Here’s a story which might sound familiar to all of you: one day, a young child is wandering around in the kitchen while his parents are cooking. This child, having never encountered a hot stove before, reaches up and brushes his hand against the hot metal. Naturally, the child experiences a physical pain and withdraws his hand. In order to recalibrate his behavior so as to not avoid future harms, then, the child spends the next week unable to get out of bed – owing to a persistent low-energy – and repeatedly thinks about touching the hot stove and how sad it made him feel. For the next year, the child returns to the spot where he burned his hand, leaving flowers on the spot, and cries for a time in remembrance. OK; so maybe that story doesn’t sound familiar at all. In fact, the story seems absurd on the face of it: why would the child go through all that grief in order to recalibrate their stove-touching behavior when they could, it seems, simply avoid touching the hot stove again? What good would all that additional costly grief and depression do? Excellent question.
In the case of the hot stove, we could conclude that grief would likely not add a whole lot to the child’s ability to recalibrate their behavior away from stove-touching. It doesn’t seem like a very efficient way of doing so, and the fit between the design features of grief and recalibration seem more than a bit mismatched. I bring these questions up in response to a suggestion I recently came across by Tooby & Cosmides, with whom I generally find myself in agreement with (it’s not a new suggestion; I just happened to come across it now). The pair, in discussing emotions, have this to say about grief:
Paradoxically, grief provoked by death may be a byproduct of mechanisms designed to take imagined situations as input: it may be intense so that, if triggered by imagination in advance, it is properly deterrent. Alternatively-or additionally-grief may be intense in order to recalibrate weightings in the decision rules that governed choices prior to the death. If your child died because you made an incorrect choice (and given the absence of a controlled study with alternative realities, a bad outcome always raises the probability that you made an incorrect choice), then experiencing grief will recalibrate you for subsequent choices. Death may involve guilt, grief, and depression because of the problem of recalibration of weights on courses of action. One may be haunted by guilt, meaning that courses of action retrospectively judged to be erroneous may be replayed in imagination over and over again, until the reweighting is accomplished.
So Tooby and Cosmides posit two possible functions for grief here: (1) there isn’t a function per se; it’s just a byproduct of a mechanism designed to use imagined stimuli to guide future behavior, and (2) grief might help recalibrate behavior so as to avoid outcomes that previously have carried negative fitness consequences. I want to focus on the second possibility because, as I initially hinted at, I’m having a difficult time seeing the logic in it.
One issue I seem to be having concerns the suggestion that people might cognitively replay traumatic or grief-inducing events over and over in order to better learn from them. Much like the explanation often on offer for depression, then, grief might function to help people make better decisions in the future. That seems to be the suggestion Tooby & Cosmides are getting at, anyway. As I’ve written before, I don’t think this explanation in plausible on the face of it. At least in terms of depression, there’s very little evidence that depression actually helps people make better decisions. Even if it did, however, it would raise the question as to why people ever don’t make use of this strategy. Presumably, if people could learn better by replaying events over and over, one might wonder why we ever don’t do that; why would we ever perform worse, when we could be performing better? In order to avoid making what I nicknamed the Dire Straits fallacy ( from their lyric “money for nothing and the chicks for free“), the answer to that question would inevitably involve referencing some costs to replaying events over and over again. If there were no such costs to replay, and replay led to better outcomes, replay should be universal, which it isn’t; at least not to nearly the same degree. Accordingly, any explanation for understanding why people use grief as a mechanism for improved learning outcomes would need to make some reference as to why grief-struck individuals are more able to suffer those costs for the benefits continuous replay provides. Perhaps such an explanation exists, but it’s not present here.
One might also wonder what replaying some tragic event over and over would help one learn from it. That is, does the replaying the event actually help one extract additional useful information from the memory? As we can see from the initial example, rumination is often not required to quickly and efficiently learn connections between behaviors and outcomes. To use the Tooby & Cosmides example, if your child died because you made an incorrect choice, why would ruminating for weeks or longer help you avoid making that choice again? The answer to that question should also explain why rumination would not be required for effective learning in the case of touching the hot stove.
It should only be a few more weeks of this until she figures out that babies need food.
One might also suggest that once the useful behavioral-recalibration-related information has been extracted from the situation, replaying the grief-inducing event would seem to be wasted time, so the grief should stop. Tooby & Cosmides make this suggestion, writing:
After the 6-18 month period, the unbidden images suddenly stop, in a way that is sometimes described as “like a fever breaking”: this would be the point at which the calibration is either done or there is no more to be learned from the experience
The issue I see with that idea, however, is that unless one is positing it can take weeks, months, or even years to extract the useful information from the event, then it seems unlikely that much of that replay involves helping people learn and extract information. Importantly, to the extent that decisions like these (i.e. “what were you doing that led to your child’s death that you shouldn’t do again”) were historically recurrent and posed adaptive problems, we should expect evolved cognitive decision making modules to learn from them fast and efficiently. A mechanism that takes weeks, months, or even years to learn from an event by playing it over and over again should be at a massive disadvantage, relative to a mechanism that can make those same learning gains in seconds or minutes. A child that needed months to learn to not touch a hot stove might be at a risk of touching the stove again; if the child immediately learned to not do so, there’s little need to go over grieving about it for months following the initial encounter. Slow learning is, on the whole, a bad thing which carries fitness costs; not a benefit. Unless there’s something special about grief-related learning that requires it takes so long – some particularly computationally-demanding problem – then the length of grief seems like a peculiar design feature for recalibrating one’s own behavior.
This, of course, all presumes that the grief-recalibration learning mechanisms know how to recalibrate behavior in the first place. If your child died because of a decision you made, there are likely very many decisions that you made which might or might not have contributed to that outcome. Accordingly, there are very many ways in which you might potentially recalibrate your behavior to avoid such a future outcome again, very few of which will actually be of any use. So your grief mechanism should need to know which decisions to focus on at a minimum. Further still, the mechanism would need to know if recalibration was even possible in the first place. In the case of a spouse dying from something related to old age or a child dying from an illness or accident, all the grieving in the world wouldn’t necessarily be able to effect any useful change the next time around. So we might predict that people should only tend to grieve selectively: when doing so might help avoid such outcomes in the future. This means people shouldn’t tend to grieve when they’re older (since they have less time to potentially change anything) or about negative outcomes beyond their control (since no recalibration would help). As far as I know (which, admittedly, isn’t terribly far in this domain) this isn’t that case. Perhaps an astute reader could direct me to research where predictions like these have been tested.
Finally, humans are far from the only species which might need to recalibrate their behavior. Now it’s difficult to say precisely as to what other species feel, since you can’t just ask them, but do other species feel grief the same way humans do? The grief-as-recalibration model might predict that they should. Now, again, the depth of my knowledge on grief is minimal, so I’m forced to ask these questions out of genuine curiosity: do other species evidence grief-related behaviors? If so, in what contexts are these behaviors common, and why might those contexts be expected to require more behavioral recalibration than non-grief-inducing situations? If animals do not show any evidence of grief-related behaviors, why not? These are all matters which would need to be sorted out. To avoid the risk of being critical without offering any alternative insight, I would propose an alternative function for grief similar to what Ed Hagen proposed for depression: grief functions to credibly signal one’s social need.
“Aww. Looks like someone needs a hug”
Events that induce grief – like the loss of close social others or other major fitness costs – might tend to leave the griever in a weakened social position. The loss of mates, allies, or access to resources poses major problems to species like us. In order to entice investment from others to help remedy these problems, however, you need to convince those others that you actually do have a legitimate need. If your need is not legitimate, then investment in your might be less liable to payoff. The costly extended periods of grief, then, might help signal to others that one’s need is legitimate, and make one appear to be a better target of subsequent investment. The adaptive value of grief in this account lies not in what it makes the griever do per se; what the griever is doing is likely maladaptive in and of itself. However, that personally-maladaptive behavior can have an effect on others, leading them to provide benefits to the grieving individuals in an adaptive fashion. In other words, grief doesn’t serve to recalibrate the griever’s behavior so much as it serves to recalibrate the behavior of social others who might invest in you.