    However, this intuition is not the same as the explicit, reflective, formulation of agent-based explanations to misfortunes. I may entertain transiently, the unpleasant, vaguely paranoid, intuition that someone malevolent is causing my roof to leak. That does not mean that I will adhere to the ludicrous belief that someone is having pleasure to wet my carpet. I may adhere to explicit beliefs of these sorts only if they are proposed to me, by trustworthy others, as legitimate explanations. But in this case, I will adhere to them only in a reflective, not intuitive manner.

    Therefore, we are exactly facing the same paradox than in the case of “magical beliefs” of the kind Pascal refers to in his preceding post. Ceteris paribus, I may prefer drinking in a glass with “H2O” written on it rather than “vomit”, because I have a slight negative evaluation of the “vomit” glass. Yet, I will never reflectively endorse the belief that a word written on a glass can change water into vomit.

    Explicit agent-based explanations to misfortunes are hence likely to be highly variable from society to society, and from social group to social group, because explanations that are legitimate in one group are not in another. To take Pascal’s example, among the Zande, when the roof of a granary collapses, everyone considers this must be a case of witchcraft. This is a legitimate reflective beliefs among the Zande. But this is not a legitimate explanation for most of us.

    What is more, Pascal not only gave, or so I believe, part of the answer in his post about “magical thinking”, he also gave another part of the answer in his post on the role that coalitions play in the explicit adherence to reflective beliefs. The explicit explanations one gives are (often) indications of one’s coalitional affiliation. And this is true also in explanations to misfortunes. For instance, scientifically shaky explanations are generally considered illegitimate and ridiculed among scientists, and among those who have a positive opinion of science, but less so among others. “Complotist” explanations are considered legitimate among those who feel coalitionnally far from the politico-economic elite. Etc. Plus, coalitional affiliations also play a particularly important role in social explanations to misfortune when these explanations involve outgroups. Endorsing the belief that the misfortune of a member of group A is caused by “evil” member(s) of group B is an efficient way to display one’s affiliation to group A!

    So, overall, if my suggestion is right, then the prize should go to Pascal (and to Dan)!

    Note, interestingly, that I have not written a word on ICCI for almost 2 years. This is really all about the money then.

  • comment-avatar
    Nick Connolly 21 September 2011 (01:51)

    Further to the various comments above. The common theme would seem to be an underlying theory of the emotion of regret. Regret is not simply sadness with regard to past misfortune but also an emotion that motivates introspection and the consideration of how past events may have been different.

    Regret makes sense from an evolutionary perspective for the reasons discussed above: it motivates us to reconsider our actions so that we avoid the unpleasant event that caused us to feel regretful. Consequently we should find, experimentally, to aspects to resolving regret: 1. finding an alternative course of action and 2. motivated by regret, implementing that action in the future.

    In the case of addicition, poor diet choices, binge drinking etc, we get past the first step but perhaps not the second.

    In the cases above we have issues with getting past first step: finding an alternative course of action.

  • comment-avatar
    Pascal Boyer 21 September 2011 (18:35)

    [email contribution sent in by Névtelen Kerem, via PBoyer]

    If I may contribute, it seems to me that most responses focus on the wrong question, from an evolutionary viewpoint. Because P Boyer asked “why” questions, we tend to search for the cognitive antecedents, the mechanism that produces misfortune-explanations.

    But from an evolutionary viewpoint, we should ask not just, not Why? but also, What for? What do these explanations do for people who produce them? What potential benefits are there, with fitness consequences, in noticing cases of misfortune and in telling oneself and others that said misfortune was sent by agents?

    Here we may take a leaf out of P Boyer’s own book, or rather out of his recent post on epistemic recklessness. Why not treat misfortune in the same way? Perhaps discourse about misfortune and its causes is one of the many cases of communication for the purpose of creating coalitional alignment, recruiting support or checking who is for whom when the chips are down.

    Consider cases of misfortune like the Zande granary, or a case of illness. Such mishaps or tragedies have been part of human lives for all of our evolutionary history. One thing that is special to humans is the extent to which they depend on support from others to snatch fitness victories from the jaws of misfortune, by recruiting other agent’s time, energy and information. No need to elaborate this any further.

    So let’s get to the questions:

    Why agents? Or rather, what are agents for? When you claim that misfortune was caused by agents, rather than focus on covering laws, you are turning a causal process into an attack from particular individuals. That is, you are making your interlocutors activate whatever mental systems support coalitional psychology. They are now led to consider, not just your misfortune, but also some agent with both power and intention to inflict fitness costs. Why is that important? Because of the second point:

    Why Why me? Or rather, what is in it for people to talk about victimization rather than misfortune? Here I must say I found P Boyer’s formulation rather misleading, insisting on the “why me?” question. In fact, most people in the world seem to have an interest in creating explanations not just for their own misfortune but also for others’. Case in point, in the Zande granary case, everyone in the village thinks witchcraft was involved, not just the victims. Witchcraft the world over is something that a whole social group, not just the patient herself, considers a good explanation for e.g. an illness. So the question is, why do people want to claim that the malevolent agents had a particular target in mind? Again, this may be because that is the most relevant factor in activating people’s coalitional psychology. If I say, “I got measles sent by the gods but they did not know it would hit me”, I am not likely to recruit anyone. But if I claim that the gods wanted to hurt me, I now communicate that the malevolent agent is after particular people, including people of my group: to the extent that you are in my social group, or to the extent that the witches can attack people in our village, you are very much a potential target too. This should motivate you to do something about the situation. Also it could motivate you to demonstrate some solidarity. This also works for third-party accusations. Naming and blaming a particular person as a witch means that you create instant polarization in the surrounding group. All members have to decide whether they are in with the presumed witch, or on the contrary are prepared to join forces against the target, sometimes a powerful or successful person.

    Why mostly bad stuff? Because you need coalitional help to palliate fitness threats, but you do not need it to enjoy fitness windfalls.

    Why only some events, and why cultural variation? I would venture a rather bold hypothesis. People will produce agent-based explanations for those events for which, given the general cultural knowledge in the group (e.g. theories about accidents or disease), and local dynamics (who is for whom), there is a prospect of either getting better social support, or at least there is an interest in checking which way social support is going.

    I realize that this proposal is to some extent in line with Hugo Mercier’s and JB André’s, and inspired by several other contributors.