I’m writing this on the train as I travel back from two fascinating days in Zurich at a workshop on structural injustice, adaptive preferences, and moral responsibility, and the deep and deeply original philosophical conclusion I’ve reached is that adaptive preferences really are puzzling. So now I’m writing about them to try to clear my head. In particular, I’m writing about David Enoch’s attempt to say when the actions we take on the basis of adaptive preferences are less than fully autonomous, where he takes it to be a consequence of their resulting ‘autonomy deficit’, in Natalie Stoljar’s phrase, that those actions are not protected in the usual way against interference by others (Natalie’s talk at the workshop was engaging with David’s paper).
Even saying what adaptive preferences are is tricky. Firstly, they’re preferences: that is, they are conative attitudes like desires, values, wants, etc. Everyone agrees on that. Some theorists focus on comparative attitudes: I prefer this thing to that. Others say they might be absolute: I value this thing greatly. That doesn’t seem to make a great deal of difference. Secondly, they must be genuine attitudes, and not simply what the person reports their attitudes to be. Again, unanimous agreement. And, thirdly, they’re adaptive. That is, a person forms them in response to a situation in which they find themselves. And, specifically, they form them in response to a situation in which their options are limited. And, even more specifically, they come to prefer one of those available options in part because it is available, and indeed come to prefer one of those available options over some alternative unavailable option because the former is available and the latter unavailable.
So, for instance, if I, as a gay man in the UK in 2012, formed the absolute preference to stay unmarried, and the comparative preference to stay unmarried over getting married, and I did so because marriage to the person I loved wasn’t available to me, then that would be an adaptive preference. And if a lesbian woman brought up in a deeply religious community that abhors her sexuality comes to prefer romantic relationships with men over such relationships with women, and does so because the latter is unavailable to her, or available only at great cost, then again that would be an adaptive preference. Another example, this one from Serene Khader: consider a woman living in a very patriarchal society who eats less than she would like to at meal times, and indeed less than she must eat to remain healthy, so that there will be more food for her husband and children, even though there is more than enough food to go around and the result is only that her husband and children eat more than they need and she ends up malnourished. Suppose she values this action, prefers it, will defend it, and will resist attempts by others to change the situation. And suppose she does all this because her society imposes great costs on women who do not, and she has developed her preference for doing it because of this pressure. Then again it is an adaptive preference. A final example, this time from Enoch’s paper: consider someone who likes both music and philosophy, but music more, and would most like to be a great composer more than anything, but who learns that she lacks the musical talent for that at the same time she learns she does have the talent to be a very good philosopher, and reverses her preferences so that now she prefers philosophy to music. Again, an adaptive preference.
Note that the lesbian women’s preference would not count as adaptive if she were simply to prefer the combination of a romantic relationship with a man and remaining in her religious community to the combination of a romantic relationship with a woman and being ejected from that community. If this were the case, she would simply have weighted the values of two components of her possible situations and come to prefer one bundle (relationship with man + membership of community) over the other (relationship with woman + lack of community). To be adaptive, she genuinely has to come to prefer the relationship with a man over the relationship with a woman regardless of what access to the religious community it gives.
Hopefully that gives a sense of what adaptive preferences are. Now let me say why they’re puzzling. Preferences play many roles in our ethical and political thinking. Of course, they also play a role in theory of action, decision theory, and so on, but I think the biggest puzzles arise from their role in ethics and political philosophy. First: for many approaches to ethics, we need to be able to determine the well-being of an individual in different situations, and on some views that’s determined by the extent to which their preferences are satisfied in that situation; but if some of their preferences are adaptive, should their satisfaction contribute to well-being? Second: the central question of social choice theory is how to combine the preferences of the members of a group to give the preferences of the group, and then to use those preferences to make decisions on behalf of the group; but if some of those preferences are adaptive, should they be added to the mix in the first place or should they be left out? Third: we typically think that we shouldn’t interfere with someone who has chosen an action of their own volition that is in line with their preferences; but if those preferences are adaptive, perhaps this presumption against interference fails? It’s this final question that occupies Enoch in his paper, and that’s what I’ll be concerned with here. But it’s quite possible that his answer there will help with the other two questions, if it works.
It is often pointed out that, for almost everyone, nearly all of their preferences are formed in response to their situation, and many will be adaptive in the sense given above. Everyone faces restricted sets of options, even the most privileged among us; and, often, people will respond by adapting their preferences so that they value some of the options that are available to them more than they otherwise would. So if all adaptive preferences are such that actions based on them are less than fully autonomous, and that means there is no presumption against interference with those actions by others, there will be very little presumption against interference in any of our actions, and that sounds awful.
But Enoch doesn’t think that all adaptive preferences suffer from the autonomy deficit. Instead, he seeks whatever it is that distinguishes those adaptive preferences that lead to actions that are less than fully autonomous from those that lead to actions that are fully autonomous; he hopes to say how to tell apart preferences that lead to actions with an autonomy deficit from those that don’t. He gives what he claims is a necessary condition: adaptive preferences that suffer from an autonomy deficit are formed as a result of oppression or injustice. So, for instance, the woman’s preference for her husband and children to have more than enough to eat at the cost of her having too little to eat is formed as a result of oppression, and so is at least a candidate for an adaptive preference with an autonomy deficit. On the other hand, the person who switches from preferring music to philosophy to preferring philosophy to music is not a candidate. The autonomy of the actions that flow from that preference cannot be impugned, at least on these grounds, for that adaptation did not arise from oppression or injustice; it arose when she realised the nature of her talents. A nearby case, in which she is prevented from becoming a great composer by a musical establishment that is prejudiced against women, will be a case in which the adaptive preference is formed as a result of oppression or injustice, but that’s not what happens in Enoch’s example.
I have a rather unruly set of thoughts about Enoch’s proposal that I’ll try to lay out here. It strikes me that there is something right about it, but I have some concerns.
I think we’ll have to use a reasonably broad definition of injustice or oppression to avoid strange consequences. For instance, consider another tweak to Enoch’s example from above. In this one, she is prevented from becoming a great composer not by her lack of talent and not by an establishment prejudiced against women, but by the machinations of a powerful figure in the musical establishment who just particularly dislikes her personally; they don’t dislike her because she belongs to any demographic group; they just take against her. In such a case, there is no oppression, but it is hard to see why the autonomy deficit should be any different—to use the metaphors that Enoch himself uses, the preference seems just as alien to her, just as much a preference implanted in her from the outside, though not by a sexist system, but rather by a mean-spirited mentor.
In this case, perhaps we might say that, while there is no oppression, there is nonetheless a local injustice inflicted by the mentor and that is what’s necessary, not the more structural, global forms of injustice that we call oppression. But we might tweak the example again so that the powerful figure thwarts her ambitions as just retribution for some terrible wrong she did them in the past. So now not even a local injustice is present, yet the adaptive preference is still as alien.
Let’s set this point aside now and move to a different concern. Part of the reason Enoch opts for what he calls a historical-causal criterion by which to distinguish adaptive preferences that suffer from an autonomy deficit from those that don’t—as he says, following in the footsteps of Rosa Terlazzo in this respect—is that he thinks content-based accounts can’t work. On a content-based account, what distinguishes them is the content of the preferences: the woman’s preference in favour of her husband’s overeating at the cost of her own malnourishment leads to non-autonomous actions because it’s simply an irrational or in some other way bad preference. The preference for malnourishment may be hard to rationalise in any situation, but there are other, less severe adaptive preferences we worry about that can be. At the workshop, Natalie Stoljar cited the following example from Virginia Woolf’s ‘Professions for Women’. Woolf dubs this character, whom she paints as a phantom who haunts her mind as she tries to write something critical about a book by a man, ‘The Angel in the House’:
She was intensely sympathetic. She was immensely charming. She was utterly unselfish. She excelled in the difficult arts of family life. She sacrificed herself daily. If there was chicken, she took the leg; if there was a draught she sat in it—in short she was so constituted that she never had a mind or a wish of her own, but preferred to sympathize always with the minds and wishes of others.
Woolf, ‘Professions for Women’
This woman has extreme preferences, in some sense, but preferences that, had they formed in a different way for a different person, we might consider strange but almost laudable. They show an extreme concern for others, a sort of self-abnegation that sometimes occurs in very altruistic people. Were it formed, then, by a man as a result of reflection and a balanced upbringing, I suspect we wouldn’t consider it bad or irrational. So, whatever is problematic about the preferences of the Angel in the House, it is not that the content of those preferences is necessarily troubling. It is the process by which they were formed, and Enoch’s criterion identifies that: it is because the preferences were formed under oppression that they trouble us.
However, that criterion doesn’t seem to be the whole story. After all, consider the following person. He’s a gay man in a severely homophobic society. He values his self-sufficiency, always preferring to rectify things that go wrong in his life on his own and not depending on others for any help. But soon he comes to realise that rectifying the homophobic oppression of his society and specifically the aspects that he experiences—attacks on the street, intimidation by the police, job insecurity—cannot be achieved by him alone. That is, the option of overcoming these problems by himself is not a feasible option. Recognising this, he grudgingly joins forces with an activist group, chafing at their commitment to collective action, but reminding himself always that this is his only route to a better life. Over time, however, he comes to value collective work towards their shared goal enormously—it becomes a key component of his life and what gives it meaning and value. If he were now given the option of overturning the homophobia on his own or banding together with his allies, he’d choose the latter, where before he’d have chosen the former. His preference for collective action to overturn oppression is adaptive, because individual action is simply not available: he cannot achieve his end on his own. And so, while he started off preferring achieving his end on his own to achieving it collectively, he now prefers achieving it collectively to achieving it on his own. His preference is adaptive and formed because of the oppression he faces, but I’d hesitate to say that actions taken on its basis are non-autonomous.
Now, in one sense, this is no problem for Enoch’s account. After all, he only offers it as a necessary condition. And so he can easily respond that this example only shows there is some further condition, beyond being formed in response to oppression, that this man’s new preference for collective action and solidarity does not have, and that renders decisions that flow from it autonomous. But my concern is that the only sort of condition that will work will advert to the content of the preference. We look at the example I just gave and I think we’re tempted to understand it as a case in which the protagonist learned the value of something they didn’t previously think was valuable. And yet all cases in which the adaptive preference is not one whose content is judged bad or irrational on its own can be conceived in this way. How are we to separate out those that suffer from an autonomy deficit from those that don’t?
Another sort of case that’s similar: oppressors sometimes restrict our options in ways that prevent us from being led into forming the bad preferences that they form. We might imagine a society that holds that men should be wholly autonomous authors of their own lives, and thinks they have an innate wisdom that will lead them to form good moral characters without ever being told off, sanctioned, or having their options restricted in any way throughout their childhood and beyond; women, on the other hand, are brought up in this society in much the way many people think counts as good parenting in our society—they are given plenty of freedom, but they are also sanctioned in ways that lead them to form reasonably sound moral values, in much the way we hope good parenting does for everyone in our society. Predictably, men end up with pretty rotten values in this society while women end up with pretty sound ones, having adapted to value the restrictions placed on them during their childhood. It seems this society is oppressive and unjust, treating women as less autonomous and lacking in certain good qualities, such as a sound moral sense, that it imputes to men. And yet the adaptive preferences of the women are preferable to the non-adaptive preferences of the men, and I’d be reluctant to say the women’s preferences lack any autonomy.
So that’s a second concern about Enoch’s criterion. Let me now turn to a third and final one. Enoch rightly rejects what he calls subjectivist accounts that say that the distinguishing feature of adaptive preferences with autonomy deficits is to be found inside the person who has the preferences, perhaps in their higher-order preferences, which disavow those at the first-order level. As he points out, it is in fact a sign of very deep oppression and very troubling adaptive preferences if they are in fact ratified at every level of the hierarchy of preferences, so that the lesbian woman not only comes to prefer romantic relationships with men, but prefers being someone who prefers that, prefers being someone who prefers being someone who prefers that, and so on. But I think there might be a similar concern about Enoch’s own position. After all, a question arises: how do we know whether the system that imposed the conditions that led to the adaptive preference is oppressive or unjust? And one natural way to answer that is to ask whether it gives rise to situations in which the people who live under it don’t get what they genuinely want, have their genuine desires frustrated, and so on. But that suggests that, in a system so perfectly designed to impose exactly the right pressures that give rise to exactly the adaptive preferences that the system will satisfy, we cannot tell whether the system is oppressive without knowing whether the preferences formed are adaptive in the bad way, and we can’t know whether they are adaptive in the bad way without knowing whether the system is oppressive. That is, just as the harmony between different levels of the person’s preferences could be a result of very deep oppression or no oppression on the subjectivist account that Enoch rejects, so giving rise to adaptive preferences that the system then goes on to satisfy could be the result of very deep oppression or no oppression on Enoch’s account. And there will be no way to tell which it is. Of course, if the content of the preferences formed is bad, then that will help, but we’ve already seen it needn’t be.
As I said above, it seems that there must be something right about Enoch’s criterion. But I think it does face these three issues: first, it isn’t clear why a whole system of oppression is required to implant a preference alien enough to be problematic; second, it isn’t clear how we can distinguish the lesbian woman who comes to prefer straight relationships from the gay man who comes to value collective effort against homophobia, both because of the limited options their respective systems of oppression impose; third, it isn’t clear that we’ll always be able to identify oppression without knowing which adaptive preferences are problematic, and which aren’t.
Richard Pettigrew's homepage 