Recommendations

Often, well before discussing a patient’s medical options with her, a physician will have already identified an option she believes best for the patient, where ‘best’ here is to be understood as facilitating the patient’s welfare, with allowances for the patient’s interests. There can be sensible debate as to what extent physicians should be concerned with patients’ interests beyond their mere biological or psychological health, and there is reason to think that physicians are far less likely to understand a patient’s interests than the patient herself; but these topics are outside the scope of this paper.

When the option the physician believes best is far preferable to a patient’s other options, adequate disclosure requires the physician to recommend this far preferable option. A physician who recommends an unnecessary treatment to benefit herself financially or professionally, rather than to benefit the patient, does not act as a physician because she does not act with appropriate concern with regard to the patient’s well-being, and it doesn’t make sense to call this a genuine recommendation.

Some proponents of nudging contend that recommendations are nudges.5 After all, patients are more likely to choose the option their physician recommends than other options.

Recommendations are not nudges. Rather, recommendations are often a necessary part of adequate disclosure. Patients see their physicians as medical experts interested in preserving and restoring their health, as such a physician is expected to employ this expertise to benefit their patients medically. If a physician does not recommend one option over another, it might suggest to the patient that one option is just as good as another. If there is a medically significant difference between options, a patient who does not understand this can scarcely be interpreted as understanding their options, and the physician has failed to adequately inform.

Suppose a physician diagnoses her patient with rabies, then informs her patient of her options—(1) treatment by a rabies vaccine, expected to result in her continued life and (2) refraining from treatment, expected to result in her death. The physician is likely to believe (1) is a far better option than (2), such that she does not need to explicitly recommend (1). However, suppose the patient has trouble deciding; the physician should take this as evidence that the patient doesn’t understand her options! In light of this, the physician should recommend (1) in an effort to adequately inform the patient. A physician that expresses agnosticism regarding whether a patient with rabies ought to get the rabies vaccine fails as a physician.

Of course, a competent patient is free to choose other than the option her physician believes best—whether recommended or otherwise, perhaps motivated by fear, scepticism, misinformation or non-medical concerns like cost and religious beliefs. Recommendations are tools to help clarify difficult-to-understand medical complexities, but if a competent patient can demonstrate that she understands, the physician has good reason to believe she has received genuine informed consent regardless of what the patient chooses.

The reason proponents of nudging mistake recommendations for nudges is that patients are more likely to consent to options that their physicians recommend than those they do not. This is to say that they predictably alter the patients’ behaviour. However, by definition, a nudge alters behaviour (a) without altering a patient’s options (b) through some means other than rational persuasion.1 2 Appropriate recommendations fail to satisfy criteria (a) and (b).

As patients see their physicians as medical experts interested in their welfare, a recommendation from their physician is a reason to choose that option over other options and it would be hard to call anything a genuine recommendation if the physician didn’t believe it best and incoherent to interpret the physician making a recommendation (whether genuine or otherwise) as doing anything other than trying to give their patient a reason to choose that option. As such, recommendations fail to satisfy criterion (b).

A recommendation from a fool is not persuasive. Of course, a physician can abuse her power to make false recommendations, but false recommendations are lies intended to give the audience false beliefs that may influence how they act. If securing informed consent requires truth-telling, false recommendations are obvious at odds with securing informed consent.

A proponent of nudging might argue that recommendations don’t change a patient’s actual options at all, but rather either their beliefs, feelings or relationship to these options. However, this semantic point is ridiculous; if we take this distinction seriously, then nothing a physician says about a patient’s options counts as substantively altering or restricting their options, as it only substantively alters or restricts their perceived options. As such, it seems that recommendations that inform a patient about their options fail to satisfy criterion (a).

One might argue that recommendations can be nudges when they predictably, but inappropriately, influence patients to doing something. After all, many patients trust their physicians qua experts interested in their welfare to recommend what’s best, and a physician might recommend one option as it is slightly better than another option, and the patient might misconstrue this as a ringing endorsement that the recommended option is far better. In such a case, though, it is clear the patient doesn’t understand. However, let’s suppose a physician makes such a recommendation, the patient understands the recommended option is slightly better than the other option, but is psychologically predisposed to be disproportionately enamoured with the recommended option merely because it is recommended. Recommendations are sometimes necessary to secure genuine informed consent, but might predictably lead to irrational behaviour. Are these nudges? The answer depends on the physician’s intent. Can a physician intend to recommend to inform and irrationally exploit? These intentions seem mutually exclusive.

Framing

To secure informed consent, physicians are required to convey information to their patient so that the patient will come to understand her condition. Physicians often have a choice in how they convey this information, and there is substantive evidence that suggests how physicians frame their disclosure can influence how patients act.2 8 14 Cohen characterises framing as a kind of nudge, claiming that a physician can try to influence a patient’s decision through choosing non-medical terminology so as to not cause the patient to be anxious.2 Blumenthal-Barby and Burroughs note that whether a physician frames a patients’ risk in absolute terms or relative terms can reliably change how the patient acts.8

As physicians are required to convey information to their patients to secure informed consent, when there is more than one way to frame information, a physician will be required to frame information in one way or another. If this counts as nudging, then nudging is unavoidable when securing informed consent as physicians need to convey information in some way.

However, appropriate framing is not nudging. Those in favour of using frames to persuade patients assume that although multiple frames might convey the same medical information, some frames can convey (or fail to convey) how patients ought to feel. For Cohen, framing can avoid patient anxiety, while Blumenthal-Barby and Burroughs suggest that framing might influence a patient’s fear.2 8 Either fear and anxiety are appropriate or they are not. If either is appropriate, and a physician frames disclosure to avoid it, then it doesn’t make sense to say that patient understands her options, and thus inappropriate framing fails to satisfy adequate disclosure.

Purposely misleading framing seems to count as lying, as it tries to give patients inaccurate reasons to choose to act, and insofar as it is concerned with giving patients reasons, it cannot be considered nudging.

Once again, let’s consider the physician who diagnoses her patient with rabies and informs her of her options. Suppose the physician frames the patient’s risks in cuddly language designed not to upset her patient, and this leads the patient to not be (appropriately) concerned with her impending painful death if she does not consent to the vaccine. The patient fails to understand the risks and benefits of her options, so the physician cannot reasonably interpret her patient as giving informed consent. Furthermore, the physician cannot reasonably suggest that she tried to inform her patient if she chose a frame to mislead the patient about the appropriate anxiety or fear she should feel.

Appropriate framing, like appropriate recommending, is consistent with, and often required during, the process of securing informed consent because it helps a patient understand her options, risks and benefits; prerequisites for her making an informed choice.

The reason proponents of nudging sometimes confuse framing for nudging is that they equivocate between a frame’s technical meaning and how a patient understands the conveyance; two frames can convey the same technical medical information in two different ways (eg, one might frame it in chances of survival, the other in chances of death), but one frame might reliably cause a patient to be more anxious, fearful or hopeful than the other due to unavoidable emotive and/or non-propositional content that can reasonably be expected to alter behaviour. If this difference is a by-product of patient irrationality, it would be a matter of nudging. If this is the case, then it’s not smooth sailing for framing–nudges, as there is some appropriate level of anxiety, fear or hope a patient should have in response to a set of information, and if the patient fails to experience the appropriate amount, this is evidence they don’t understand and thus cannot give truly informed consent.

Note, however, that proponents of misleading framing cannot reasonably suggest that the patient interprets two frames that elicit different emotional responses differently as conveying the same information; after all they predictably react differently. They may be unable, unwilling or simply uninterested in explaining how the patient misunderstands certain frames, but their difference in reaction just is evidence of a difference in understanding. To assume this is irrational simply because it is unexplained is highly dubious, but in many cases the difference is rational, as one frame emphasises one thing and another doesn’t (eg, a frame that conveys information in terms of chance of survival might emphasise survival, while a frame that conveys information in terms of chance of death might emphasise death). Physician emphasis, like physician recommendation, gives patients reasons to act upon, and thus appropriate framing, like recommendations, would give patients reasons to act, rather than influence them irrationally.

Nudging during disclosure

In ‘Nudging and Informed Consent’, Cohen proposes a variety of methods in which physicians can attempt to nudge their patients towards their performed option while securing informed consent; here I will summarise his most developed nudging proposal: discouraging disclosure.2 Cohen argues that in some situations, physicians should steer patients towards their preferred option by deliberately crafting their disclosure to discourage the patient from continuing to listen. If successful, she will have led the patient to consider only a select portion of the information she would otherwise have disclosed to the patient. He says:

There is rarely an objective measure to the appropriate amount of information the doctor should provide; there is rather a spectrum of reasonableness. Now, within that spectrum, the doctor may legitimately offer the larger amount of relevant information, anticipating it may test the particular patient’s patience to listen. A typical reaction by the patient could then be to forgo his or her right to hear more and request the first option presented, provided it is sufficiently good. By explicitly declining further information, the patient allows the doctor to not even present those options the doctor thinks are better not chosen.2

Here, nudging involves ordering the disclosed information to annoy the patient, rather than to inform the patient. The physician purports to act as an educator, but chooses their words not to educate, but to annoy. She does not want the patient to come to choose an option through rational deliberation and command of the relevant facts of the situation, but rather because the patient is frustrated by the physician. The physician buries information she doesn’t want the patient to understand under trivial information and/or technical terminology (which, in turn, might need to be explained, further obfuscating), hoping the patient will get annoyed and either ask the physician to stop disclosing or merely stop listening to her before the physician gets to the information she believes the patient will find compelling.

Proponents of nudging might contend that the physician has two goals here—educate and nudge. However, this doesn’t seem to be the case, as rather than present information in such a way as to encourage the patient to listen and understand or to test the patient to make sure the patient understands, the physician deliberately delivers information to discourage the patient from paying attention. Such actions would be unthinkable in the classroom. Cohen clearly doesn’t expect patients on the end of this disclosure to walk away understanding all of their options; instead, he seems content that the patient says that she understands them and waves her right to further disclosure.

Non-propositional conveyances are bullshit

Proponents of nudging might contend that nudges, in principle, cannot be truth-telling or lying, as they lack propositional content. For example, expressions like ‘Yeah!’ and ‘Boo!’ cannot be either true or false; perhaps some nudges share the same lack of true value. Non-propositional conveyances of this kind are frivolous and pointless if they’re not designed to influence anyone, and bullshit if they are intended to get a target to do something.

If adequate disclosure involves non-propositional conveyances, one might argue that my assumption that truth-telling requires truth-telling is false, as it is inconsistent with non-propositional conveyances. However, this is a mistake. On Frankfurt’s definition of truth-telling, one tells the truth if and only if one conveys x, believes x and wants their audience to believe x as well. One can tell the truth about something even if one expects her audience to misinterpret or misunderstand their conveyance; however, for a physician to interpret her patient as giving informed consent, she also has to believe that her audience understands x. If a physician tells the truth and the patient doesn’t understand, she cannot reasonably interpret this as receiving genuinely informed consent from her patient.

However, this seems to be beside the point. Either the physician chooses to convey x to inform the patient or to nudge them. If the physician chooses the former, even if the patient misunderstands, the physician tells the truth. If the physician chooses the latter, even if the patient comes to understand, the physician bullshits.

Perhaps proponents of nudging can construct a situation wherein the only way a patient can come to understand her situation is if her physician bullshits her. This would be quite an interesting scenario, but I don’t think it is a genuine threat to the theory that adequate disclosure requires truth-telling. Students in introductory ethics classes routinely construct scenarios where a physician produces utilitarian benefits by lying to the patient; consider the following:

Introductory Problem Case for Informed Consent: Jones has a rare condition such that she will die if her physician tells her she’s sick and will live if her physician tells her she is healthy. She is of no danger to anyone else.

Perhaps in cases such as these, and in cases where understanding requires bullshit, the normal requirement that physicians tell the truth is void. If we were interested in constructing a full conception of the scope and appropriate behaviour of physicians, we might require some foundational principle that usually requires truth-telling during the process of informed consent, but sometimes, in these rare cases (which, by their very nature, seem to have to be exceptions to the rule, else they would undermine the patient’s willingness to trust the physician at all) lying or bullshit is acceptable. However, this project is outside the scope of this paper and is nothing new.

However, for all intents and purposes, this discussion is irrelevant. The notion that nudging involves genuinely non-propositional conveyances is quite suspect. In discouraging disclosure, the physician orders his disclosure in a particular way to get the patient to choose one of the few options she bothers to listen to. Although this ordering doesn’t explicitly say that the early options are better, it clearly implies this.

A competent nudging physician might be able to nonexplicitly imply that some options are better than others simply through ordering them. A physician can only intentionally deceive a patient via ordering if the physician believes the patient will, consciously or otherwise, map value onto the order. Furthermore, outside of some transparently arbitrary ordering (perhaps listing options on pieces of paper drawn at random from a transparent hat), the physician should assume the patient ascribes some relevance to the order (even if this relevance was merely ‘of like kind’ rather than a value ascription).

Cohen discusses a variety of nudging methods that might be misconstrued as non-propositional, suggesting physicians can choose their mannerisms or attire to influence patients.2 However, Scott D. Gelfand notes that this kind of behaviour can have proposition content, implying something like (appropriate or inappropriate) optimism or urgency that can give patients reasons to choose one option over another.22 If a physician chooses to dress or act in such a way as to convey such things so that their patient believes them, they tell the truth or lie. When the physician conveys this information, believing it to be true, so that her patient will believe it to be true and act as though it were true, she tells the truth. When the physician conveys this information, believing it to be false, so that her patient will believe it to be true and act as though it were true, she lies in the very same way she might lie if she was to give a false recommendation or misleading frame to persuade her patient.

Proponents of nudging might argue that patients are unaware of any propositional content, but remember that Frankfurt’s account of truth-telling, lying and bullshit turns on the speaker’s intent, not the audiences’ awareness of the speaker’s intent.

Gettier-style cases of informed consent

Gettier-cases are counterexamples to the theory that one has knowledge if and only if one has a justified true belief.23 For example:

Little Red: One day Little Red goes to her grandmother’s house to deliver baked goods. Her grandmother is feeble and spends most of her day sleeping in her bed. She enters her grandmother’s house and sees someone in her grandmother’s bed. She hears snoring. She comes to believe that ‘My grandmother is home asleep.’

As it so happens, her grandmother was tied up by an anthropomorphic wolf, thrown in the closet and put to sleep using sleeping gas so that she couldn’t struggle or make noises. The wolf then dresses in her clothes and hides in her bed, playing a tape recording of Little Red’s grandmother’s sleeping.

Little Red has a justified true belief; her belief is true because her grandmother is asleep in her closet. However, her belief is justified by unrelated features of the case—the implausible tape recording and the implausible wolf in her grandmother’s bed. Thus, although Little Red has a justified true belief, we’re inclined to not call it knowledge.

In ‘The Gettier Problem in informed consent’, Shlomo Cohen discusses cases of informed consent that he says are Gettier-case like; for example:

Gambler: Competent patient Alfred is diagnosed with a cataract in his right eye, and his physician explains his options, their risks and rewards and recommends surgery. A stickler for informed consent, his physician requires Alfred take a written exam, which he passes. His answers are novel and nuanced and better than any textbook, demonstrating his nuanced understanding. However, he refuses treatment and leaves.

Later, Alfred wagers on the outcome of a game; the loser must fulfil one request of the winner. Alfred loses. The winner requests that Alfred get the surgery. Alfred returns to his physician and consents (modified from Cohen’s case).24

Cohen wonders whether Alfred has given genuine informed consent here, suggesting that the kind of consent Alfred gives is given for the wrong reason.

It is easy to construct a similar case where a patient chooses based off of unintended but unavoidable misconceptions during adequate disclosure:

Foolhardy: Competent patient Bert is diagnosed with a cataract in his right eye. As it so happens, prior to disclosure, his physician came to learn that Bert loves to get surgery. His family explained that he thinks it’s brave, and foolhardily opts for any surgeries regardless of context. However, the physician chooses to disclose as normal, explains Bert’s options, their risks and rewards and recommends surgery. He is aware that the term ‘surgery’ might incline Bert to choose the surgery regardless of all this, but uses it anyway. A stickler for informed consent, his physician requires Bert take a written exam, which he passes. His answers are novel and nuanced and better than any textbook, demonstrating his nuanced understanding. Bert consents to the surgery.

The physician has good reason to believe that Bert’s consent is given because of Bert’s preconceived notions about surgery rather than for the "right" reasons; however, it does not make sense to say that his physician nudges him, as his physician doesn’t choose his disclosure to unduly influence him, even though the undue influence is predictable and perhaps even unavoidable.

This does not constitute bullshit, nor nudging, as the physician does everything he can to tell the truth. We might say that the physician should make it clear that he’s aware of Bert’s predilections and disapproves of them, but the key feature here is that truth-telling turns on the intent of the conveyor, and informed consent requires only understanding and competency, not that one consents for the right reasons.

Cohen’s contention that such cases are relevantly similar to Gettier-cases turns on a confusion, as informed consent does not require a patient make a rational choice, even if a paternalistic physician wishes their patient to make such a choice.

Multitasking

Suppose that a physician wants to do two unrelated things during the process of securing informed consent—tell the truth about a patient’s options and bullshit a patient into choosing an option. Is this acceptable? Perhaps the proponent of nudging contends that one can have two different intentions with a single utterance, to inform and to deceive. I have my doubts that this could be done, but I look forward to proponents of nudging constructing cases meant to satisfy both truth-telling and bullshit at the same time.

That said, I think this enterprise is doomed before it starts because it constitutes multitasking, and multitasking is prima facie inconsistent with obtaining informed consent. By assumption, adequate disclosure during the process of securing informed consent requires truth-telling. But consider the following cases:

Juggling: Carol is diagnosed with a cataract in her right eye. Her physician explains her options, their risks and rewards, makes a recommendation and tests Carol, who passes with flying colours. As he does this, he juggles chainsaws because in the off hours he is a chainsaw juggler for a local circus. A professional, he juggles the chainsaws behind a glass divider such that he doesn’t risk Carol’s life. Carol consents to surgery.

Seduction: Diane is diagnosed with a cataract in her right eye. Her physician explains her options, risks and rewards in a seductive tone, makes a recommendation by whispering it in her ear and tests Diane while reserving a hotel room for what he thinks is a possible post-doctor/patient relationship sexual liaison. Diane passes the test with flying colours. Diane is rather flattered by the attention, but is not sure it is appropriate. Still, she consents to the surgery because he is an expert, resolving to decide whether to report him for sexual harassment and/or consent to his advances after the surgery.

Twin Disclosure: Mary Kate and Ashley are competent conjoined twins diagnosed with brain cancer. Mary Kate is blind and knows English while Ashley is deaf and knows sign language. Their physician knows both spoken English and sign language and chooses to explain their options, risks and rewards simultaneously. After all, he’s a busy physician. He recommends surgery to each patient, then tests them. They pass with flying colours. Both consent to the surgery.

These examples are absurd. Although some truth-telling in adequate disclosure may be easy, most involves educating patients about complex issues that they are psychologically and intellectually at a disadvantage to understand. The notion that the physician would juggle chainsaws while expecting his patient to get the same education as though he did not multitask is ridiculous. The behaviour in Seduction is similarly unprofessional. In Twin Disclosure, we might be inclined to think that because both twins are selectively unable to interpret the physician’s other behaviour, the disclosure is unproblematic. I think not; the fact of the matter is that in addition to multitasking’s unprofessional nature, it also needlessly opens up the possibility of error. No doubt there are people who can successfully text and drive without substantially increasing the chances of getting into an accident, but it seems foolhardy to expend much philosophical ink arguing that professional ambulance drivers have the right to text and drive.

Thus, the proponent of nudging who argues that nudging can be a distinct, and simultaneous, act with truth-telling has two hurdles to overcome—(1) she must be able to construct a scenario where a single action is both intentionally truth-telling and bullshit and (2) argue that the bullshit uncontroversially does not threaten the physician’s professional moral obligation to inform her patient. However, either (a) the patient is aware that the physician is nudging or (b) she is not. If (b) she is unaware, it’s hard to believe that we can interpret her consent as truly informed because she doesn’t know what is going on! However, if (a) she is informed that in addition to telling her the truth her physician might be trying to exploit cognitive weaknesses to predictively coax her to choose a particular option, her job of interpreting her physician’s conveyances has become that much harder. If either (a) or (b) is the case, then it seems (2) is false.