Many years ago, I was watching one treacly movie or another about a grieving widow. It wasn’t good, but it was embarrassingly effective. I was moved. As I am wont to do, I quickly dissipated my sadness by contemplating a (superficially simple) question: Why do we, as humans, grieve at all. I had, at this point, thoroughly imbibed the evolutionary psychology literature and I approached human behavior from a relentlessly Darwinian perspective. Grieving, I thought, is costly and it is not so obvious what return a griever could expect from his or her anguish. Putting the question this way may seem unduly callous and excessively analytical. Grief is a powerful human experience. Why kill it so we can perform an autopsy on the lifeless cadaver? Putting aside a suite of powerful rationalizations, I will simply confess that the subject piqued my curiosity. I wanted to understand why we might possess a capacity to grieve.
Many years later, my colleagues (Tania Reynolds, Roy Baumeister, Ben Winegard, and Jon Maner) and I believe we may have discovered one possible solution to the puzzle of grief.
We begin by noting that grief is indeed puzzling:
“A bereaved wife every weekend walks one mile to place flowers on her deceased husband’s cemetery stone. Neither rain nor snow prevents her from making this trip, one she has been making for 2 years. However poignant the scene, and however high our temptation to exclude it from the cold logic of scientific scrutiny, it presents researchers with a perplexing puzzle that demands reflection. The deceased husband, despite all of his widow’s solicitude, cannot return to repay his wife’s devotion. Why waste time, energy, effort, resources—why, in other words, grieve for a social bond that can no longer compensate such dedication?”
Although this is a poignant tableau, the real costs of grief might be much more severe:
“It may require significant interruption of a person’s ability to perform daily tasks, to participate in social activities, and to seek out romantic partners (Archer, 1999; Averill, 1968; Schwab, 1992). Grief also increases mortality rate, susceptibility to illness, visits to physicians or other health professionals, suicidal ideation, and possibly even suicide (Fredrick, 1971; Hart, Hole, Lawlor, Smith, & Lever, 2007; Maddison & Viola, 1968; Parkes, 1964; Parkes & Brown, 1972; Phillips et al., 2006; see also Hendrickson, 2009 for a cautious review of the literature on parental grief and health).”
From an evolutionary perspective, these costs are puzzling because they would appear to reduce one’s inclusive fitness–especially because the ostensible target of the displays, namely the deceased, cannot possible recompense the mourner’s dedication (the sadness, longing, et cetera).
But what if the real target of the displays is someone else? Perhaps, that is, grief is a signal that is directed not at the dead, but at the living. Grief, then, might be a signal of (1) one’s underlying propensity to form strong, stable social bonds and/or (2) one’s current level of commitment to a group/tribe. We believe this is one possible reason for the existence of humans’ prolonged grief response. Our scenario for the evolution of grief is as follows:
- The original grief response was a byproduct of the attachment system. At first, the grief response was probably relatively muted compared to today’s average grief response.
- Observers began to note (not consciously) a correlation between a person’s grief response and his or her general propensity to form strong social bonds (or between the grief response and the person’s commitment to a group.)
- Observers began to make social decisions based on the grief response
- The grief response was shaped and elaborated by those decisions just as trees and flowers were shaped and elaborated by the preferences of pollinators.
Grief, then, separates/d the social wheat from the social chaff. Of course, this is a speculative reconstruction of grief, and it is not without problems. Robert Kurzban wrote an excellent blog (see bottom for link) in which he forwarded two basic objections to this proposition. The first objection was that there doesn’t seem to be an asymmetry in costs between a prosocial griever and a more exploitative non-griever. The second was that our proposal relies upon signaling “types,” or relatively stable personality characteristics that persist across domains (e.g. a general “prosocial” personality). These criticisms are insightful and important, and I am thankful that he raised them. Although I concede that they might be entirely correct, I would like to forward a few possible answers. Let me tackle them in order.
We argued that grief is a costly signal that has different costs for a loyal person who has a propensity for forming strong social bonds than for a relatively disloyal person who does not. That is, we argued that a person who forms strong commitments faces relatively fewer costs for grieving than does a person who is unlikely to form strong commitments. Kurzban argues that this is implausible:
“It seems clear that the woman in question has more than just the two options of either grieving on the one hand or exploiting others on the other. People have many things they might be doing at any given moment besides those two activities. In short, it seems from the opening vignette that the authors not only concede but require that it be true that grieving carries very big opportunity costs, even if one is a prosocial sort of person. Yet their argument also requires that the opportunity costs of grieving people to be small, at least relative to non-grieving people.”
This is an excellent point. The argument is that the grief signal must be costly because otherwise dissemblers would deceive others by displaying it. Upon the other hand, there must be a differential cost between those who grieve intensively and those who do not because no one would grieve if it didn’t benefit him or her in Darwinian terms (i.e., inclusive fitness). So what gives?
I think one possibility, one that is addressed in the article, is that the rewards for grieving are larger for a person who is prone to forming strong commitments than for a person who is not. And this is an important part of the of the cost/benefit analysis of signaling. A relatively commitment-free person, a person who cultivates a number of superficial associations, is simply not rewarded as much as a person who forms intense commitments by the addition of strong social bond. If this seems a bit muddy, consider an analogy. Researchers have proposed that men engage in different mating strategies. For simplicity, we can dichotomize these: short term and long term. Now, let us consider a short-term man (Steve) and a long-term man (Bill). Suppose that Steve and Bill both zero in on an attractive woman. They would both like to sleep with her. However, let us suppose that she is particularly fastidious and requires significant investment before sleeping with a man. Both men could, of course, attempt to woo her with expensive dinners, elegant poems, and exquisite displays of investment. But the cost/benefit structure of such displays is different for the two men.
Steve desires a few sexual encounters, not a long-term relationship; Bill desires a long-term relationship. Therefore, if Steve accomplishes his goal, he will obtain access to a few sexual episodes. This is certainly a positive evolutionary outcome, but it is not so important to Steve as is a long-term relationship to Bill. Presumably, Steve could procure sexual access somewhere else and without extreme costs. This means that, ceteris paribus, Bill is more likely to emit displays of investment. And this is, in fact, why women use such a vetting system: it deters those who do not actually intend on forming a stable pair bond.
It seems plausible that grief could operate in a similar manner. So, take two men: John and Dilbert. Now suppose that John is a social butterfly who flits from person to person and group to group, making the most of a high volume of low commitment friendships and group bonds. Dilbert, upon the other hand, practices a high commitment social strategy. Once he bonds with a group, he remains loyal until the bitter end. Same with individuals. John would probably avoid groups that require large initial investments or costs because his strategy is best advanced by seeking low cost social opportunities, whereas Dilbert might be attracted to them. Grief might function as a costly initiation practice, deterring people like John from extracting the benefits of a close-knit, highly committed group.
In fact, this can be quantified, and the math shows that it could work. Please excuse the ridiculously crude nature of this equation. The basic point is simply that the math works. So, suppose that grief costs 1 unit of inclusive fitness (RS). And suppose that the rewards for grief are different for a person with a propensity for forming strong social bonds (Dilbert) than for a person who lacks such a propensity (John) such that the reward to Dilbert = 1.1 RS and the reward for John = .7. Both receive rewards for their displays, but the rewards are differentially rewarding. In this (very simple, crude model), Dilbert achieves a .1 RS gain for the signal and John loses .3 RS for the signal.
Of course, as I said, this is remarkably crude; the real world is much more complicated. The costs/benefits would most likely be distributed across a continuum (i.e., from low grief to high grief, low benefit to high benefit). This also doesn’t consider the signaling component from a group perspective. Nevertheless, the important point is that the math could work. This, of course, doesn’t mean that it does. That is an empirical question, and I am hopeful that future research is conducted that either refutes or supports the theory.
Kurzban’s second objection is that the theory relies on signaling a “type.” That is, according to the theory, a griever is signaling that he or she is prosocial. Kurzban points out, quite correctly, that underlying propensities for prosociality do not necessarily cross relationship domains. In other words, a person might be a very cooperative and committed husband, but an irascible and uncooperative friend. Although I think the general point is correct, I do not believe that it is fatal to the signaling theory. It is a fair assumption that people who commit and cooperate with one person are more likely to commit and cooperate with another person. I suspect that this is relatively domain general, although there are some egregious counter examples (e.g., Bonnie and Clyde—although, contrary to popular understanding, Bonnie and Clyde weren’t loners, and they may have been quite loyal to their fellow gang members). We do, for example, use small snippets of behaviors to predict future behaviors, and we do so because such small snippets are, at the very least, somewhat informative. Likewise, we use a person’s behavior in one relationship as a gauge of his or her behavior in another relationship.
Again, though, this is an empirical question. And it does raise intriguing questions. For example, would women be more affected by a husband’s grief for a deceased wife than men? Would men be more affected by a man’s grief for a fallen group member than women? Who would respond more to grief displays about children, brothers, sisters, or other kin? At any rate, the problems forwarded by Kurzban are important and thought provoking—and I appreciate the discussion.
Last, before concluding, I want to note that there are several reasonable proposals about the evolutionary function of grief. I suspect that the human grief response was shaped by multiple selective forces, so there isn’t any one single function of grief. Too often, we as evolutionary thinkers attempt to ascertain the reason or the function for a particular trait as if reality were a multiple-choice test with only one correct answer. Randolph Nesse, Tooby and Cosmides, John Archer, and others have forwarded compelling theories about the function (s) of grief. The signaling theory of grief does not argue that any of these theories are entirely incorrect. It is rather unlikely that we will discover the special selective force that led to our species’ incredible intelligence because there isn’t one; similarly, it is unlikely that we will discover the single selective force that shaped the human grief response because it doesn’t exist. Instead of searching for some absolute answer, some absolute perspective on such puzzles, we should look through multiple windows. Initially, this might make our view more fragmented, but ultimately it will lead to a more comprehensive understanding.
If the signaling theory is part of the story about grief, a griever’s laments may not be ineffectual communications to the dead; instead, they may be powerful signals to the living.
Kurzban, R. (2014). Good Grief. http://www.epjournal.net/blog/2014/02/good-grief/
Winegard, B. M., Reynolds, T., Baumesiter, R. F., Winegard, B., & Maner, J. K. (in press). Grief Functions as an Honest Indicator of Commitment. Personality and Social Psychology Review.