Defective Normative Powers: The Case of Consent

The Justification and Ontology of Consent

To fully appreciate the difference between the behavioral view and the mental state view, we need to clarify the role that communication plays within these two approaches. Defenders of the mental state view do acknowledge that communication plays an important role in our practice of giving consent. For once you have consented to my φ-ing by forming the relevant mental state, you need to let me know that this is the case, so that I can adjust my behavior accordingly. However, the role of communication here is simply to inform me that the relevant normative change—the creation of a permission for me to φ–has already occurred (Alexander 2014: 102). The ‘magic of consent’ (Hurd 1996) happens when you form the mental state.

Not so for the behavioral view, according to which the relevant normative change happens only once the communication, or at least an attempt to communicate, has taken place. At the heart of the behavioral view is the claim that ‘the communication of consent matters independently of its instrumental role in informing other agents about the content of our will’ (Healey 2015: 357, 358; see also Dougherty 2015: 249; Manson 2016: 3318).

How should we go about adjudicating this matter? The debate so far has proceeded in an unsatisfactory way. Some have noticed that the early literature tended to rely excessively on intuitions, with philosophers in each camp selecting cases that were intended to support their own view but which ultimately seemed to trigger different responses in different people (Dougherty 2015: 230–31; Millum and Bromwich 2018: 62–63). To avoid this problem, the main strategy pursued in the most recent debate has been to look for an account of consent that coheres with other normative powers, such as abandoning, promising, or commanding. For example, Tom Dougherty has argued that since consent is the counterpoint to promising (consent turns an existing duty into a permission, whereas promising turns an existing permission into a duty), we should expect the two to work similarly. And since promising requires communication, we should expect consent to require communication too.⁵ On the other hand, Larry Alexander and Kim Ferzan have observed that with abandonment, all that is required to waive one’s right over property is that certain mental states are formed, whether or not they are communicated to anyone. If this is sufficient for abandonment, why doubt that consent could operate in the same way (Alexander 1996, 2014; Ferzan 2016)?⁶

This strategy, however, is not a significant improvement on the one it meant to replace. For how should we establish whether consent operates like promising or abandonment? The worry is that in presupposing that it works like promising, Dougherty is already assuming that consent needs to be communicated, and in presupposing that it works like abandonment, Alexander and Ferzan are already assuming that it doesn’t.

Dougherty, for example, writes, ‘If communication is needed to release someone from a promissory duty, why is communication not also needed to transfer the consent that releases someone from a duty?’ (Dougherty 2015: 236). But why think that we are released from our consent-sensitive duties in the same way in which we are released from our promissory duties?⁷ Given that these two normative powers are importantly different, this is hardly surprising. For example, many believe that whereas promises need to be accepted to be morally binding, consent does not. If so, this might explain why promising requires communication whereas consent does not (Pallikkathayil 2020).

On the other hand, Ferzan argues that insofar as the power to consent, like the power to abandon, is grounded in the value of autonomy, we should accept that no communication is required for its exercise (2016: 404–07). But according to Ferzan, the power to promise is also grounded in the value of autonomy (404). So how can the appeal to autonomy justify the claim that consent operates like abandonment rather than like promising?

Far from helping us answer the ontological question, the strategy of modeling consent on other normative powers seems to replicate the problem of relying on conflicting intuitions about whether communication is required for consent. The strategy is unhelpful, unless much more is said about why consent is relevantly similar to these other normative powers.

Here, I will pursue a different strategy, one that ties the question of the ontology of consent more closely to the question of the justification of the power to consent.⁸ Start by thinking about why we have the power to consent. We have the capacity to change our normative relationship with others by way of consent because having this capacity is valuable to us in some respect (Raz 1972). But plausibly, there is a connection between how the power operates (the ontological question) and why we have the power to begin with (the justificatory question). If the existence of the power to release others from obligations they owe us is explained by the fact that being able to exercise such power is valuable to us in some respect, it is only to be expected that the power will operate so as to facilitate the realization of whatever makes the exercise of the power valuable.

Of course, there will be moral constraints on how the power can be exercised, since the autonomy of the consenter is only one of the interests protected by consent. (More on this below.) But with this caveat in mind, what could possibly be the reason to think that the power operates in any other way? If what justifies the existence of the power is that having it serves some valuable moral purpose (a big if, you might think, but I’m asking you to accept this here for the sake of argument), then we should expect that the functioning of the power will be shaped so as to best enable us to realize the relevant purpose.

So, what is it that makes the exercise of the power to consent valuable? A plausible answer appeals to the value of autonomy. Being able to control the consent-sensitive duties that others owe us is valuable because it increases our capacity to exercise our self-determining agency and shape our life as we wish. For example, I can realize the plan of staying healthy by (among other things) consenting to surgery; I can realize the plan of keeping plants in my apartment by consenting to my neighbor watering them while I’m away; I can realize the plan of feeding my child by consenting to you taking my money in exchange for pizza.

It would be much harder for me to pursue any of these projects if I weren’t able to make it the case that others do not wrong me by touching me, by entering my apartment, or by taking my money. If I lacked that power, others might well refuse to do any of these things, which would constitute acts of wrongdoing. And even if they could be persuaded to play along, the realization of my plans would be secured only at the cost of turning the consentees into wrongdoers and of me becoming a victim of wrongdoing. Wrongful interference with some of our most important rights would then be necessary to realize the basic goals of human existence.

But an important aspect of what it takes to exercise our autonomous agency is that we have the capacity to pursue the plans of life we have chosen for ourselves without becoming the victims of wrongdoing. Indeed, a crucial aspect of shaping our life as we wish consists in determining whether others wrong us by φ-ing, even when we do not have an independent interest in whether they φ. Sometimes the only thing that is in our interest is being able to change our normative relationship with them with respect to φ. For example, I might have an interest in consenting to you using my car as a way of fostering our friendship and expressing my special concern for you, even if I don’t have an interest in you using my car. The value of fostering our friendship is realized simply by giving you the permission and is not affected in any way by whether you use the car or not.

This important insight is explored in David Owen’s work (Owens 2011, 2012; see also Tadros 2016: 216–19), whose central claim is that consent serves our ‘normative interests’ in controlling the moral status of certain interactions (i.e., in controlling whether such interactions wrong us or not). But whereas Owens is certainly right that one important function of consent is to serve such normative interests, he is mistaken in suggesting that this is the only function of consent. In addition to serving our normative interests, the power to consent also serves our interest in getting people to behave in certain ways,⁹ as my earlier remarks suggest. (A fact that will play a central role in my account of the ontology of consent). This is because both kinds of interests have an important role to play in enabling us to pursue the plans of life we have chosen for ourselves.

Now, existing autonomy-based justifications of consent tend to focus on how the power to consent protects the autonomy of the consenter in the way described thus far (Hurd 1996: 123–24; Shiffrin 2008: 501–02; Ferzan 2016: 405). But consent is also meant to serve the autonomy of the consentee by enabling them to choose how to act while having a fair opportunity to avoid becoming a wrongdoer. While the consenter has a powerful interest in controlling the duties others owe them, so as to avoid becoming the victim of wrongdoing, the consentee has a powerful interest in knowing which duties they owe the consenter, so as to avoid becoming a wrongdoer.

If you have good reasons to believe that I have consented to φ, when in fact I haven’t, you are constantly at risk of becoming a wrongdoer (Bolinger 2019; Pallikkathayil 2020). By acting in ways that you reasonably believe to be permissible, you might end up violating my consent. One way to neutralize this risk is, of course, to never engage in activities that require consent, but that would make it much harder for you to have a good life, since it would rule out some of the activities we cherish the most—for example, those involving physical contact. It would also make it harder for you to realize those plans that rely on others consenting to you acting in certain ways—for example, the plan of saving for your retirement with the money I consent to give you in exchange for your pizza, or the plan of having special relationships whose flourishing can be favored by the power to allow conducts that would otherwise wrong you.

Thus, the power to consent is not best conceived, as traditional autonomy-based accounts suggest, as protecting the capacity of the consenter to exercise their autonomous agency, but rather as facilitating what we might call ‘interpersonal exercises of autonomy’. Consent is a tool that enables both consenters and consentees to pursue the projects they have chosen for themselves while reducing the risks of being involved in acts of wrongdoing, either as victims or as perpetrators.¹⁰

Consent and Autonomy

Suppose this autonomy-based justification of consent is plausible. What are the implications for how we should think about the ontological question? It looks as if we have reasons to abandon at least one version of the behavioral view—the one which requires uptake for consent, where this means that the message communicated by the consenter must be understood by the consentee.

If we value having the power to consent because we value having control over the consent-sensitive duties that others owe us, we have reason to abandon this view because it would be much harder to control such duties if uptake was required for consent (Tadros 2016: 207). Any attempt to exercise such control would be hostage to fortune because whenever our attempts to communicate turn out to be unsuccessful, we would be unable to change our normative relationship with the consentee as we wish. Maybe you are distracted when I tell you that you can borrow my bike; maybe the email where I give you permission is caught in your spam filter. Vice versa, if all it takes to give valid consent is that we form certain mental states, then we have full control over the consent-sensitive duties others owe us because we can determine what those duties are simply by forming the relevant mental states.

But perhaps the behavioral view can be rescued if it is reformulated as requiring attempted, as opposed to successful, communication. Attempting to communicate consent does fall within our control, and thus we’re not left hostage to fortune if this is what it takes to give morally valid consent. Should we then adopt this weaker version of the behavioral view?

Here’s a powerful argument offered by Victor Tadros to this effect (2016, ch. 3).¹¹ We have seen that when I consent to you φ-ing, I release you from a duty you previously had not to φ. Now, the primary function of duties is to constrain our practical reasoning. If you have a duty not to φ, that means that in deliberating about how to act, φ-ing is in some important sense not an option. It would be inappropriate for you to see yourself as free to φ, even if there are respects in which φ-ing would be valuable. (For example, it would be inappropriate for a doctor to see herself as free to let a patient die and transplant his organs in five patients, even if it would be valuable to save five lives.)

But if the function of duties is to constrain our practical reasoning, when I consent to you φ-ing, I am executing an intention to affect your practical reasoning. With my consent, I release you from a constraint that you would otherwise have not to φ. This is why, according to Tadros', attempting to communicate is necessary for consent. I could not execute the intention to affect your practical reasoning without attempting to communicate with you because it would be impossible for me to alter the role that φ-ing has in your practical reasoning without communicating that intention to you. Given that I cannot intend to do something I believe is impossible, I cannot intend to alter the role that φ-ing has in your practical reasoning without attempting to communicate with you:

This is why it seems incoherent of Y to claim to consent to X φ-ing, but intend to keep this a secret from X—she can desire that X φ-s without attempting to communicate this to X, but she cannot consent to X φ-ing without attempting to communicate this to X. (Tadros 2016: 208–09.)

Two replies can be offered to this argument. First, it might well be incoherent for me to want to consent to you φ-ing while intending to keep this a secret from you. This, however, is compatible with the claim that as long as I don’t have this intention, consent is given when I form the mental state associated with waiving my right that you don’t φ before any attempt to communicate the presence of that mental state is made. A further argument is required for the conclusion that some sort of behavior, such as attempting to consent, is necessary for the relevant normative change to occur.

More importantly, we should reject Tadros' claim that when I consent to you φ-ing, I am necessarily executing an intention to affect your practical reasoning. This is not always the case, as the following scenario illustrates.

Suppose that I see that my estranged son is about to take some money from my car without my permission. I cannot attempt to communicate with him, as I am observing the scene from a security camera. However, I don’t want my son to become a thief, nor do I want to be the victim of his wrongdoing. So, I consent to him taking the money by forming the relevant mental state (the mental state of waiving my claim that he refrains from taking the money). When the police catch him, I can honestly say that he was not stealing, because I consented to him taking the money.

Here we need to tread carefully. I am not denying that my son’s conduct is wrong all things considered (assuming he lacks a justification for it), nor am I denying that his conduct wrongs me, insofar as trying to take someone’s money without seeking their consent is disrespectful. Having the right to consent to φ only gives us the moral power to make it the case that we are not wronged by φ in the way we would have been wronged had we not consented. It gives us neither the power to make it the case that φ is not wrong all things considered, nor the power to make it the case that φ does not wrong us in other ways. (Our power only ranges over the consent-sensitive duties we are owed, not all the duties we are owed) (Owens 2012: 181; Dempsey 2013; Schaber 2019). My claim here is that even if my son’s conduct is wrong all things considered, and even if it wrongs me in that it’s disrespectful, it does not involve the distinctive wrong suffered when consent is violated. I did not suffer this pro tanto wrong because I managed to consent to him taking the money.¹²

No doubt some will disagree. If there is a clear case of consent on which everyone’s intuitions will converge, this is definitely not it. The case is unusual and, at least at first sight, different analyses of what happens seem available. Some, for example, might be tempted to interpret it as a case where I forgive my son for acting without my consent, rather than one where I consent to him taking the money.¹³ As we have seen, however, there is only so much work that intuitions can do in supporting a particular account of the ontology of consent. Thus, my strategy here is not to rely on our intuitive responses to the case, but rather to consider how my analysis fits with the autonomy-based justification of consent sketched in the previous section.

My argument goes like this: if we accept that a central component of the justification of consent is that it serves our autonomy interests, we have reason to believe that we have the power to consent in cases like the one I have described. But if we have the power to consent in cases like this, that means that consenting need not involve an attempt to affect the practical reasoning of the consentee. Thus, we should reject Tadros' argument for the weaker version of the behavioral view.

So, start again with the idea that we value being able to control the consent-sensitive duties that others owe us because this increases our capacity to shape our life as we wish. If this is correct, it seems plausible that such power should include our capacity to prevent others from wronging us in the way attempted by my son. For an account that gives us the capacity to control other people’s duties in cases like this serves our interest in being able to shape our life as we wish better than an account that doesn’t. The latter would leave me at the mercy of my son’s intention to wrong me by violating my consent, whereas the former gives me the choice between being wronged in this way or not. If my preference is to be wronged, perhaps so that I can take advantage of whatever remedy morality affords me, I am free not to consent. If my preference is not to be wronged, I am free to consent. Consent gives me the valuable power to frustrate my son’s effort to wrong me by stealing my money, should I wish to do so (though I would still be wronged by his attempt).¹⁴

This is why the notion of forgiveness is not aptly invoked in this case. The case differs, for example, from the scene of Les Misérables where the bishop first lies to the gendarmes, pretending to have donated the silverware to Jean Valjean and then forgives Valjean for stealing them. The reason why the power to consent is so valuable in my example is precisely that it enables me to prevent my son from wronging me (in the relevant way)¹⁵ in the first place, so that forgiving him will not be subsequently necessary. The bishop, by contrast, lacks the valuable opportunity to prevent Valjean from wronging him. All he can do is suffer the wrong and then forgive.¹⁶

If this understanding of the autonomy-based account of consent is correct, we should conclude that the consenter can change the normative status of the conduct of the consentee without necessarily trying to affect his practical reasoning. This, in turn, undermines Tadros' argument for the weaker version of the behavioral view (according to which attempted communication is necessary for consent), lending further support to the mental state view.¹⁷

As we have seen, the mental state view and the weaker version of the behavioral view à la Tadros' are the accounts that best serve our capacity to act as autonomous agents, because they best protect our interest in having control over the consent-sensitive duties others owe us. If we have reasons to abandon the second view, all things equal, we should adopt the first as the most plausible answer to the ontological question.

But are all things equal? We might worry that this argument focuses too much on the autonomy of the consenter, failing to pay enough attention to the autonomy of the consentee. And if I am right that the function of consent is to protect interpersonal exercises of our autonomy, this is a mistake. Wouldn’t the consentee’s autonomy be best protected by an account that requires the consenter to at least attempt to communicate the presence of the mental states associated with consent? Not necessarily. The consentee’s autonomy can be adequately protected by the mental state view, provided that the capacity of the consenter to give and revoke consent by forming the relevant mental state is suitably constrained.

Some constraints are conceptual: Consent is the power to release others from the consent-sensitive duties they owe us. Absent delegation or other forms of authorization, I cannot waive the consent-sensitive duties you owe a third party. (I cannot meaningfully consent to you taking someone else’s car. All I can do is unsuccessfully try if, say, I believe that it is my car.)

Other constraints depend on the best way to conceive the functioning of consent if it is to adequately serve the value that justifies its existence. For example, we cannot give or revoke consent to conduct that happened in the past, for that would make it impossible for consentees to even try and respect our autonomous agency by responding to our consent. If we had such power, the consentee would be, once again, left hostage to fortune, in that the permissibility (or impermissibility) of his conduct could be constantly reversed ex post. For example, in my attempted theft case, my son could have tried to respect my consent-sensitive duties by securing my consent before taking the money. But how could he do that with respect to conduct that happened in the past if I can give or revoke consent at will with respect to the money that he has already taken?¹⁸

For the same reason, we lack the moral power to give or revoke consent when we intentionally mislead others about our consent-related mental states. For example, I lack the power to revoke my consent to you taking my car if I intentionally give you reasons to believe that you are permitted to take my car, despite no longer having the relevant mental state. Having such power would mean having the ability to intentionally turn consentees into wrongdoers through the exercise of our normative powers simply by lying about our mental states. This would be incompatible with respecting the autonomy of the consentees, since they would lack a fair opportunity not to become wrongdoers. Thus, whenever consenters intentionally mislead someone into believing that a consent-related mental state is present, when in fact it isn’t, they forfeit whatever claim their consent is meant to protect (Ferzan 2017). If I intentionally give you reasons to believe that you are permitted to take my car, while lacking the relevant mental state, I forfeit my claim that you don’t take the car.¹⁹

With these (and perhaps other) constraints in place,²⁰ the mental state view seems able to adequately protect the autonomy of the consentee. Of course, it cannot guarantee that well-intentioned consentees will always be able to correctly grasp and respond to the relevant mental states of the consenters. But neither can the behavioral view, since attempts to consent can easily fail, as we have seen, when the world fails to cooperate. People can be distracted when attempts to consent are made, and spam filters do occasionally get in the way.

What’s Missing?

Go back to the attempted theft case. Suppose you agree that it is valuable for me to have the power to consent to my son taking the money. Even so, something seems missing from the picture presented so far. What is it? The problem can be framed in terms of direction of fit. What happens in the attempted theft scenario is that I change my will so as to prevent my son’s behavior from wronging me. But this is not how things should ideally go. Ideally, it should be the other way around: We normally hope to use consent as a way of guiding other people’s behavior so that it does not wrong us.

This feature of consent is surprisingly neglected in the most recent philosophical literature, which largely focuses on the idea that the point of consent is to affect the normative status of the conduct being consented to. Indeed, some deny altogether that consent has a guiding function. David Owens, for example, contrasts the significance of choice, whose function is to control what is done to us, with the significance of consent, whose function is to protect our interest in controlling the normative status of what is done to us (Owens 2011: 407–08; Owens 2012: 172–76). He acknowledges that the two typically go hand in hand. (Though, strictly speaking, consenting to dental treatment does not entail choosing to receive it; normally by consenting to it we signal that we intend to receive it.) But he stresses that the functions are distinct. Consent is not a tool for guiding how others behave toward us but rather a tool for controlling whether what is done to us wrongs us. (See also Gardner 2018: 57.)

My attempted theft scenario casts doubt on this claim. The reason something seems amiss in that case is that the point of consent is not simply to ensure that a given behavior does not wrong us but that it doesn’t wrong us in the right way. Ideally, we want our consent to ensure that we are not wronged because our consent guides how others behave toward us, and not because we adjust our will to what is going to happen to us anyway.

True, there might be cases where I consent to you φ-ing while being indifferent as to whether you φ, and even cases where I consent while having a positive interest that you do not φ (Owens 2012; Tadros 2016). But this is not to say that my consent doesn’t have a guiding function. My consent guides your behavior insofar as it directs you to not φ unless my permission to you φ-ing has been secured. Thus, at least when things go well, consent enables us to control whether what is done to us wrongs precisely by controlling what is done to us.

This fits with the autonomy-based justification of consent. If the point of consent is to enhance our capacity to shape our life as we wish, we should care not only that those who happen to take our money have our permission but also that they would not take the money if they hadn’t our permission. A view that gives people permission to take our money irrespective of whether they have taken into consideration our decision to let them do so (by forming the relevant mental state) is one that does a worse job at protecting our autonomy than a view that makes the permission conditional on their tracking our decision. Not only because without this constraint it’s more likely that there will be cases where others will take our money even if we do not intend to give them permission, but also because in those cases where the intention has been formed, it will be respected in one of two ways: either through luck or because we have adjusted our will to match what is going to happen to us anyway (as in the attempted theft case). And either way, our capacity to exercise our self-determining agency is poorly served. For we cannot reliably use the power of consent to shape our life as we wish if it is only through luck that others respect our intention to give them permission, let alone if our choice to give them permission merely tracks what they have decided to do.

What does it take for the consentee to be guided by the exercise of our power to consent? First, the consenter must intend to change the normative relationship with the consentee by forming the relevant mental state (i.e., the mental state of waiving or refusing to waive the claims correlating to the consent-sensitive duties). Second, the consentee’s behavior must be sensitive to the presence of that mental state. For the normative relationship between the consenter and the consentee to be changed in the right way, the latter must acknowledge the intention of the former and do his best to respond to the relevant mental state. (More on this below). This kind of sensitivity to the mental state of the consenter is what is missing from my attempted theft case.

I believe that something like this idea is what pushes some toward the behavioral view. The thought that communication, or an attempt to communicate, must be present for consent to be given is prompted, at least in some cases, by the belief that only by incorporating some form of communication in our view can we vindicate the role that this kind of sensitivity ought to play within a theory of consent. Thus, Dougherty writes that ‘[c]ommunication enters the picture as the only reliable way to guarantee common belief across the range of cases that will regularly arise in our lives’ (Dougherty 2015: 240).²¹

And Healey writes,

[I]n order for consent to play [the role of allowing X and Y to relate to each other as autonomous agents], it must be possible for X and Y to have mutual knowledge regarding whether consent has been given or revoked. That is, both X and Y must, in principle, have access to the information that determines whether or not consent has been given. (Healey 2015: 358)

This, however, is mistaken. The tendency to think in these terms is prompted by an assumption that drives much of the contemporary debate on consent, namely that in answering the ontological question, we should focus on what the consenter needs to do: Does the consenter need to form a mental state? Do they need to attempt to communicate it? Do they need to successfully communicate it? I want to suggest that this is the wrong way to frame the question of the ontology of consent. To adequately address this question, we also need to focus on what the consentee does.

Consent, at least in the ideal case, requires the consentee to do something: the consentee has a responsibility to adequately track the mental state of the consenter and do their best to respond to it. Of course, communication greatly facilitates this process, but it is not necessary for it. People in close relationships, such as partners, siblings, or friends, might be good at reading each other’s behavior and realize when they have formed the mental state associated with consent (to sex, to borrowing something the other person is not using, etc.), even before any attempt to communicate is made. When they do, we have our ideal case of consent, despite lack of any communication.

In other words, once we pay attention to the responsibility of the consentee to adequately track the mental state of the consenter, we see that the sort of ‘common belief’ or ‘common knowledge’ that defenders of the behavioral view appeal to can be secured without communication.²² All that is required for the relevant common knowledge to be in place is that the consentee is adequately tracking the mental state of the consenter and that the consenter is aware of that. This condition can be fulfilled before any attempt to communicate is made.²³

This claim might strike some as unrealistic, since the risk of mistakes in tracking the mental states of the consenter might be thought to be too high, unless some communicative behavior is enacted. This is an empirical question, and I don’t have an argument against this sort of skepticism. But luckily, we don’t need one. Any attempt to challenge the mental state account by raising doubts about how effectively the mental states of consenters can be tracked, absent an attempt to communicate, reinforces the case for the mental state view instead of weakening it. For any such argument presupposes that the role of communication is to provide epistemic access to those mental states, and this is precisely what defenders of the mental state view believe.

To see this point better, suppose technology were to provide us with a device capable of reading each other’s desires. The device would reliably tell you when I desire to take a nap, eat pizza, or be kissed. Such a device would not help us significantly in our practices of consent. Of course, you could rely on its verdicts in planning how to act: if you know that I desire to let you use my car, you might wait to ask someone else. If you know that I desire to be kissed by you, you can leave the room or conveniently place yourself next to me. Either way, you still need to wait for the magic of consent to happen before taking the car or kissing me. Even if we imagine that the device never fails, you would have to wait for I might always decide not to authorize the conduct I desire.

But now suppose that a different device were available, one that tells you when I form the mental state associated with waiving my claim that you don’t take my car or my claim that you don’t kiss me. Suppose that we are all connected to such devices, whether we want to be or not, so that our being connected to them cannot be plausibly understood as an attempt to communicate our mental states to anyone. Assuming that you are dutifully tracking the presence of that mental state, it is not clear why it would be wrong for you to take my car or to kiss me once the device lights up, before I even try to communicate the presence of the mental state in question.²⁴ The point of communication here seems to be limited to ensuring that you have the correct information about how I have exercised my will. (Perhaps I want to make sure you’ve read the device’s signal correctly.) What produces the authorization is how I have exercised my will.

Once again, this is not just a matter of intuition. It is obviously valuable to me that I am in control of deciding whether a certain behavior wrongs me or not, but it is not clear why I should value being the one who communicates the result of this decision, apart from the fact that I am generally best placed to reliably do so.²⁵

Now, of course there is a sense in which when the device lights up, it ‘communicates’ the presence of my mental state, in the same way in which my sweating communicates to you that I’m hot or the sun’s position communicates to you that it’s midday. But this is not the notion of communication that defenders of the behavioral view operate with. Communication involves listener and speaker acting cooperatively and mutually intending their inputs to be taken in a certain way by the other party (Grice 1989). This is not what happens when the device lights up. The device lighting up is merely a sign of the fact that I have formed the relevant mental state, in the same way in which my sweating is a sign of the fact that I am hot. Once again, what we have here is merely evidence of the fact that the mental state is present.²⁶

This, however, raises the question of whether the presence of a sign of this sort is necessary for the ideal case of consent. In other words, we need to choose between two accounts:

1. Ideally, consent requires

   1. that the consenter has formed the relevant mental state, and

   2. that the consentee is tracking it.

2. Ideally, consent requires

   1. that the consenter has formed the relevant mental state,

   2. that there are signs of the fact that they had done so, and

   3. that the consentee is tracking the relevant mental state by responding to the signs associated with it.²⁷

What is at stake in this choice? Option 2 restricts the way in which the consentee ought to track the relevant mental state to doing so via identifying a manifestation of the mental state (i.e., a sign of the fact that the mental state has been formed). Option 1, on the other hand, allows for the possibility that the consentee might track the mental state in a more indirect way—say, by inferring its presence from the fact that it is reasonable for the consenter to form it, or perhaps from the fact that they have previously formed it in similar circumstances.

Consider for example, the following case by Mollie Gerver:

Hurricane: Katya’s neighbor is abroad, and there is a hurricane that risks destroying his porch if it is not propped up with special reinforcements. Katya can only reinforce his porch if she enters his property. Though her neighbor hears about the hurricane, he is far away, and cannot tell her to prop up his porch. Given that Katya knows that humans in general care a great deal about the porches of their homes, she can infer his mental state is such that he has decided to waive his right against her entering his property, or at least decided he would like to waive a right against her entering his property. She can also infer that he can infer that she infers this, given that he has access to news about the hurricane, and given that he knows Katya knows that humans generally care about saving their porches from hurricanes. (Gerver 2020: pp. 39–40)

In this scenario, ideal consent could be secured under 1 but not under 2.

This is, I think, a reason to adopt 2. For 1 does not make the neighbor’s conduct sensitive to Katya’s decision to permit the conduct in question.²⁸ If, on a whim, Katya decides not to permit the conduct, her neighbor is justified in drawing the exact same inference he would be justified in drawing had Katya decided to permit it. Since his inference is based on what he expects Katya’s mental state to be, given his general knowledge of her, any behavior based on such inference will not be sensitive to how Katya has in fact exercised her will. But how consenters have exercised their will is precisely what consensual behavior should ideally respond to, even when their decisions are unpredictable or out of character. The requirement to respect other people’s consent is the requirement to respect their decisions to grant us certain permissions, regardless of whether they do so in a way that conforms to our expectations.²⁹

2 aptly captures that. In requiring that the consentee tracks the relevant mental state in the consenter by responding to the sign(s) associated with it, the account makes the consentee’s behavior sensitive to how the consenter has exercised their will in the case at hand, whether or not they have done so in a predictable way. So, consider again the world in which we have consent-detecting devices and assume that given my interests, desires, and beliefs, it makes sense for me to permit you to take my car and for you to believe that.³⁰ Here, if for some reason I decide not to form the relevant mental state, the device will not light up and thus, under 2, you will not be released from your consent-sensitive duties. It is by making your permission to take the car conditional on your tracking my mental state by responding to this sign that we ensure that your behavior is sensitive to how I have exercised my will.

Now, this is the intuitively correct result. But why is that? Why should we care that the behavior of the consentee is sensitive to the mental state of the consenter in this way? Why not choose between 1 and 2 simply by looking at which view is more accurate overall in tracking the relevant mental states? Suppose that relying on the sort inference allowed by 1 would, in the long term, prove to be a more accurate strategy in tracking consent-related mental states than having to stick with the restrictions imposed by 2. Why not go for 1?

The answer is that the value of autonomy, which consent is meant to protect, is one that we should primarily respect rather than promote (Pettit 1989). Leaving aside cases of justified paternalism, disregarding someone’s autonomous decisions to give or withhold consent is impermissible, even if doing so would enable us to do a better job in conforming with their autonomous decisions in the long term. This is why sensitivity to the existence of consent-related mental states is more important than overall accuracy in tracking such states. We want the conduct of the consentee to be sensitive to the mental states the consenter has in fact formed in the case at hand, even if we concede arguendo that an inference-based method would enable us to do a better job in protecting the autonomy interests of the consenter in the long term.³¹

This is not to say that Katya should not reinforce her neighbor’s porch, of course. Whenever the conditions for giving consent are absent, it might well be permissible for us to rely on reasonable expectations of the kind of behavior others would be willing to authorize, if they had a chance. But it would be a mistake to treat these as cases of consensual behavior.

Ideal and Defective Cases of Consent

I have argued that, ideally, consent requires that the consentee is sensitive to the presence of the consenter’s relevant mental state, in the sense that the consentee must track such mental states by responding to the signs associated with it.³² If I am right, we should conclude that although cases like my attempted theft scenario are indeed cases of consent (for the reasons discussed in the section on “Consent and Autonomy”), they are cases of defective consent.

The distinction between defective and ideal (or full-blooded) consent might sound odd. After all, either I have waived my right that my son does not take the money or I haven’t. If I haven’t waived my right, the right is still present, and thus my son acts impermissibly. If I have waived my right, there is no right violation, and thus my son is permitted to take the money. Either way, what is the point of saying that ‘defective’ consent has been given?³³ The answer is that in cases of defective consent, a moral remainder persists. An apology, or even compensation, might be owed to the consenter, which is not owed in cases of ideal consent (where the consentee is sensitive to presence of the relevant mental states). The consenter, however, has no claim to it.

This is not an exclusive feature of consent. It is true of other normative powers that, like consent, can be exercised defectively. The most famous example is probably forgiveness.³⁴ Many think that the central case of forgiveness is one where forgiveness is granted once the wrongdoer has acknowledged their wrongdoing, repented, and apologized (Murphy 2003; Griswold 2007; Hieronymi 2001). However, there seem to be genuine cases of forgiveness of wrongdoers who have not done, or cannot do, any of these things. Having the power to forgive when these ideal conditions are not fulfilled can be valuable for a number of reasons (Pettigrove 2004; Allais 2013; Garrard and McNaughton 2003). For example, we might value being able to forgive someone who’s deceased, or someone who stubbornly refuses to acknowledge their responsibility, because we love them, because doing so will help us move on with our life, or because we want to overcome feelings of resentment and other negative reactive attitudes.

These two views of forgiveness are sometimes presented as alternatives to each other, but this is a mistake. ‘Conditional’ and ‘unconditional’ forgiveness, as they are often called, are related but distinct moral phenomena that coexist in how we normally think about forgiveness (Bennett 2003; Fricker 2018). The former identifies the ideal case of forgiveness, the latter a defective one.

The point of saying that unconditional forgiveness is ‘defective’ here is not to suggest that it is somehow less valuable than conditional forgiveness, but rather to mark the presence of a moral remainder associated with it. Once unconditional forgiveness is granted, there are no further claims that the forgiver has vis-à-vis the wrongdoer. However, this is not to say that there are no moral demands that apply to the wrongdoer. Clearly, the wrongdoer still ought to acknowledge that what they did was wrong, repent, and, if possible, make amends. That their victim no longer has a claim that they do any of this doesn’t mean that they do not have powerful moral reasons, perhaps even a duty (though not a directed one),³⁵ to do it.

I take this judgment to be intuitive, but how can we explain it? We do so, once again, by considering what justifies the existence of the power to forgive. While unconditional forgiveness might in a sense be even more admirable than forgiveness issued in response to apologies or repentance, it is nonetheless regrettable that the reconciliation between the wrongdoer and his victim is not realized through a process in which the former acknowledges their wrong and takes appropriate steps to make amends for it. If we think that within a plausible account of the justification of forgiveness this valuable process is what should ideally lead to repairing the relationship between the wrongdoer and the victim, the fact that such process hasn’t occurred explains why there is a moral remainder associated with unconditional forgiveness.

Another example of a normative power that can be exercised defectively, and which, unlike forgiveness, has not received any attention so far in its defective form, is commanding. Imagine a battalion that is so enraged at the sight of the enemy on the battlefield that they are about to charge without waiting for the general’s command. The general realizes this and hastens to order the attack so as not to lose face.³⁶ How should we think of such a case? I take it that the general does successfully exercise their normative power. The command is clearly binding and as such it does change the normative relationship between the general and the soldiers in the way orders normally do. (For example, I take it that if one of the soldiers refused to charge, they would be liable to whatever sanction is attached to disobedience.) But, once again, something seems missing. As in the attempted theft case, there is the wrong direction of fit between the exercise of the normative power and the behavior of those over whom the power is exercised: instead of guiding the behavior of the soldiers, the general’s command is issued only to prevent the wrong of disobedience. This is not how ideally things should go.

Explaining why would require offering an account of what justifies the power of generals to issue commands, a task which I cannot undertake here. But presumably any credible justification will have to appeal, at least in part, to the idea that in guiding the soldiers’ behavior, commands facilitate coordination on the battlefield. If that’s true, what might plausibly explain the presence of a moral remainder in cases of defective commands is that our interest in facilitating coordination on the battlefield is best served when soldiers are sensitive to their superiors’ attempts to exercise such powers, rather than when the power is exercised in order to prevent the wrong of disobedience.

It might be valuable, at least under certain conditions, that the general has the capacity to issue a binding order even when this sensitivity is absent. If they didn’t, soldiers could effectively deprive the general of the normative power to command simply by ignoring the general’s attempts to do so. But this is not to say that nothing is amiss when soldiers act without trying to follow the command. This, in turn, explains the presence of residual moral demands that apply to them.

Notice how this explanation mirrors the one offered earlier to defend the possibility of defective consent. There, we saw that the value of shaping our life as we wish is best served when consentees are sensitive to our attempts to exercise our power to consent by forming the relevant mental state, rather than when we exercise the power in order to prevent the wrong of nonconsensual treatment. This is what explains why cases of consent that lack such sensitivity leave a moral remainder behind. Even if there are good reasons to give the consenter the power to consent in those cases, the behavior of the consentee is such that some of the interests underlying the justification of this power are not being served as well as they should. Hence, the residual moral demands that apply to the consentee.

We should not think, of course, that there are no differences between how various normative powers can be exercised defectively. For example, while the problem with defective exercises of the power to consent and command is that they display the wrong direction of fit, defective forgiveness does not seem to have this feature. This is unsurprising since forgiveness normally does not have the sort of guiding function that consent and commanding have. On the other hand, defective forgiveness and defective consent are similar with respect to the type of moral remainder that they generate.

We have seen that in the case of unconditional forgiveness, this moral remainder takes the form of moral reasons (and possibly a nondirected duty) to apologize and make amends, even if the forgiver has no residual claims vis-à-vis the wrongdoer. Defective consent operates in the same way. Once defective consent is given, the consenter waives any claims correlating to the relevant consent-sensitive duties of the consentee. And yet, there are residual moral demands that apply to the consentee. For example, we might think that although I no longer have a claim that my son does not take my money, once I defectively consent to him doing so, he might still have powerful moral reasons, perhaps even a non-directed duty, to apologize, make amends, and possibly even return the money to me. (Although I gave him permission to take it, the morally right thing to do might be not to take advantage of this permission.)³⁷

This is where my account differs from traditional formulations of the mental state view. These views consider the behavior of the consentee only in order to determine whether the consentee is culpable, not in order to determine whether the relevant conduct is consensual. (Their answer to the ontological question is determined exclusively by the mental states of the consenter.) For this reason, in dealing with cases like my attempted theft scenario, they cannot say that there is any moral remainder associated with the consent received by my son in the attempted theft case. Although my son acts culpably in taking the money, for the traditional mental state view, my consent is as good as the one he would have received if he had asked for permission (Alexander 1996; Ferzan 2016; Alexander, Hurd, and Westen 2016; see also Husak 2006).³⁸

Now, perhaps this is not a problem if it turns out that the same moral remainder my view associates with defective consent can be generated by the fact that my son acts culpably. If so, traditional mental state accounts can maintain that my son’s conduct is perfectly permissible while also saying that he ought to apologize and make amends for his culpable behavior, i.e., for attempting to act without my permission (despite failing to do so). But this analysis seems to me less perspicuous than one capable of flagging the problematic nature of the permission received by my son.

More importantly, the analysis runs into trouble if we imagine that my son has a justification, or even just an excuse, for taking the money. Suppose he needs the money to prevent some impending moral tragedy. In this case, we could not say that my son acts culpably. (He might even have a duty to take the money.) Still, it would make perfect sense to say that, insofar as he is not tracking my intention to give him permission, the consent he receives is defective, and thus leaves behind a moral remainder, which might take the form of a duty to apologize, compensate, or at least explain his behavior. Traditional mental state accounts, however, cannot explain this, since my son in this case is, by hypothesis, not culpable.

One might be tempted to reply that these accounts have at least the advantage of being more parsimonious in positing only one type of consent, while allowing the culpability of the consentee to explain the presence of the moral remainder. But if I am right that other normative powers, such as forgiveness and commanding, are also subject to ideal and defective exercises, the view I am offering in fact adds parsimony to our broader theory of normative powers. The view uses a simple distinction to explain a feature that all these powers share.

More would need to be said to explore the similarities and dissimilarities between how the powers to consent, forgive, and command can be defectively exercised. This, however, is a task for another day. My aim here was to show that the distinction between defective and ideal exercises of a normative power is plausible and is not limited to the case of consent. In elucidating the importance of this distinction, I’ve made three suggestions.

First, the distinguishing features of defective exercises of these normative powers is that they leave behind a moral remainder, though one that, at least in the case of forgiveness and consent, does not correlate to any claim of those exercising the power.

Second, in order to explain what counts as an ideal or a defective exercise of a normative power, we need to consider what justifies the existence of the power in the first place. For example, it is the fact that the interests underlying the justification of consent or command are best served when those subject to them are sensitive to how these powers are exercised that explains why any exercise of such powers that is not accompanied by this sort of sensitivity counts as defective.³⁹

Third, that defective exercises of these powers can be morally transformative is explained by the fact that they serve an important moral function: they provide us with valuable control over the normative situation when others are unwilling to cooperate (i.e., when their behavior precludes the possibility of an ideal exercise of the relevant power).⁴⁰ Specifically, we have seen that the reason to believe that defective consent is morally transformative is that it is valuable to us to retain some control over whether our consent-sensitive duties are violated even when the consentee’s behavior precludes the possibility of giving ideal consent.

Conclusion

I have suggested that in answering the ontological as well as the justificatory question, we need to pay more attention to the role of the consentee(s). If we do that, we’ll see that two types of consent are worth distinguishing, each grounded in certain autonomy interests we have.

In developing an account of these two types of consent, we’ve ended up with a rather complex picture, one that relies on a novel idea (the thought that normative powers can be exercised defectively) and a number of subtle distinctions (such as the one between tracking the consenter’s mental states in a way that is sensitive to their presence or not). But for all these philosophical niceties, the motivation behind the account is simple and can be easily summed up: we have an interest in being able to control the normative status of consent-related interactions by guiding others to act in conformity with the conduct we have decided to authorize. But we also have an interest in being able to control the normative status of the same interactions when others cannot, or will not, be guided by what we have decided to authorize.

What follows from this is that although it is enough that the consenter forms certain mental states for consent to be given, it is only when the consentee is sensitive to the presence of such mental states that we have full-blooded consent. Failing that, what we have is a defective case of consent. Having the power to give defective consent is valuable, but it would be a mistake to think that this is all there is to consent. Defective consent differs from ideal consent in that it leaves behind a moral remainder, which, depending on the circumstances, might take the form of a duty to apologize, make amends, compensate, or perhaps simply explain one’s conduct.

Notice that this view is not simply the combination of the behavioral view and the traditional mental state view. For almost any debate between two views of X, one can adopt a deflationary response that consists in distinguishing two types of X, each responding to the motivations of the two views and then combining the two. Even when successful, however, this sort of strategy is of limited philosophical interest.

This is not what I am offering. The notion of ideal consent I have defended is categorically different from the one employed by behavioral views, because at no point does it attribute to communication anything more than an epistemic function.⁴¹ At the same time, my view differs from traditional mental state accounts because it sets the bar for full-blooded consent higher than they do. Instead of depending exclusively on facts internal to us (i.e., whether we have formed certain mental states), the existence of full-blooded consent also depends on the world being a certain way (i.e., whether the intended recipients of our consent are connected to those mental states in the right way by tracking the signs that are evidence of those states).⁴² True, like traditional mental state accounts, my view allows the formation of private mental states to realize a normative change in cases of defective consent, but the change in question is different from the one realized by full-blooded consent, in that it leaves behind a moral remainder (in the same way in which defective cases of forgiveness and commands do).

Traditional mental state accounts have tried to explain the presence of this moral remainder by appealing to the culpability of those who act while mistakenly believing that they lack consent, but we have seen that this approach struggles with cases in which the agents are justified or excused for so acting. Moreover, insofar as the ideal/defective distinction applies to other normative powers, such as forgiveness and commanding, a theory of consent that also relies on it fits a broader picture of normative powers that seems independently plausible. If I am right about this, the conceptual framework developed in this paper advances not only our understanding of consent but also our understanding of this broader picture.