Different people expect different things from theories of well-being. Some expect that they systematise in a maximally general way intuitions about goods that constitute well-being, others that they states most important causes of well-being, still others that they help them to lead a good life. I for one ask that a theory tells me which constucts of well-being scientists should use and why. Otherwise their research is not value-apt, as I explained in my previous post. There I also gave reasons to suspect that there is no one concept of well-being, but instead many and which one is invoked depends on context.
Suppose you have settled on the context – you are looking to assess well-being of such and such population for such and such purpose. Now you need a theory that characterises the goods that make up well-being for this case. Dutifully you turn to the philosopher of well-being and ask her to tell you what those goods are. Being a good philosopher she lays out the options for you: you could define well-being as prevalence of positive mental states over negative ones, you could define it as a fulfillment of the deepest goals of the individual, or as this individual’s excellence at being a rational human. These are the Big Three theories and there are many variations of each of the themes. As I said in the first installment of this series, these options do not take you very far in terms of justifying the constructs of science, because they are built to satisfy philsophers’ preferences for generality and abstractness and it is very hard to see what these theories imply for the scientist looking for a construct.
In Chapter 2 “Is there a single theory of well-being?” I survey this dynamic and its consequences. Philosophers test their theories by counterexamples: the happy resident of the experience machine, the satisfied pig, the successful counter of blades of grass, the virtuous sufferer, etc. When a counterexample makes a theory look bad, the philosopher amend this theory in one of three ways: biting bullet (‘yes, it is possible to fare well in the experience machine’), going intricate (‘the pig has to experience high quality pleasures’), or going hybrid (‘the grass counter has to be happy as well as successful in his goals’).
But whatever the counterexample, there is one thing philosophers never do and that is to restrict the scope of their theories to contexts or populations. No matter how bad a counterexample is, you will never hear a philosopher say ‘I suppose my theory is not universal but it works in some important set of cases and that’s enough for me’. This is not a done thing. And so the theories and their defenses continue to get very intricate indeed and this moves them further and further away from operationalisability. The theories of well-being that are most successful by the lights of philosophers are not the ones that are usesable by the lights of scientists. I call the theories that philosophers favour – simple, abstract, and applicable to all humans – the high theories of well-being.
High theories have a place. There are plenty of abstract and incredibly idealised theories in sciences that play important roles in measurement and explanation. But they only succeed in these roles thanks to the richness of mid-level theories, i.e. theories that are less universal in scope but more adapted to goals of specific applications (think of the difference between theoretical physics and materials science). This is what is missing in case of well-being: on one extreme there are high theories belaboured by philosophers and on the other there are measures in science. Nothing in between. Much effort goes into ‘validating’ measures and I will talk about psychometrics in the last post of this series, but validation is not a replacement for what is sorely needed in my view: mid-level theories of well-being.
I define mid-level theories as theories of well-being relativised to either purpose, or a kind, or to both of those. Well-being of children in care, well-being of refugees and migrants, well-being of urban youths, well-being of caretakers of the chronically ill, well-being of psychosis patients, and so on. I hope you get the idea. Mid-level theories are as specific or as general as their purpose and use require: there is no one way to carve up nature, it depends on the goals of the project and also on the ability of the kind to support generalisations.
Mid-level theories are not just a necessity if like me you pursue value-aptness for the science of well-being. They are also interesting methodologically – building them makes you realise both the power and the weakness of high theories such as the Big Three. I came to this area from philosophy of science where the practice-turn since the 1980s revealed the limits of fundamental theories and the importance of models, experiments, and instruments. My mentor and advisor Nancy Cartwright forever innoculated me against the worship of the fundamental and gave an appreciation of the diversity and richness of local and tacit knowledge. I brought this attitude to ethics. High theories of well-being are toolboxes full of interesting concepts such as preferences, emotions, capabilities, fully informed selves, virtues, etc. These concepts function as raw material and constraints for building mid-level theories. But the high theories neither imply, nor determine the mid-level ones. (They are not vending machines, I say in the book, borrowing Cartwright’s expression from The Dappled World).
In Chapter 3 I illustrate this process by proposing a mid-level theory for the case of children. Children are not exactly a very specific kind – encompassing teenagers, tweens, toddlers, and babies. So mine is rather high as mid-level theories go. But it is still a good start. When I started looking for a theory of child well-being in about 2009 (after I had my first child) I was distinctly unimpressed with what I saw: philosophers of well-being hardly mentioned children (mostly negatively) and rarely bothered to extend their theories to them. Anyone who ever lived with a baby knows that these are some strange creatures whose well-being is not just a positive mental state, nor the satisfaction of their desires, nor virtuous rational activity. The Big Three are just insultingly bad when applied to children. I argue that child well-being is a function of the nature of children. Their nature is dual: they are creatures of their own kind, but also future adults. So their well-being is dual too: they flourish to the extent that they engage with the world in child-appropriate ways (i.e. with curiosity, exploration, spontaneity, and emotional security), and also develop capacities that prepare them successful future (i.e. grow!). Various concepts derived from high theories are very useful for this theory – flourishing, development, realisation. But these concepts only become a fully fledged theory when coupled with background knowledge about the nature of the kind that children are.
This takes us to about half of the book. Good science of well-being requires good theorising about well-being and a variety of concepts and theories, especially mid-level ones, is needed to secure value-aptness. But there is a lot more to a good science than that. Objectivity and measurement – two non-negotiable constraints on responsible science, at least today – are the subjects of Chapters 4-6 and I’ll tell you about them next time. Thank you again for reading.
I appreciate this delineation of mid-level theories and the need therefor. I have been puzzled by philosophers’ demand for only high theories. When I ask why we should demand this of theorizing, I do not often get arguments. And the arguments I have heard for the demand have not been compelling. Moreover (and as you mention), there are plenty of reasons to think that high level theories cannot always deliver what we need. So, based on my limited experience at least, the demand for only high theories ends up looking like mere convention (or, worse, dogma). So it would seem that your discussion of mid-level theories (if generalized) could improve not only the literature on well-being, but also philosophy more generally. So thanks for that! 🙂
Hi Nick, thank you for this! Of course I share your attitude and delighted to hear you like the idea of mid-level theories. But I wonder if you and I should be puzzled at the preference of philosophers for high theories. They may not be sufficiently informative for some purposes but they provide a common language to connect many different projects in metaethics, normative ethics, political philosophy. I respect this power. What is wrong is that these are the only kinds of theories that philosophy recognises judging by the literature. I hope that is changing. There are philosophers actively working on what I would call mid-level theories (Haybron&Tiberius’s pragmatic subjectivism for policy, Alicia Hall’s theory of well-being for medicine, probably more I am ignorant of…).
Are the mid-level theories intended to be neutral with respect to the high-level theories? Or to replace them altogether?
The theory of child well-being outlined above doesn’t sound neutral with respect to high-level theories; it sounds Aristotelian and virtue-based. Presumably, a committed hedonic utilitarian will not be too worried by children engaging with the world in child-inappropriate ways if it makes them subjectively happier.
Hi Jonathan, thank you for reading and for this very pertinent question. Yes and no is the answer. Yes, mid-level theories can be neutral wrt to high theories because they are evaluated on different grounds. A good high theory systematises intuitions and desiderata in a maximally simple and general way, whereas a mid-level theory enables measurement, informs deliberation and decision-making, etc. Because of these different requirements a mid-level theory can be developed on the basis of one high theory on grounds that do not show that a different high theory is incorrect. So as you wisely notice my theory of child well-being is built up from Aristotelian resources (in particular Richard Kraut’s developmentalism) but I don’t think that this alone shows that high theories in terms of mental state are somehow disconfirmed or even undermined. But at the same time I believe it matters whether or not a given high theory is fruitful for mid-level ones. If it was consistently useless for this purpose, then I would think it should count against it, its theoretical virtues notwithstanding. I hope I understood your questions right.
Thanks Anna! Yes, I see what you mean about the criteria of evaluation being different. At the same time, it also sounds like some combinations of high- and mid-level theories will mesh with each other much better than others.
For example, if your favoured high-level theory is hedonism, and then your preferred mid-level theories are all about cultivating virtues appropriate to your nature, the overall picture looks incoherent. Some serious work will have to be done to argue that it is in fact coherent. Mid-level theories that tell you the virtues a particular kind of person should cultivate seem to mesh far better with a high-level theory that is also virtue-based.
Yes it does sound incoherent, you are right. I just don’t know if I want to require a theory holder to obey any particular coherence constraints that connect her preferred high theories to her mid-level theories… Hmm… I am tempted to say that a commitment to a high theory is not a commitment to any claim about the nature of well-being, but a commitment to usefulness and importance of certain categories for various purposes (that’s the thought behind the toolbox view in Ch2). So in your example of a ‘high hedonist’/’mid-level eudaimonist’ the problem is not that they accept one set of claims at the high level and a totally different one at the mid-level, but that their high hedonist tools are laying fallow when it comes to their mid-level goals. The failure for such a theorist is a certain epistemic inefficiency. Thank you for pushing me on this, I probably did not think this through properly in the book…
> I am tempted to say that a commitment to a high theory is…a commitment to usefulness and importance of certain categories for various purposes…The failure for such a theorist is a certain epistemic inefficiency.
Eeps. I’ll duck most of that, and set aside theorist failures, and bite on a corner only…the notion of philosophical integrity across (in philosophy of science terms) high- and mid-level theories is unduly neat, and likely sometimes highly inefficient. I feel rationality crowding me with your above, Anna, even if your definition (high theory as a commitment to particular usefulness) seems ok. There’s an amorphousness to high-level theory we should feel comfortable abandoning. First, of all, it’s hard enough to get any mid-level model to make sense; we’re like children trying to throw rings at a carnival to win a bear, and we’ll take what we can get. One should run with what seems to predict best in given environs, with a strict agnosticism on the belows and aboves. Anyway, any lower-level useful schema will typically get more and more pragmatic and contingent and changeable in ways that often has to ultimately be addressed using very strange high-level theory indeed, as what has happened with quantum theory and QCD. So, no ‘but-but-but-I’m-a-Stoic’s allowed, at least not without exhausting contrasting mid-level theories that, yes, may leave ideas fallow, or contrast enough with high-level guidance to drive you to drinking.
I believe we do this melange all the time, akin to moving from digital and analog choices for design; or in implementation, such as believing a particular political paradigm while using an ironically ideologically contrasting implementation on the ground. Or addressing one another across racial or gender or species divides, as contingent high- and medium-level models are shuffled on and off stage vigorously. It’s not just a question about science method, per se. Rorty helped get across this essentially ironic nature of life as a mishmash of high- and mid-level models that simply can be observed to conflict in important ways when one is optimizing science well, or working a successful actual life on the ground. To me, that’s much of what constitutes the grand mystery nowadays: that the levels and competing useful models don’t get to be reconciled (at least for me.) That we must hold such disparate views all at once to capture complex information about life, and that potential weighty differentials such as the hedonic-vs-virtuous dichotomies across hierarchical levels of models are there to be exploited, not to winnow away. In that sense, Jonathan, your use of the term “coherent” also felt like an unnecessary damming up of possibility, even if it strains the notion of rationality to relax a requirement for coherence, particularly for hedonists 🙂
I think we have to shed a part of our rationality with our shoes at the door. That buys us a lot. Contrasting-by-design high-level requirements (differential between accepted high-level rqmts and a mid-level’s high-level requirements) lends a certain breathing room to the whole affair. We’re leaving the back door open and the back porch light on for a guest to arrive, and maybe stay. If there’s some normative value to being a thru-and-thru anything, I’m not sure it would amount to more than clear toasts at family events about one’s reliable nature some long day hence. Typically, when I see such unification projects in a person– one does– I wonder more about the cost than the gain.
It does get more involved, in the spirit of the nonlinearity we’re awash in, and various uncertainty principles: multiple competing high-level and multiple mid-level theories should be used in experiment, if possible, to capture effects usefully. This has parallels to the notion of only sub-personalities being recruited to handle life, never one’s whole personality or any predictable “essence.” Imagine, for instance, if the Skinnerians hadn’t been so goddamn sure of themselves, and so obstinate about the ‘whole truth’ of their model, their “sub-personality”; imagine all the shit we could’ve learned, both high- and mid-level, if we could’ve piggy-backed on those studies with other models while they plodded onward for decades, getting the wrong half right, with all that funding. But that’s changing the subject a bit. The point being the reminder that heterodoxy in high-vs-medium models typically allows for both the capture of more and/or better information (dimensionality) and incremental explanatory/narrative power.