Nudges and Incomplete Preferences

Nudges have become increasingly popular, but there are many objections to their widespread use. Some allege that nudges do not actually benefit people, while others suspect that they do not really respect autonomy. Although there are many ways of making sense of this latter concern, in a recent paper, I develop a new version of this objection, which takes as its starting point the observation that people often have incomplete preferences.
Written by Sarah Raskoff
(Post is based on my recently published paper in Bioethics
First, and again borrowing some terminology from Ruth Chang, the nudge might cause someone to “drift”: they might never settle their preferences, and so find themselves in a career they do not prefer, never having settled whether the other career path would better align with their values. They choose to be a lawyer, not because of a decision about what to value or what kind of person to be, but rather, say, because it feels like the path of least resistance. Such a person lives a less thoroughly autonomous life than someone who has committed to a career as a lawyer, and “thrown their agency” behind the decision. This may come out in the sense of alienation they feel from themselves and their own life pursuits: they never really settled on being a lawyer, and so were not able to rationally decide to become one, but rather simply found themselves with this identity and career. 
The possibility of incomplete preferences raises new and, until now, overlooked ways that nudges might violate autonomy. Specifically, nudging might violate or fail to respect what I call formative autonomy or the sort of autonomy someone exercises when they settle their incomplete preferences by deciding how to weigh their values or what kind of person to be. Let’s consider two ways this might occur, focusing on the example of nudging someone into choosing one career, say, being a lawyer, over another. 
     Should one accept a life‐saving treatment that violates strongly held religious convictions?
Each of these examples involves a decision that is often difficult in the colloquial sense, and where this difficulty isn’t always resolved by slightly improving one of the options, suggesting that such choices are indeed sometimes hard in the technical sense that they involve incomplete preferences.
Second, they might “adapt”: after drifting into being a lawyer, they develop a preference for being one, but only as a result of an adaptive preference for their new status quo. In this case, the cost to their autonomy comes not from the fact that they lack a settled preference, but rather from the fact that their preferences were settled by something external to them, which deprived them of the opportunity to exercise their formative autonomy. In this latter case, the cost to formative autonomy is arguably greater. Rather than being allowed to settle their own preference, the nudger managed to impose preferences on them. One can hardly think of a more paradigmatic violation of autonomy than others imposing their ends or preferences on someone in this way—a clear case of “heteronomy.” 
     Should one medically transition in light of gender dysphoria, and if so, how?
Discussions of nudges often assume a background model of rational and autonomous choice, on which people have values and they have settled tradeoff rates between them, yielding all-things-considered preferences between any two options. For example, suppose Alex is out to breakfast and must decide between oatmeal or pancakes, and that Alex values both health and tastiness. Then, according to this common way of thinking, Alex must have some settled tradeoff rate between these two values, implying that, all-things-considered, they either prefer oatmeal to pancakes, pancakes to oatmeal, or else are perfectly indifferent between the two.
     Should one get a cochlear implant later in life?
     Should one continue with invasive chemotherapy that is unlikely to work or die comfortably in hospice care?
Nudges are small changes in the presentation of options that make a predictable impact on people’s decisions. Proponents of nudges often claim that they are justified as paternalistic interventions that respect autonomy: they lead people to make better choices, while still allowing them to choose for themselves. A classic example is changing the location of food items in a cafeteria so that healthier choices are more salient. The salience of healthy foods predictably leads people to select them, even though they are still free to select the unhealthy options, too.
Chang argues persuasively that we often face hard choices. I agree, and in the paper, I consider several examples, focusing especially on high-stakes clinical decisions, such as:
     Should one continue an unexpected pregnancy or obtain an abortion? 
Of course, this is only a rough sketch and, anyway, not all violations of formative autonomy are all-things-considered wrong. I discuss some of these issues in greater detail in the paper, but leave others for future work. For example, one remaining puzzle, even for those of us attracted to the idea that autonomy can sometimes involve a formative action of settling one’s incomplete preferences, is: how can one possibly do this, especially in a rational or autonomous way? I leave this question as an exercise to the reader (but please do let me know if you figure it out 🙂). 
What this common model overlooks is the possibility of what Ruth Chang calls “hard choices”: cases where a chooser prefers one option in some respects, another option in other respects, but has no settled tradeoff rate between those respects. In these cases, individuals don’t all-things-considered prefer either option, but neither are they perfectly indifferent. Cases of incompleteness are similar to indifference, in that both involve failing to positively prefer either option to the other. But the hallmark of incomplete preferences is that, unlike with indifference, “small improvements” or “sweetenings” fail to resolve this. To take Chang’s most famous example: suppose I neither prefer becoming a philosopher to a lawyer, nor becoming a lawyer to a philosopher. If I am perfectly indifferent between the two and you marginally increase the salary of one of them, I should come to prefer that “sweetened” option. But if my preferences are incomplete, then even after this small improvement, I will still lack a positive preference in either direction.