Field of Science

Atonement, self-punishment, and guilt

Atonement is a funny concept. Essentially, it's the idea that you can cancel out a wrongdoing not by doing a good deed, but by engaging in some act of self-punishment.

Although the classic example comes from Christianity (the tortured death of Jesus) similar concepts of penance are widespread in other religions. Penance goes beyond the more normal concepts of justice (revenge and punishment) because it's voluntary.

Perhaps there's more going on here than meets the eye. Rob Nelissen and Marcel Zeelenberg at Tilburg University in The Netherlands speculate that people might indulge in self-punishment because it makes them feel better.

They set out to test whether people self-punish when they are made to feel guilty, but only if they can't make good the wrong doing directly.

The basic idea was that subjects had to perform a test that they were told was a measure of how hard they concentrated. As usual it was no such thing - whether the subjects succeeded or failed was entirely manipulated by the investigators (why do the subjects fall for this every time, I wonder!).

They were paired up in this game with another player (OK, so the player was fictitious too, just there to help manipulate their guilt).

Basically the deal was that some of the subjects were made to feel that they had underperformed on the second round of the game, so that they had let the other player down.

In the third round, they were given the opportunity to self punish. Instead of just receiving points for correct answers (as in the previous rounds), now they would get points taken away for wrong ones.

The key to the experiment was that some participants chose the level of their own punishment, while others got to chose the level of their partner's punishment.

The graph sums up the results nicely. In the control condition, there was no guilt and the level of self-punishment and partner punishment were similarly low.

In the guilty condition, there was no change in the partner punishment. But there was a large increase in self-punishment.

It seems likely that this self punishment only takes place when there is no opportunity to right the wrong. In another experiment, they asked people to envisage a variety of scenarios about borrowing money for college from their parents, and then goofing off. They were then given some options on what to do next.

Some students were presented with the scenario where they had no opportunity to make up for their wrongdoing by working harder. These students were more likely to choose the self-punishment course (denying themselves the pleasure of a skiing trip).

So, there you go. Do the religious ideas of penance and atonement result from a subliminal need to self-punish? And, if they do, what could possibly be the function of it (from a biological/evolutionary perspective)?

Nelissen RM, & Zeelenberg M (2009). When guilt evokes self-punishment: evidence for the existence of a Dobby Effect. Emotion (Washington, D.C.), 9 (1), 118-22 PMID: 19186924

Creative Commons License This article by Tom Rees was first published on Epiphenom. It is licensed under Creative Commons.


  1. Justice, equality, balance, are these genetic? If one person eats more than his share and then sees other suffer, and responds by fasting, that would help the group.

    If the self-punishment is seen, and/ or communicated, that would be a deterrent to others committing the same "sins" and it could save help the "sinner" from possible exile.

    A recent example is assumed environmental damage to poor nations as a result of predicted global warming. The US recently pledged 100 billion dollars a year to help poor nations cope. Would that (I'm not asking if it's enough) help our species? I suppose that would help, even if the skeptics are wrong- we'd have less cash for cruise bombs and mercenaries.

    Here in Detroit we're starting to build electric cars. Electric cars! Now that's atonement!- David Mc

  2. Does this also explain the "blushing" response? They should have observed some of that.

    Also, how many responded by instinctively covering their faces, as if expecting punishment? Did anyone stub their toe? Make painful noises? This is hardwired in non-sociopaths. It's fun to see it quantified. David Mc

  3. Tom, the explanation that immediately comes to my mind is costly signalling. The people are self-punishing to show others they regret their actions. Was there any work done on whether people who engage in more religious practices also engage in more self-punishment?

  4. Total random speculative stab in the dark from a total layman with an interest in evolutionary topics:

    It's well-known that primates have an evolved tendency to punish perceived cheaters, even when there is no net gain to the punisher or to the group. IIRC, this has been well-demonstrated in chimps as well as in humans. (I don't have the citations right in hand, but I'm sure Tom or others with relevant expertise are far more familiar with the work than a lowly computer engineer like me.)

    It seems to me that it could just be a simple misfiring of the no-freeriders imperatives, i.e. perhaps in many individuals that imperative is not sufficiently inhibited to prevent the instinct from never acting on the self.

    Monkey A observes monkey B getting unfair benefits, so monkey A wants to punish monkey B. Perhaps if monkey A starts guilt-tripping monkey B, it might sometimes cause monkey B to empathetically experience the same imperative to punish monkey B...?

    I suppose one way to test this would be to set up two groups, one in which the participant is being induced to want to punish a perceived freerider, and another in which the participant is being induced towards "atonement", and monitor the brain activity in both groups. If similar parts of the brain are firing, that would support my wild speculative hypothesis. :)

  5. @Konrad,

    I pretty well agree. I suspect that we'd see self-punishment decrease as people became more certain that others can't observe their outcomes.

    Game theory and Atonement

  6. It might not be just a case of insufficient effort. They may have perceive that they are less skilled, and thus less worthy of an even share in reward- defective in some way at the skill at hand. We all have different gifts and modes of operating. Our success often depends on the task at hand. That's why we have specialists. I can't draw a straight line or produce good handwriting no matter how hard I try, but I excel at other tasks. I want to be rewarded by the group for those, but not for things I do crappy. David Mc

  7. Costly signalling is an attractive explanation. Unlike other alleged costly signals related to religious acts, this one is genuinely hard to fake (i.e it can only indicate that you understand the offence caused and that you want to make amends).

    But the problem is that this experiment was anonymous - so who is the self-punisher sending a signal to? It could be an artefact - it might be that truly anonymous encounters were so rare in our evolutionary past that we aren't prepared to deal with them.

    Religious atonement is never anonymous, of course. So it could be that it's a misfired costly signal sent to a supernatural peer group.

    They didn't separate out religious and non-religious, but given that similar experiments have found no differences, I wouldn't expect one here. What might make a difference is hitting them with a religious prime (or secular one) prior to the experiment.

    The authors suggest that self-punishment may simply be irrational and harmful - the 'dark side' of guilt. They caution that encouraging guilt may cause harm if that guilt can't be undone without self harm. But then they do go on to offer a signalling explanation:

    "On the other hand, there may be functional relevance to guilt-induced self-punishment. Self-punishment may signal appreciation of and future compliance with violated standards (for a similar position on the function of guilt-induced action see Caplovitz-Barrett, 1995). The Dobby effect then, is a public sign of reconciliation that occurs if actual reconciliation (by compensating the victim) is impossible. Alternatively, self-punishment may serve a self-affirmation function, by which individuals restore their self image after violating personal standards. Indeed, people seem to have a desire to wash away their sins after threats to their moral identity (Zhong & Liljenquist, 2006)."

  8. This comment has been removed by the author.

  9. I would agree with Konrad that costly signalling might point to the answer - which is not anonymous if you believe yourself to be observed by supernatural agents.

    But I do feel that the "self-affirmation function" Tom cited from Zhong & Liljenquist also has its point: As with private prayer (which e.g. Jesus endorsed), self-punishment may support specific self-perceptions.

    Personally, I found the Credibility enhancing display (CRED)-model by Joseph Henrich an impovement to the older term "costly signalling". Accordingly, self-punishment could be (comparable e.g. to prayer) a CRED directed at either social others, the supernatural or (and) the self.

  10. In addition to costly signalling, couldn't self-punishment for an act be an advantageous for future individual fitness over punishment by others? Leaving punishment up to unsympathetic others in the group may result in much more damaging punishment. You also know your own weak spots (weak back, weak knee, etc.) and could avoid them. It could give back a degree of control over the punishment.

    This explanation could only work if self-punishment for an act prevents others from punishing. Perhaps a prediction would be that the vigour of self-punishment is proportional to the perceived ability to damage of the others' punishment(s).

    The excessive levels of self-punishment in God intoxicated people could be a result of the perceived divine level of punishment available to God and the incessant failing to live up to the near impossible standards of behaviour required by this God.

  11. A number of good points and suggestions have been made in the comments I am getting somewhat concerned that we're all (and I include myself) getting ahead of the data. And I'm a philosopher!

  12. There should have been a 'however' somewhere in the previous comment. Insert as needed.

  13. True, but if you can't indulge in out-of-field speculation in a blog comments, then where? :) Lot's of fertile hypotheses, now we just need a psychologist to test them!

  14. It makes me think of a possible t-shirt design: Need More Data!
    More seriously, once I get through my current load of writing I plan to find a friendly psychologist to start working on precisely these kinds of things. I already have an open date with people at the psych department in Warsaw to see if anyone is keen. If you're a naturalist philosopher you ultimately have to put up or shut up.

  15. Hmmm. Perhaps, making public atonement might be done in the hope that it would preclude the group from imposing their own punishment. Maybe if you gotta pay a price for your misdeed, you are best off picking your own punishment?

    Just speculating here....


Markup Key:
- <b>bold</b> = bold
- <i>italic</i> = italic
- <a href="">FoS</a> = FoS