The Morality of Lying and the Murderer at the Door

The article engages with some of the main claims in chapter 1 of Seana Shiffrin’s book Speech Matters. There, Shiffrin sets out a case for a general moral prohibition on lying, based on the conditions required for reliable speech, and circumscribes the permissible falsehoods that could be uttered to would-be moral criminals, such as Kant’s familiar murderer at the door. I raise a few questions about the case for the general moral prohibition on lying and about Shiffrin’s basis for distinguishing between the sorts of lies that, on her view, one is and is not permitted to tell would-be moral criminals so as to avert harm.

Is lying always wrong? Is it even generally wrong? In the first chapter of her thought-provoking book about the philosophy of free speech, Seana Shiffrin seeks to shed some light on these and other questions about the morality of lying through an extended exploration of a moral problem well known to Kantian scholars: the murderer at the door. The basic features of the example are familiar to most ethicists. A would-be murderer knocks on your door and demands to know the whereabouts of his intended victim, which you know. Only by lying to him can you thwart his murderous enterprise. Are you permitted to lie?
As Shiffrin recounts, contrary to the common intuition that that lying to the murderer is the only reasonable course of action, Kant is famously supposed to have claimed that the moral prohibition on lying makes no exception for the murderer at the door scenario. I take it that Shiffrin's interests, like my own here, lie less with exegetical questions such as whether Kant should really be taken to have endorsed an exception-less prohibition on lying, and more so with ethical questions about when, if ever, it is morally acceptable to lie. I therefore set aside all exegetical issues in what follows. I am also going to set aside questions about the conceptual distinction between lies and deceptions, which is one of Shiffrin's interests, and simply take her definitions as given.
Shiffrin defines a lie as a false proposition about the contents of the speaker's mind, intended by the speaker to be taken as true by the addressee. This definition accounts for the fact that one can say something which happens to be true (such as, 'it's the Queen's birthday today') and still be lying if one does not actually believe it or is completely indifferent as to its truth (i.e. you don't know and you don't care). Shiffrin furthermore claims that there is no 'success' element to lying: you do not have to be believed to be lying, or even think you stand a good chance of being believed, so long as the proposition is still offered to be taken as true (Shiffrin's humorous example is of the undergraduate with the succession of grandmothers' funerals to go to, explaining her absence from class). And finally, it is an important part of Shiffrin's definition of lying that not all assertions which the speaker does not believe are, ergo, lies. False assertions made in what Shiffrin calls 'suspended contexts' (e.g. storytelling, joking, devil's advocacy), where the presumption of the truthfulness of statements has been consciously put aside, are not lies. More fully, what Shiffrin refers to as 'justified suspended contexts' arise where there is a 'normatively justified reason for the suspension of the presumption of truthfulness' (17). This could be true in many scenarios where there is some value in the ability to make false statements not meant to be taken as sincere, as in dramatic productions, or fiction, or some forms of social etiquette ('Nice haircut!').
As I understand it, one of Shiffrin's main aims in this chapter is to construct an account of the wrongness of lying which avoids being cashed out in terms of maximising good consequences. Another aim, I take it, is the vindication of a general moral prohibition on lying which, although admitting some exceptions, including in the case of the murderer at the door, holds lying to be typically wrong. Her ambition is to sketch a theory for a 'qualified form of Kant's absolutism about lying' (6) based on Kantian or neo-Kantian forms of reasoning: a theory which is non-consequentialist, retains the general prohibition, but cuts us some slack with the murderer at the door (MATD) and analogous dilemmas. Finally, Shiffrin aims to show that although the general prohibition on lying allows for some exceptions, there are limits on the kinds of lies one can tell even the MATD, especially lies ventured after one makes the point of reinforcing one's commitment to honesty, despite the unusual circumstances. I want first to look more closely at Shiffrin's defence of the general moral prohibition on lying, before considering the rationale and boundaries of the allowable exceptions under her theory.

I. THE GENERAL PROHIBITION
Shiffrin begins her analysis by laying out a certain tension which surfaces when we think about the rule against lying. Those of us who think that lying is generally wrong will much prefer a deontological to a consequentialist grounding for that moral prohibition. We do not want to say that selfish lying is permissible as long as the personal gains of the liar outweigh the costs of the lie, and that would seem to follow from a consequence-based prohibition. However, there are cases in which all reasonable people find it hard to resist the conclusion that lying is permitted. This is especially true of an altruistic lie told to a would-be moral criminal in order to avoid serious harm to another -the MATD. Yet it seems that the only way to explain exceptions of this kind is by appeal to consequences. Things will just be better, perhaps in some gravely serious respects, if one lies. This creates a problem for a general, deontological prohibition. For if it is acceptable to break the rule against lying in order to maximise good outcomes (or avoid bad ones) in the MATD case, it is not clear why it isn't always acceptable to lie for the greater good, including where that good is one's own. But if lying is acceptable just whenever it will maximise good states of affairs, it seems we can potentially make an exception of any lie, be it little and white ('Great article!') or a big fat whopper ('Yes, we did find weapons of mass destruction in Iraq'). In that case, the general prohibition appears to dissolve altogether.
As I see it, that is the fix from which Shiffrin is trying to extricate us with her account of the wrongness of lying, an account which she claims is both deontological and able to resist complete capitulation to the murderer at the door. The kernel of Shiffrin's account of the general wrongness of lying is the value of preserving reliable chan-nels of communication about the contents of our minds. For all sorts of reasons, we need to be able to trust each other's testimonies about what is going on in our heads. Reliable speech is important for many ends: co-operation, mutual understanding, human relationships, moral assessment of others, and so on. Since we cannot get along without trustworthy channels of communication, the argument goes, we cannot commit ourselves to the maxim that it's fine to lie when doing so maximises good consequences. Presumably, that would just too often be the case (or arguably be the case) for reliable communication to be preserved.
The Kantian dimension of the account comes in the importance of universalising the maxim on which we act when we lie. Shiffrin claims that: …the wrong of lying is that it operates on a maxim that, if it were universalized and constituted a public rule of permissible action, would deprive us of reliable access to a crucial set of truths and a reliable way to sort the true from the false. (23) In other words: what would happen if everyone behaved that way? There is something distinctly consequentialist sounding about this question (maybe the kind of consequentialism which, on one view, inheres in all Kantian maxims). It is the potentially bad consequences of consequentialist-like thinking about lying that explains why we cannot accept that maxim as universal. Reliable communication cannot withstand everyone lying just whenever it seems like the best thing to do, the argument goes. Shiffrin is keen to fend off any suggestion that her explanation of the wrongness of lying is really consequence-based. She says (26) that although her argument rests on the moral purposes and value of truthful communication, that does not make it consequentialist, since she is not claiming we should try to 'maximize the realization of moral communication', or that the permissibility of any particular lie 'hangs on the test' that it individually diminishes the goods associated with truthful communication, which we know cannot be true of all lies. It is more the principle of the thing: we cannot countenance everyone lying whenever it seems like the best thing to do, so we shouldn't do that ourselves.
A consequentialist argument for the prohibition could well acknowledge that not every instance of lying contributes to the erosion of reliable communication and still be consequentialist. A rule-consequentialist approach might hold, for instance, that our epistemic shortcomings about which lies will or will not erode communal trust can justify a rule of thumb against lying as the best way to ensure reliable enough speech. Still, we can, I think, accept Shiffrin's claim that her account of the wrongness of lying is deontological rather than consequentialist insofar as it is not interested in the maximisation of any goods per se, but in the acceptability of the maxim we act upon when we lie. That maxim, she claims, is simply not reconcilable with the conditions required for trustworthy communication and all kinds of related goods. Since we cannot endorse it as a universal maxim 'without frustrating our compulsory end of ensuring that communication remains reliable' (26), it is a principle of action we cannot accept.
Whether one formulates the argument as Kantian or consequentialist, the key question is, I believe, the same. Can the need to preserve reliable channels of communication somehow ground a general moral duty not to lie? First, it is obvious that not every instance of lying contributes to the disintegration of trust in speech, even between speaker and listener, let alone more broadly. In fact, one essential condition of that relationship is what we can call the 'discovery condition'. Couples do not experience trust issues when one of them is systematically lying to the other; they experience trust issues when it is discovered that one is systematically lying to the other. Discovery is the bridge between any instance of lying and a breakdown of trust. The discovery condition might suggest that, at most, Shiffrin's argument only establishes a duty to lie well. One might reply that the risk of being found out and damaging trust is present in every lie, but then, every liar already knows that. The shrewd liar will simply plug that risk into her initial calculations.
In reply to this, it might be pointed out that not everyone can behave in this calculating way all of the time without resulting in too much discovery and too much damage to trust for reliable speech to be preserved. And perhaps that is true. But many people can take calculated risks with lying a great deal of the time without threatening the reliability of all communication. In fact, a fair amount of lying does go on, and we somehow manage to get along regardless, with enough reliability of communication to achieve the ends with which Shiffrin is concerned (effective co-operation, mutual understanding, relationships, and so on). The would-be liar could simply defend herself by saying that as long as enough people are honest enough of the time, which they are, the threat of destroying general trusted speech simply does not loom, in which case there is no reason to abstain from any individual act of lying so as not to risk contributing to it.
Thinking in terms of universalisable maxims, however, it may well be objected that, on Shiffrin's account, it doesn't matter whether an act of lying is likely to contribute to a genuine threat of communication break-down, but only that, since not everyone can lie whenever it seems optimal, it is wrong to take the privilege to do so for oneself, thus acting on a maxim one cannot will everyone else to act on. One potential problem with this argument is that it is not entirely clear that reliable enough speech could not withstand everyone acting on the maxim that it is acceptable to lie when, say, the gains are appreciable and risk of discovery very small. Operating on this maxim does not amount to pathological lying, which certainly would threaten productive communication if practised by all. Neither is it obvious that someone adopting this maxim will be moved to lie frequently at all -even less frequently will they be caught in a lie. It cannot be said with much assuredness that a universal maxim such as this presents any serious risk of deteriorating trust or knowledge to the point of damaging the basic goods of communication.
Of course, there are plenty of occasions when lying is an unreasonable thing to do. This is especially so where the relationship of trust between speaker and listener is particularly important, so that it is not worth risking discovery. Perhaps the parent-child relationship is a good example of a context in which a breakdown of trust is too disastrous to countenance almost any lie, even silly ones about the existence of Santa Claus. 1 Much of the time, however, perhaps even most of the time, calculated misreporting of our thinking does not hinder co-operation, damage trust, or contribute to the erosion of the general reliability of all speech (particularly so where the lie is not discovered). Thus, the need for us to remain truthful enough of the time to maintain generally reliable channels of 1 It might be suggested that a parent's attestation about the existence of Santa Claus is an example of what Shiffrin calls a 'suspended context', where the speaker's assertions are not meant to be taken by the listener as true reports of the speaker's mental contents, and as much is clear to both parties. However, this would only be so where the ''Of course Santa is real'' assertion is said with a nod and a wink that is well understood by the child as a signal that the parent is humoring him. communication only seems, to me, to yield a highly contextdependent prohibition, not the sort of general prohibition that Shiffrin supports. (Police organisations lying to the public time and again about their own misconduct may be one very striking instance in which potential damage to general trust is not worth the gains of lying.)

II. THE MURDERER AT THE DOOR
Despite arguing for a general prohibition on lying, Shiffrin does want to allow for some exceptions. With qualifications, we can lie to the murderer at the door. Why? Our instinct here is to say that the consequences of not lying are just too dire in this case. But we have already seen the problem there. As Shiffrin says, '…the deontologist cannot merely stipulate that declarations need not be true when truth-telling will harm another or cause poor consequences.' (29). If we can bend the rule because the benefit of lying in this particular instance outweighs the costs, then why can't we do it wherever that is the case ('Yes, our food is lovely, thank you.')?
To escape this difficulty, Shiffrin ventures a non-consequentialist explanation for why the prohibition relaxes with the Murderer at the Door: the murderer is not owed truthfulness about things to do with his criminal enterprise, and we would do wrong to cooperate with that enterprise through the use of honest speech. She writes: One in hot pursuit of an evil end should have no reasonable expectation that the world or other people cooperate with her evil enterprise. She has no entitlement to those warrants about the world that would materially further those ends. (34) Generally, the murderer has no moral right to assistance, and you have no moral right to provide it. (35) Importantly for Shiffrin, this explanation does not make any lie to a moral criminal acceptable, only some lies. One cannot say that someone on an immoral enterprise has no general right to the truth from others, since the wrongdoing of moral criminals should not disqualify them from all standard moral treatment, she says. Some false representations do not appear to be acceptable. This particularly includes so-called 'strategic lies', such as falsely telling the would-be murderer that the death penalty is regularly applied in that state, or falsely telling a hostage-taker that her demands have been met so as to effect the release of the hostages (31).
Instead, Shiffrin argues that the right to lie to moral criminals only covers utterances the content of which would, if truthful, amount to a kind of complicity in the immoral enterprise. The moral criminal is not entitled to truthful utterances that would materially further her criminal end. There exists, she argues, a 'justified suspended context' around such utterances whereby (as in fiction or social etiquette) the normal presumption of sincerity is justifiably suspended. On this approach, although we can misrepresent information which would, if truthful, assist a crime (such as the whereabouts of the victim), we cannot just falsify anything in order to deter the crime (such as the applicability of the death penalty). But why is this?
Shiffrin reasons that, even when he is on an immoral enterprise, it is vital that we preserve for the criminal a route to redemption and conflict resolution. She worries that if no duties of truthfulness were owed to the would-be murderer, we would place him in something like an epistemic 'solitary confinement', 'with no access to reliable information about the world and no way out' of the context of insincerity (38). This need to preserve 'a way out' means that one kind of misrepresentation is particularly forbidden: where one purports to exit the 'justified suspended context' of insincerity by declaring that one is now going to be honest and then lies (e.g. 'Ok, look, I'll level with you, the wallet is in the glove compartment.'). Shiffrin follows Kant here in saying that the fact that a moral criminal is not entitled to the truth on particular matters (such as 'Where is the victim'? or 'Where is the wallet'?) creates a special context around those communications where sincere communication is presumptively and justifiably suspended, much like it is justifiably suspended in joke-telling, devil's advocacy or fictional narrative. However, to retain for the criminal a route back into the moral community, it is vital to preserve the possibility of reliably exiting this 'suspended context' of insincerity. Like abusing the white flag of surrender in war, the kind of lie where one makes a pretence of exiting the suspended context of insincerity and volunteering the truth (the now-I'm-really-telling-the-truth lie) destroys this possibility and keeps the wrongdoer in inescapable moral isolation.
Shiffrin draws on the harm of 'solitary confinement' to convey the severity of the situation in which we place the MATD if we can tell him any kind of lie to thwart his harmful aims. As she rightly notes, 'solitary confinement is one of the cruelest forms of punishment'. To be sure, literal solitary confinement is torture. But Shiffrin here is only using the notion of solitary confinement as a metaphor to depict the kind of epistemic situation in which the moral criminal finds himself if placed in a 'truth-free zone' (38). The tortuousness of literal solitary confinement can hardly be compared with the harm that is done to a person who is treated as having no right to the truth from others whilst on an immoral enterprise. And this difference matters a great deal if the harmfulness of that epistemic isolation is the main consideration we are weighing against the benefits of various misrepresentations to the moral criminal.
Further to this, the degree to which the murderer is cut-off from the truth is still extremely limited if we think he forfeits his right to the truth not just about the whereabouts of the victim, but about all matters the misrepresentation of which could deter the crimewhether or not there is a death penalty, for example. It is difficult to accept that the difference between telling the would-be murderer that the intended victim is running to the beach (a lie Shiffrin deems permissible) or telling him falsely that the death penalty is regularly used, is the difference between his being placed in complete moral isolation and not being so placed. And it seems reasonable to think that if he forfeits his right to the truth about the whereabouts of the victim, he also forfeits his right to the truth on other matters the falsification of which could deter his wrongdoing. Complete moral isolation only follows if his wrongdoing means that anyone can lie to him about anything at all, but this is not a claim anyone would be tempted to make.
Completely gratuitous and pointless lying by anyone who might knowingly come into contact with a moral deviant could still be impermissible, but I do not think this is where the tension between harm avoidance and a deontological prohibition on lying truly resides. That tension only comes into its own where a certain misrepresentation seems to be the only way of avoiding harm. This can be true of both reactive and volunteered falsifications, and might well be true of the now-I'm-really-telling-the-truth lie.
Shiffrin recognises that there are cases in which such a lie 'is not gratuitously volunteered, but [is] necessary to avoid direct exposure of one's initial [permissible] misrepresentation' or of what one concealed (43). But this is potentially a very big problem for Shiffrin's 'content-based' approach to permissible lies. In some emergency scenarios, an elaborate, embellished, volunteered lie is the only credible lie. As any well-seasoned liar will tell you, the more detail ventured, the better the lie. In situations like these, it is always going to be open to re-describe what looks like a gratuitous misrepresentation as a reactive lie (or part of one) in response to an emergency. (Consider: 'The victim is running to the beach! I know because I saw her as I was driving back from the garage, where my car was getting new brakes fitted, and I got a good look at her as I was waiting at the traffic lights.') Importantly, this can equally be true of misrepresentations that pretend to exit a suspended context (the now-I'm-reallytelling-the-truth lie). And where the consequences are serious enough, one may just think that preserving an exit from insincerity for the moral deviant is not the most important thing.

III. EXITING THE SUSPENDED CONTEXT
Shiffrin seems to maintain that the consequences of averting harm are always subordinate to the values protected by the prohibition on lying, including where that value is maintaining 'a way out' for the wrongdoer. This is where I think someone can reasonably disagree. Yet, according to Shiffrin, a more general value is at issue whenever a person misrepresents about exiting a suspended context. As she writes of Kant's mugger example (in which a mugger's victim makes a show of now telling the mugger, (but falsely), where his wallet really is), when we falsely declare our intention to break of out of the presumed context of insincerity, we 'implicitly affirm a principle that, if public, would undermine our ability to exit the suspended context' (42). In other words, faking this exit with the now-I'm-really-tellingthe-truth lie puts in jeopardy not just the moral dignity of the criminal, but the value of maintaining the possibility of ever exiting suspended contexts and re-establishing sincere moral relations. Like abusing the white flag of surrender, such lies might be thought to threaten a crucial avenue for conflict resolution, even where one party remains in the wrong. For Shiffrin, then, the imperative not to lie this way rests not only on the need to preserve an exit from insincerity in the present interaction, but also on the value of maintaining the option in general. We cannot operate on the maxim that 'fake exits' can be justified since, as I understand her, it cuts against the conditions for exiting presumed insincerity and conflict resolution.
Depending on how one formulates the maxim, however, and the levels of harm involved, we might yet disagree about whether it is a maxim we can act upon. Shiffrin contends that when someone falsely breaks out of a suspended context, 'he threatens the safety of all of us, because he has rendered his word unreliable and given us reason to distrust people's word more generally' (41). However, if this is what needs to be true for the 'fake exit' maxim to be rendered permissible, it brings back into the fore questions about the resilience of trusted speech to lying in many contexts, particularly so contexts involving moral criminals, which might be regarded as bracketed in so many ways. Is a maxim permitting 'fake exit' lies to moral criminals in order to avoid harm really incompatible with trusted speech in general, or with the general reliability of declarations to exit contexts of suspended truth?
One worry could certainly be that 'fake exit' lies to moral criminals damage the reliability of such declarations in those contexts in particular, which would be of some negative value. Perhaps there is an effectiveness requirement that, as with the white flag, one, or others, ought not 'cry wolf' too many times to make a purported exit from insincere relations dependable. But this is a consideration the strength of which will still always be context-dependent (sensitive to features such as discovery and promulgation), and can always itself be counterbalanced by the harm sought to be avoided in the given instance. When it comes to the white flag in particular, I imagine that the dire importance of retaining trust in that particular communication gives rise to powerful reasons against ever abusing it, almost regardless of the consequences.
It is worth remembering that, on Shiffrin's account, it is not actually the harmful consequences which permit us to tell some falsehoods in the MATD case, but rather the addressee's wrongful enterprise and our obligation not to be complicit in it. This, she says, gives rise to the kind of context in which there can be no real expectation of truth between the parties, as Kant said is true when a mugger demands to know where his victim's wallet is. On this understanding, the misrepresentation to the murderer is less of a justified lie and more so not a lie at all, the way that fictional narratives or certain forms of social etiquette -'Lovely to see you!', when it wasn't -are not really lies. Following this, Shiffrin's contentbased approach to the false utterances that we may permissibly make to a moral criminal is itself a guide to what is or isn't truly a lie in the context. This reading would have the somewhat counter-intuitive (to some) implication that harm avoidance itself is not what renders some false statements to the MATD permissible. The suspended context around those permitted falsehoods ('She went that way!') would mean that they are not lies, and thus not in need of any harmbased justification. But the false utterances to the MATD certainly seem to be justified lies, rather than non-lies. One might wonder why the MATD scenario is a puzzling case for the general prohibition on lying if the permitted falsehoods are not really lies, because stated in a context of presumed insincerity.
Shiffrin may respond to this by saying that there is still the need to provide a justification for the suspended context, as is true of other contexts of suspended truth-telling, like fiction or social etiquette. This is where the relevance of the moral criminality comes in. Here, what renders the suspended context justified around certain 'lies' to the MATD is the fact that those utterances would assist in his immoral enterprise and the fact that he consequently has no right to honesty about such matters. But the justified suspended context in which the would-be murderer has no right to expect honest speech from the addressee at the door is meaningfully different from the suspended contexts in which there is a mutual understanding that, for some good reason, we should not be taking each other's statements (like 'nice haircut') as really true, or even offered as true. The suspended context surrounding the MATD is not like this. What creates the justified suspended context there, it seems, is just that so much hangs on your need to lie, rather than there being some general (probably shared) value in not taking each other's words seriously in the context. Indeed, for the lie to achieve its important end, it must be taken somewhat seriously by the would-be murderer.
On this analysis, the MATD case is not a 'suspended context' at all, but rather an instance of justifiable lying.

IV. THE DUTY OF VERACITY
In sum, whether they are non-lies because justified, or justified lies, it might be thought that a whole range of false utterances to the would-be murderer can be defended on the reasoning that, given what he wanted to do, he simply had no right to the truth from you about anything that could potentially stop him, and that Shiffrin's concerns about the moral isolation of criminals and preserving exits from insincerity will simply not always weigh heavily enough in the circumstances.
The question of the murderer's right to the truth leads to my final point. Shiffrin is willing to concede that moral criminals lack a right to the truth from certain people about certain things. However, once our attention is fixed on this right to the truth, and the corresponding duty of veracity, we might wonder whether the variability of both goes far beyond this. It seems to me unexceptional, maybe even usual, for people to lack a right to the truth on a variety of matters, and not only because they are moral criminals. We may think, for instance, that the waiter in the restaurant is simply not entitled to know my real thoughts about the quality of the food if I would rather not interrupt my conversation to have a whole discussion about it. If, on the other hand, I had trained the restaurant's chef, then he or she could plausibly claim a right to know my genuine judgment. Partners in monogamous relationships have a right to the truth about the question of each other's fidelity, but other people do not. And there are plenty of private thoughts or beliefs that we all have which no one can claim a right to know the truth about. As soon, then, as the inquiry turns from the justifications for lying to the conditions for the duty of veracity, anything in the way of a general duty to be honest becomes elusive. This is not only a question of non-disclosure and privacy. Shiffrin of course does not contend that we all have duties to speak the whole truth about matters that do not concern other people, and which we have the right to keep private. The problem here is that, as I noted earlier, resisting unreasonable demands for the truth can require far more in the way of volunteered lies than may often be appreciated. Sometimes, only a volunteered and embellished lie can secure a person's right not to tell the truth. And a whole host of considerations may determine when addressees are owed a duty of veracity, of which moral criminality is only one.
This does not resolve the tension between deontological and consequentialist approaches to the morality of lying, but it might reposition it. That tension will simply not arise in the case of the MATD where the would-be murderer does not, in any case, have a moral right to the truth about the matters we are driven by necessity to misrepresent. Rather, that tension will arise in cases where it seems that the addressee does in fact have a right to know the truth, but where there are strong consequence-based reasons to lie ('We don't think your father died in any pain'). However we would seek to resolve this problem within the morality of lying, it seems that the MATD case will not be the example on which to focus.

OPEN ACCESS
This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Hertford College, Oxford OX1 3BW, UK E-mail: Kate.greasley@hertford.ox.ac.uk Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.