Flickers of Freedom

Paternalism and Consent

This is going to be my last substantive post for the week. Huge thanks to Thomas and the broader Flickers community for the opportunity, it’s been a blast!

I started this week’s blogging with a problem for certain hierarchical theories of autonomy. If we are only autonomous when our desires/actions align with some higher order attitude, then it looks like paternalism will be less prevalent, and consent much harder to come by, than typically thought. Akrasia is a paradigm example of the kind of mismatch between higher and lower order attitudes that hierarchical theories will treat as non-autonomous. Yet we typically think it’s paternalistic to intervene in akratic actions, and that (absent other considerations) akratically given consent is still valid.

The theory I’ve been sketching over the course of the week can avoid this problem. Decisions, I claim, have authority of their own, even when they fail to align with other elements of the agent’s psychology. This provides a simple and straightforward explanation of the connection between autonomy, paternalism, and consent. It does, however, raise difficulties of its own. It’s those difficulties I’ll be focusing on today.

In my last post I talked about the problem of manipulation. To recap, the problem is that if the agent’s values, beliefs, decisions etc are merely the product of manipulation, the agent shouldn’t be considered autonomous with respect to them. My solution was to note that such agents would have low autonomy because many of their values, beliefs, decisions etc would clash with other of their core attitudes: for instance, a policy of only treating as reasons considerations that are true. What’s distinctive about this solution is that it doesn’t deny the authority of the values, decisions etc that are the product of manipulation; it just suggests that their presence lowers the agent’s autonomy because they clash with other elements. It’s this continued authority that needs to be accounted for when thinking about consent and paternalism.

Both consent and paternalism are responsive to the authority of the agent’s decision: consent is only valid if the agent has the authority to grant it through her decision; more straightforwardly, paternalism is wrong because the agent’s decision to act has authority. When an agent is otherwise competent, informed, and free from coercion, we typically take decisions to consent and to act to be authoritative. We do this even when the agent is deciding akratically, suggesting that those decisions don’t need to be fully autonomous in order to be authoritative. What, though, if the agent’s decision to consent/act is the product of manipulation?

It might be tempting to say that, for whatever reason, the agent’s decision cannot have authority if it is the product of manipulation. That’s the temptation I’m hoping to resist. That doesn’t mean, though, that I think manipulated consent is fully valid; or that intervening in manipulated action is fully paternalistic.  The reason is that, while the agent’s decision has authority, so too do the rest of her values, policies, etc.

Most of us, I suspect, have some kind of standards concerning when our decisions should be granted authority, even if not fully articulated. For instance, I don’t think my drunken decision to send an angry email should be respected; if someone’s in a position to stop me, they should. I do, however, think my drunken decision to eat a kebab on the way home should be respected; how dare anyone presume to tell me what to eat! I’ll call such standards policies (even though they don’t neatly fit with what Bratman calls self-governing policies). They are policies concerning the authority I take my decision to have, under certain conditions.

In order to head off a potential misunderstanding at the outset: I am not suggesting that these policies trump the decisions themselves; that because I have a policy denying authority to certain decision, those decisions then lack authority. Instead, I am suggesting that these policies have authority alongside the decisions themselves. If we then make one of these decisions, we have placed ourselves (and others) in a bind – there is nothing we can do, or that they can do, that will fully preserve and respect our autonomy. Whatever happens, some element with authority has been overridden. 

What should we say, then, about consent and paternalism under conditions of manipulation? Exactly the same thing, I think, that we should say about consent and paternalism when the agent is akratic. In each case, the agent’s decision retains authority – to prevent her from following through would be to frustrate/disrespect her autonomy. However, the agent’s policies also have authority – if she has a policy denying the authority of an akratic or manipulated decision (in the relevant circumstances) then granting her consent, or failing to intervene in her action, will also frustrate/disrespect her autonomy.

When we’re deciding whether to intervene, or whether consent has been validly given, I don’t think it’s helpful to focus on whether or not the agent’s decision is autonomous. I think it’s more fruitful (and more accurate) to ask how we can respond so as to best preserve/respect the agent’s autonomy, acknowledging all the while that whatever we do will involve some costs: some aspect of the agent’s autonomy will be frustrated.

Why is this fruitful? A key reason, I think, is that it brings to our attention an analog of what Judith Jarvis Thomson calls ‘moral remainder’. If we assume that the manipulated agent’s decision has no authority, then there is no autonomy-related harm done to her if we override that decision, or refuse to grant her consent. On my approach, by contrast, we are forced to acknowledge the ways we are – albeit justifiably – overriding something that actually has authority. In thinking about how we engage with agents at the margins, such as the mentally ill, I think this sensitivity has much to recommend it.

The Problem of Manipulation

Today I’ll be talking about the problem of manipulation, which poses a critical challenge for any theory of autonomy to address. The problem of manipulation runs, roughly, as follows: if autonomy is simply a matter of getting one’s first and second-order attitudes to line up in the right way, then it looks like agents can be manipulated and still be autonomous. All the would-be manipulator would need to do is alter the agent’s second-order attitudes so that they support whatever first-order attitudes the manipulator has instilled in the agent.

Philosophers love to come up with examples of nefarious neuroscientist and such to play this role, but real-life examples are not difficult to find. At the extreme, there are cults which engage in indoctrinations that border on brainwashing; more relevantly, some worry that everyday gender socialization has this shape, too.

The problem of manipulation is a big problem, and I’m certainly not going to solve it here today. What I hope to do, though, is offer just a little bit of pushback on the claim that content-neutral theories of autonomy (of which mine is one) have no resources to address this problem.

The problem of manipulation looms largest, I think, for theories such as Frankfurt’s that require coherence only between a first-order desire and a second-order volition. Even in his later work, where the requirement of satisfaction is introduced, the goal seems to be that the agent should experience no conflict regarding her identification with her first-order desire. Such lack of conflict is possible, however, in cases of indoctrination or oppressive socialization – indeed, the more effective the indoctrination/socialization, we might worry, the more satisfied the agent will be with her first-order desires.

What happens, though, if we broaden the scope of autonomy beyond higher-order volitions, to include values, beliefs, self-governing policies, and so on? My hunch is that in the vast majority of indoctrination/socialization cases, there will be significant conflict within the agent’s practical identity, and hence – at least on my account – we have resources to say her autonomy has been damaged. This is true even though she may be completely unaware of this conflict, and hence be satisfied in Frankfurt’s sense.

Take an extreme case, of an agent raised in a cult to believe the leader is a God-like figure who must be obeyed, and the government is just a conspiratorial plot. Such an agent would strike many (me included) as having very low autonomy. She may well, though, meet Frankfurt’s conditions – for any given first-order desire, no matter how outlandish, she may well be satisfied with the corresponding second-order volition. She will most likely not, however, meet my conditions. This is because the indoctrinated agent will most likely have a self-governing policy of only treating as reasons those considerations that are true – indeed, the cult itself will have to instill such a policy in the agent to motivate her allegiance to the cult, and her suspicion of the government. This policy will be violated in the vast majority of her cult-related activities. Similarly, the indoctrinated agent will most likely value trustworthiness – again, the cult itself has reason to instill such a value in agents whose trust they hope to exploit. And again, this value will conflict with the agent’s actual practices of trust. The broader point is that long-term, systematic, indoctrination typically piggy-backs on values, beliefs, and self-governing policies that have broader scope than the manipulator’s immediate goals, and which conflict with other of the induced values, beliefs, and self-governing policies. There is thus a fairly direct route from indoctrination to damaged autonomy.

In order for a manipulated agent to be fully autonomous, then, the manipulator would have to ensure full coherence across the full range of the agent’s beliefs, values, self-governing policies, etc. In other words, they would have to shape every facet of the agent so as to avoid even highly abstract values such as truth. To lay my cards on the table: I think such thoroughly indoctrinated agents would in fact be relatively autonomous – just as autonomous, perhaps, as regular people. While this might seem highly counter-intuitive, I think we need to be particularly cautious in leaning on intuitions here. Although such a thorough process of manipulation is conceivable, it is also well beyond anything that actually happens, even in extreme cases of indoctrination into cults, so it can be tricky to get our intuitions to focus in on what such agent’s would actually be like. Moreover, I think we also need to be careful not to presuppose some ‘real self’ that lies underneath such manipulations, waiting to be set free.

There are of course alternative ways of trying to get around the problem of manipulation, the most common of which incorporate historical conditions that directly rule out values/desires etc that are the product of manipulation. I’ve argued elsewhere that such alternatives typically run into counterexamples of their own, in that consistently applying their principles would require us to count many normally socialized people as non-autonomous. If that diagnosis is correct, I’d be curious to hear how Flickerers would assess the dialectic: is it better to have a theory that bites the bullet and counts some manipulated agents as autonomous, or to have a theory that requires us to see normally socialized people as non-autonomous?

Addiction

Over the last couple of posts I’ve been sketching an account of autonomy that locates the ‘self’ of self-governance in a range of elements or processes, rather than trying to pinpoint a single authoritative center. Going forward, I’ll try to motivate this approach by seeing how it fares when brought to bear on a series of different applications. Today I’ll focus on addiction; in my next post I’ll talk about oppressive socialization; and I’ll wrap things up with a discussion of paternalism and consent.

Harry Frankfurt famously drew a distinction between what he called the willing and the unwilling addict: while the willing addict wants her desire for the drug to lead her to take the drug, the unwilling addict doesn’t want her desire for the drug to lead her to take the drug. Because the unwilling addict does not have the will she wants, Frankfurt claims, she is unfree.

Frankfurt’s early work fits with the general approach I talked about earlier – it picks out an element or process that is supposed to speak for the agent, and hence ground her autonomy. In Frankfurt’s case, that element is a second-order volition; for Bratman it is a self-governing policy. In either case, what we are doing is checking to see whether the element or process that speaks for the agent authorizes the motivating desire.

While this approach allows us to acknowledge the different degrees of autonomy exhibited by different addicts, I don’t think it goes far enough. Because it only recognizes one source of authority in the agent, it fails to take note of the  range of different ways in which an addict can be unwilling, as well as the possibility that these different kinds of unwillingness might compound.  

Autonomy and Authority

In yesterday’s post I suggested that it’s a mistake to pick out a single element or process of the self as authorized to speak for the agent. Today I’ll say a bit more about the multiple sites of authority I think underpin autonomy. Future posts will try to make good on the claim that understanding autonomy in this more diffuse way provides better resources for thinking about other domains within which autonomy is appealed to, such as moral responsibility and paternalism.

Before heading down that path, it’s worth pausing to reflect on why the search for authority is so important. The heart of the concept of autonomy (at least as I’m using that term) is self-governance. If an agent is to be self-governing, then there must be something analogous to governance taking place – something has to play the role of laws or commands, directing the agent. Moreover, those laws or commands have to in some way emanate from the agent herself, otherwise autonomy collapses into heteronomy. [I’m taking a lot of this for granted - if there are Flickerers who disagree I’d be very keen to hear from you!]  

Since I think it’s a bit grandiose to suppose that human agents do much in the way of giving ourselves explicit commands, I’m looking for something more psychologically basic to play this role. Basically, I’m looking for any element or process within the agent that has authority over the agent – not necessarily in the sense of ‘speaking for the agent’, as Bratman puts it, but in the sense that the agent has some kind of pro tanto obligation to heed the demands made by that element or process. I think the element Bratman picks out – policies about what to treat as reason-giving in practical deliberation – clearly fit this bill. Such policies have authority for an agent, they create pro tanto obligations for the agent to deliberate in accordance with them, and she is less autonomous to the extent that she fails to do so. Where I differ from Bratman – and from many others – is in thinking that a wide array of elements and processes also fit this bill, and that none of these elements or processes trumps the others.

Here’s a (non-exhaustive) list of the kinds of elements or processes that I think also have authority, and hence also have to be cohered with if the agent is to be (fully) autonomous: values; beliefs; practical deliberation; and intentions.

I’ll start with a closer look at the first two, since they share with Bratman’s policies the feature of being part of what Korsgaard calls an agent’s ‘practical identity’, and hence forming part of the agent’s point of view. [A lot of details are being glossed over here – I’d be happy to share the book-length version with anyone who’s interested!]

A Unified Theory of Autonomy?

Hello Flickerers, and thanks for having me on board! Thanks especially to Thomas for extending the invitation. I won’t be starting my substantive posting for another week or so, but I wanted to take the opportunity to introduce myself, and say just a little about what I’m working on.

Autonomy is one of those tricky concepts that is both widely used and deeply contested. It’s taken as central in debates on the nature of paternalism, consent, political legitimacy, weakness of will, and adaptive preferences, to name just a few. Yet it’s not always clear that participants in those debates are using the term in the same way. For instance, it’s common to assume that people suffering from weakness of will are less autonomous (at least with respect to those actions). But it’s almost never assumed that weakness of will is sufficient to undermine consent, even though consent is typically thought to require autonomy. Similarly, a number of theories of autonomy claim that actions motivated by oppressive norms are non- (or less) autonomous. And yet, I think very few people would agree that it is OK to intervene in the actions of someone, just because they are motivated by oppressive norms, even though paternalism is only meant to be wrong insofar as it offends against autonomy.

One way we might respond to this predicament is to declare autonomy a useless concept, too hopelessly muddled to be salvageable. Another way we might respond is to develop distinct accounts of autonomy for distinct purposes (autonomy means something different if we’re talking about weakness of will than it does if we’re talking about consent, and so forth).

My goal over the last few years – hopefully not a fool’s errand! – has been to see if it’s possible to develop a theory of autonomy that can make sense of these seeming inconsistencies, and hence offer some common ground across the range of debates within which autonomy is a key concept. I’ll be sharing the core features of this theory, and showing how they can be brought to bear on the relevant debates, in the weeks to come. I very much look forward to hearing your thoughts!

Passing the Torch!

It's once again the time for the monthly changing of the guards here at Flickers. But first, I wanted to thank Josh Knobe one more time for an illuminating series of posts as the Featured Author in July. He did a splendid job of fanning the philosophical flames during his stint. It was a great way to relaunch and revamp the series and a nice foray into some of the exciting interdisciplinary work being done on people's ordinary intuitions about agency, free will, and the self. Hopefully, others got as much out of Josh's posts as I did.

That said, now that August is here, we have a new Featured Author to look forward to in the coming weeks! Suzy Killmister is an Assistant Professor at the University of Connecticut, with a joint position in the Philosophy Department and the Human Rights Institute.  She is  currently writing a book on personal autonomy, provisionally titled Taking Autonomy's Measure. She is also working on a number of papers on the concept of dignity, and the connection between dignity and human rights. Needless to say, I am very much looking forward to seeing what she has planned to share with us this month. Please join me in welcoming her to the series (and the blog)! She will be introducing herself in the days ahead.  So, stay tuned.

The upcoming program for the Featured Author series can be found below the fold.