• Autonomy
• Behaviour Modification
• Health Promotion
• Public Health Ethics

Policy makers are understandably interested—for both political and moral reasons—in following Thaler and Sunstein's recommendation to use ‘choice architecture’ (as in Cafeteria ), or other ‘nudges’, to promote desirable behaviour in ways that are allegedly compatible with personal freedom.1 Yashar Saghai's intricate analysis shows that simply maintaining the target's choice-set is insufficient to preserve the target's freedom when the nudge bypasses the target's deliberative capacities—as it is specifically designed to do.2 In his friendly amendment to Thaler and Sunstein's project, Saghai advances a more robust account of nudges in which fewer policies would count as nudges, but those that do count as nudges would be less troublesome. In this commentary, I briefly discuss several issues raised by Saghai's project, some of which go beyond the topic of nudges.

• How important is definitional salvaging? If a ‘substantially controlling’ influence were more efficacious in promoting healthy behaviours, we would have a trade-off between the efficacy of a policy and its compatibility with freedom. Saghai states that an influence can be morally permissible even if it is ‘substantially controlling’ and, therefore, does not qualify as a nudge. So even on Saghai's definitional project it is not clear how much work it does in distinguishing the morally permissible from the morally impermissible.

• The choice-set. Saghai claims that A interferes with B's freedom if A reduces what is B's choice-set prior to A's influence. Not necessarily. If B is overwhelmed by her current array of options, removing some unimportant options may enhance B's deliberative capacities and, perhaps, her freedom.

• Substantial non-control. Even when policies preserve the target's choice-set (as in Cafeteria), they do not preserve freedom of choice if the influence is ‘substantially controlling’. This condition raises several issues.

  1. Efficacy. It is not clear whether an influence can be both extremely efficacious and also substantially non-controlling. Saghai argues that a choice is substantially non-controlling only if it is easily resistible. Despite Saghai's considerable efforts, I find ‘resistibility’ to be an elusive concept. Can an influence be both easily resistible and yet rarely resisted? If it is frequently resisted, then it is clearly non-controlling, but does it follow that if it is rarely resisted, then it is substantially non-controlling and incompatible with freedom of choice? Could people easily resist an influence but (virtually) always choose not to do so?

  2. The Controllability Continuum. I do not think it best to think of forms of influence as lying on a continuum from the fully controlling to the fully non-controlling. It is the mechanism of control that affects its compatibility with freedom and not its degree of control. If ‘carrots’ are more effective in influencing B's behaviour than ‘sticks’, it does not follow that they are less compatible with B's freedom. In that sense, it is simply not true that coercion or threats are always more controlling than incentives or offers, unless we say that A does not coerce B to do X unless A's threat to harm B actually gets B to do X, but then coercion is controlling or efficacious simply by definition. And incentives can be extraordinarily powerful. If A offers to make B much better off than B's present situation if B does X, then A's proposal may be extremely efficacious, difficult to resist and, therefore, substantially controlling. But is there anything wrong with making attractive offers that are genuinely too good to refuse, and that the target is happy to receive? If not, does it follow that they are non-controlling?

  3. Resistibility and wants. Saghai maintains that A's influence to get B to do X is substantially non-controlling when B could easily not do X if she did not want to. But, what if A's influence changes B's wants? Suppose B has not been much concerned about his diabetes and has resisted A's attempts to get him to exercise. A tells B that diabetes may cause impotence, and that changes (nudges?) B's desire to exercise. Is A's influence substantially controlling or not?

  4. Resistibility and deception. It is not clear why the veracity of information affects its resistibility or controllability. In Less than you think, it is the students’ non-deliberative desire to conform to what they take to be the non-binging norm that seems to be doing the work. I am not (but might be) persuaded that the technique is substantially non-controlling if truthful, but substantially controlling if accurate.

• Framing. Also, one can tell the truth in various ways. Suppose a physician could tell his patient that a drug would: (a) reduce the risk of a disease by a proportion of 50%, or (b) reduce the risk of a disease by the absolute amount of 1% (from 2% to 1%). Both are truthful, but (a) is more likely to get the patient to take the drug. If a physician thinks the patient should take the drug and therefore chooses (a), is this a nudge? Is it controlling? Is the patient's consent valid?

• Authority. A can attempt to influence B by implicit or explicit appeal to his authority, as when the oncologist says, “I think you should have surgery”. B could resist by not deferring to A's judgment, but he does defer. When is A's authority (or B's trust in A) compatible with B's freedom?

• A self-determined life. Saghai states “some health-related choices are so fundamental for leading a self-determined life that they ought to be fully noncontrolled by others”. He specifically mentions reproductive choices. Suppose that the tax code affects people's decision to have children by providing a tax exemption for each child and deductions for certain child-related expenses. Does the tax policy compromise ‘full non-control?’ If so, are such policies therefore morally impermissible?

Finally, we might say that some choices are so fundamental to one's health that the opportunity to live a life, or self-determined life in the future, would be compromised if people made the wrong choices now. Can we justify policies that compromise present self-determination for the sake of future self-determination? Saghai's position implies that the answer is ‘no’. This is worth further exploration.