Bennett’s Expressive Justification of Punishment

Abstract

In this paper, I will critically assess the expressive justification of punishment recently offered by Christopher Bennett in The Apology Ritual and a number of papers. I will first draw a distinction between three conceptions of expression: communicative, motivational, and symbolic. After briefly demonstrating the difficulties of using the first two conceptions of expression to ground punishment and showing that Bennett does not ultimately rely on those two conceptions, I argue that Bennett’s account does not succeed because he fails to establish the following claims: (1) punishment is the only symbolically adequate response to a wrongdoing; and (2) punishment is permissible if it is the only symbolically adequate response to a wrongdoing.

1 Introduction

Christopher Bennett has offered some rich and thought-provoking justifications for retributive punishment that appeal to the concept of expression in The Apology Ritual: A Philosophical Theory of Punishment (hereafter, The Apology Ritual) and a number of papers. I argue that they do not succeed. While I take Bennett’s works as my primary target, I believe that the discussion below will have a more general interest as it illustrates some general difficulties with any attempt to justify punishment by appealing to the concept of expression.

Before I proceed, let me confront a significant ambiguity. What does Bennett mean when he speaks of a justification of punishment? It is now customary to draw a distinction between justifying the goodness of punishment and justifying the permissibility of inflicting punishment, i.e., showing that it does not violate the offender’s rights.¹ The two issues are distinct: it can be permissible to do something that is not good, and vice versa. For example, it is permissible to swear at others (in the sense that such an act does not violate their rights), but it may not serve any good; we may achieve some good by killing a person to save two others, but it may be impermissible to do so. Which question (or both) is Bennett trying to answer?

I take it that Bennett intends his account of punishment to provide an answer to the permissibility question in addition to answering the goodness question. This is because he is clearly aware of the distinction between the two questions;² and his criticism of instrumental justifications of punishment, such as deterrence, is that instrumental justifications ignore individual rights.³ If Bennett’s account of punishment can only establish the goodness but not the permissibility of punishment, then his account falls prey to a similar criticism: even if we can achieve some good in punishing offenders, we ignore their individual rights in doing so.⁴ In what follows, I will assess Bennett’s account in terms of whether it provides a plausible answer to the permissibility question; that said, I believe many (though not all) of my objections will apply even if we understand Bennett’s account as only trying to answer the goodness question.

Let me consider one objection to my claim that Bennett has to be at least partly concerned with the permissibility issue. The objection is that the permissibility of punishing can easily be established by appealing to the consent or voluntary undertaking of the offender, assuming that the law stipulating the penalty is announced well in advance. As such, the only real issue facing a retributivist is that of explaining why there is a non-consequentialist good in imposing punishment. In reply, voluntary undertaking and consent cannot account for our intuition that the right to punish, if there is one, incorporates a proportionality constraint. If the law stipulates that life imprisonment is the penalty for petty theft well in advance, then there could be no objection based on lack of consent or voluntary undertaking to imposing such a punishment.⁵ But most people think that this would be a violation of the petty thief’s rights. Accordingly, I think the permissibility issue cannot be disposed of so easily.⁶

2 Three Conceptions of Expression

To start with, we should observe that the idea of expression can be understood in different ways; without being clear about what amounts to expression, it is hard to assess the plausibility of an expressive justification of punishment.⁷ Let me start by distinguishing between three senses of expression.⁸

Sometimes we take an act to be expressing a certain thing if it communicates that thing. For example, we may say the act of raising a thumb towards my son expresses approval, since it communicates such message to him. Let us call the sense of expression here communicative expression.

But this cannot be the only sense of expression, as sometimes we say an act expresses something even if the agent did not communicate anything.⁹ Communication requires a certain intention on behalf of the agent and the existence of an audience.¹⁰ But it is possible to say an act expresses something even if there is no communicative intent or audience. For example, suppose a relative of a deceased person forgot to attend the funeral and no one noticed; we may still say that the relative’s act expresses something (say, his lack of affection for the deceased) even if there is no communicative intent or audience.

This brings us to the second sense of expression, according to which an act expresses a certain attitude if and only if the act is motivated by that attitude.¹¹ Let us call this second sense of expression motivational expression. Since the relative’s failure to attend the funeral is due to an improper attitude, his conduct expresses something, namely his improper attitude, in the motivational sense, even if it does not express anything in the communicative sense.

But communicative expression and motivational expression do not seem to exhaust all senses of expression, for two reasons. First, sometimes we say an object expresses a certain thing even if there is no communication (and hence the thing is not expressed in a communicative sense) and even if the object does not possess any attitudes (and hence the object cannot express anything in the motivational sense). For example, we can say that an effigy expresses the person it represents or a picture expresses sadness, even if they have not been seen by anyone.¹² The second reason comes from the fact that there seems to be a sense of expressive acts not captured by communicative expression or motivational expression. Expressive acts need not be communicative—banging my fist on the table privately can be an expressive act, for example. On the motivational sense of expression, any act motivated by an attitude, like going to work, can be called an expressive act—since the act expresses its motivating attitude.¹³ But, while it is possible to use the idea of expressive acts in this broad sense, we often use the idea in a narrower sense, as referring to acts like mourning and performance of rituals rather than any other ordinary acts.

These two reasons point to the idea of symbolic expression: an effigy can symbolize a person even if no message is communicated and the effigy does not possess any attitudes; when agents engage in expressive acts in the narrower sense, they are thought to be acting on a special type of reasons, namely, symbolic reasons. Let us call this third sense of expression symbolic expression: an act expresses a certain thing in this sense if and only if it symbolizes it. What an act expresses in this sense does not solely depend on what the act communicates or what attitudes motivate the agent in acting. As an example, if a king passes a racist law, some may want to say that his act symbolizes racism even if he was not motivated by racist thoughts and even if no one takes the king to be communicating racism.¹⁴

I wish to highlight one more difference between symbolic expression, in the sense intended by Bennett, and communicative expression (besides the fact that communicative expression requires a communicative intent and an audience). Some symbols are purely conventional: for example, there is no correct answer, beyond conventions, as to whether the character “5” is the proper symbol for the number “five.” But, according to Bennett, some symbols are not conventional and accordingly some symbolic reasons, including the ones that can help to justify punishment, are not reasons based on convention.¹⁵ Since I am trying to assess Bennett’s account of punishment, below when I refer to symbolic reasons I shall mean non-conventional symbolic reasons. As convention is central to communication,¹⁶ symbolic reasons, in Bennett’s use of the term, are very different from reasons based on communication.

I have just drawn a distinction between three senses of expression: communicative, motivational, and symbolic. Correspondingly, we can understand an argument for punishment that relies on the concept of expression in three different ways: first, an argument that claims that punishment is justified because it is necessary to communicate the proper message; second, an argument that claims that punishment is justified because it is necessary for one to act properly in the motivational sense—in other words, it is justified because if a person has the proper attitudes in response to a wrongdoing, he will be motivated to punish; third, an argument that claims that punishment is justified because it is the only symbolically adequate response to a wrongdoing. Let us survey these three arguments below.

3 Communicative Expression

Bennett sometimes frames his theory as one based on communicating censure,¹⁷ but I think it is clear that the idea of communicative expression does not play a central role in his theory.¹⁸ Moreover, there are two well-known problems with a justification of punishment that focuses on communicative expression.

The first problem is that, in order for punishment to be justified by communicative expression, it has to be the only effective way to communicate the proper message; but that seems to be incorrect. To understand this problem, let us start by asking what amounts to a successful communication. A rough working account would suffice for our purpose. Successful communication between A and B means roughly that A intends to send a message Y to B by doing X, and B is aware of such communicative intent when he perceives X.¹⁹ A sufficient condition for B’s grasping A’s communicative intent is that there is a shared linguistic convention between A and B covering the relevant message. To illustrate, if I say Guten tag to my friend in Germany, I intend to send the message that translates to Good day in English. If my audience knows the convention (i.e., he speaks German), then he will grasp my communicative intent; the communication is therefore successful.

Given that the intended audience of criminal justice (the offender or the public) and the state generally share a sophisticated linguistic convention, it is hard to believe that the proper message, whatever it is, cannot be communicated by mere words. Perhaps mere words do not do symbolic justice to the situation, or a properly motivated person would act instead of speak—but, even if this is true, it only shows that, in deciding what we should do, we should not only care about whether we are communicating properly; it does not show that we cannot communicate properly using mere words.²⁰ It does not serve any intellectual purpose to jam all normative considerations into the idea of communication.

There are two related errors that I want to briefly discuss here. The first is the idea that, for a punitive act to be successful as an instance of communication, the offender must truly understand that his conduct was wrongful after our act.²¹ This is an error, because successful communication only requires a grasp of the communicator’s message, not agreement. In any case, appealing to the persuasion of the offender is, as many writers including Bennett have observed, an implausible consequentialist route to justifying punishment.²² The second related error is the thought that, for communication to be successful, the audience must believe that the communicator is sincere.²³ For example, it may be thought that, if my wife does not believe that I love her when I say “I love you,” the communication is unsuccessful. This is an error because it seems that in the case just mentioned the communication is actually successful; only if she grasps what I mean can my wife criticize me for being fake or insincere. In any case, appealing to the audience’s belief of the communicator’s mental states cannot provide a plausible justification of punishment. Suppose a society wrongly believes that if the state does not execute a petty thief then the state is condoning theft. On the theory just mentioned, we must execute the petty thief in order to send out a condemnatory message that is perceived as sincere by the audience.

The second reason communicative expression cannot justify punishment is that, even assuming that punishment is the only way to communicate a certain proper message, it is not normally permissible to use force to communicate messages. Suppose we cannot communicate a proper message to a hostile neighbour, who does not want to talk to us, unless we detain him; clearly that would not make it permissible for us to detain him.²⁴

4 Motivational Expression

Bennett sometimes writes as if the justifiability of punishment depends on the justifiability of the retributive emotions;²⁵ this may appear to some to suggest a defence of punishment based on motivational expression. The strategy currently under consideration justifies punishment by arguing that, first, holding certain emotional attitudes, like resentment, is a proper response to a wrongdoing; second, an emotion, in addition to having a phenomenological component, has a dispositional component: as Bennett writes, emotions are not mere feelings but are partly constituted by the actions that they motivate us to perform;²⁶ in the case of resentment, holding it will motivate one to act punitively.²⁷

Before we assess this strategy, we should distinguish between two conceptions of the propriety of emotions, depending on how tightly the propriety of emotions is connected to the rightness of acts.²⁸ On the first conception, the propriety of emotions is only loosely connected to the rightness of acts. Whether an emotion is proper, on this conception of emotional propriety, depends on independent criteria that are not concerned with the rightness of acts motivated by the emotion: an emotion’s being proper does not entail that the act motivated by it is right (even assuming that the agent has the correct factual beliefs and his other mental states are proper).²⁹ On the second conception, the propriety of emotions is tightly connected to the rightness of acts: if an emotion is proper, then the act motivated by the emotion must be right (assuming that the agent has the correct factual beliefs and his other mental states are proper).

Of course, if we understand emotional propriety in the first sense, then even if a properly motivated person would punish, it does not entail that punishing is the right thing to do. Michael Moore, however, argues that this can be a piece of good, even if defeasible, evidence for the rightness of punishment.³⁰ I do not think Moore’s strategy can help to justify punishment. The main problem is explaining why the fact that an emotion is proper, according to the sense of emotional propriety under which retributive emotions can be said to be proper, provides a reliable, even if fallible, epistemic indicator for the rightness of acts motivated by such an emotion. Let me illustrate this problem with a brief critique of Moore’s ideas. According to Moore, whether an emotion is proper does not depend on whether the acts motivated by the emotion would be right, but on whether it satisfies some other criteria, including intelligibility and the consistency of the emotion over time.³¹ One may wonder whether retributive emotions are more proper than anti-retributive emotions on Moore’s criteria, since Gandhi’s anti-retributive emotions seem quite intelligible and consistent over time as well. Moore may, of course, amend his criteria for emotional propriety in such a way that retributive emotions, but not anti-retributive emotions, would satisfy them;³² but the problem is that we then need an explanation for why the fact that an emotion is proper in that particular sense gives us good reason to believe that the acts motivated by the emotion are likely to be right. This cannot be answered without some empirical evidence of the correlation between the propriety of emotions (in the specified sense) and the rightness of acts; but no such evidence has been offered.

In the passages where Bennett refers to retributive emotions, he does not seem to take the justifiability of retributive emotions as merely of evidential significance in justifying punishment.³³ Can punishment be justified by motivational expression, if we understand the propriety of emotions in the second sense, as tightly connected to the rightness of acts? The problem here is that an appeal to emotional attitudes seems superfluous in justifying acts, including acts of punishment.³⁴ (My criticism, as we will soon see, is a variation of the oft-made circularity charge against virtue ethics, namely that virtue ethics cannot help in answering the question of what acts we should perform).³⁵

An emotion is not proper, in the sense specified here, unless the acts that would be motivated by such an emotion are right; the problem is that we seem to be more confident about what acts are right than what emotions are proper in this sense. To put it in another way, the judgment of whether an act is right enjoys epistemic priority over the judgment of whether an emotion is proper (in the sense of emotional propriety relevant here). Using an example from Bennett to illustrate this, consider a person who hugs a stranger after hearing a piece of good news about his own daughter’s health situation. According to Bennett, the action is not justified because the emotion is excessive: while one should act in excitement, one should not be so excited as to hug a stranger without his consent.³⁶ But it seems that actually we decide that the emotion is excessive only because we know about the permissible boundaries between persons; the emotion motivating the hugging is excessive because it motivates him to violate personal boundaries, not the reverse.³⁷

I want to offer a diagnostic explanation of why some people may find the idea that the question of the propriety of emotions (in the second sense of emotional propriety) is epistemically prior to the question of rightness of acts attractive. We are very confident about the propriety of some emotions, like love or care; we are not very confident about the rightness of some acts, like what to do in trolley cases. It may be easy to infer from those cases, therefore, that judgment of emotion is epistemically prior to judgment of acts. But that would, of course, be an error, as we are not comparing like with like. In order for judgment of emotions to be really epistemically prior to judgment of acts, we have to compare our confidence about an emotion with our confidence about the acts that would be motivated by that particular emotion. Since we are very confident about the rightness of acts motivated by love or care and not very confident about what emotion a properly motivated person would have in trolley cases, a consideration of these cases does not show that the judgment of emotions is epistemically prior to the judgment of acts.

The above explains why I believe that motivational expression cannot help in justifying punishment. However, I do not think that this is fatal for Bennett’s expressive account of punishment because—despite some appearances to the contrary—on a proper reading of his work, Bennett need not be seen as relying on motivational expression. Bennett does discuss the issue of appropriate emotional responses to wrongdoings in The Apology Ritual; but the question that he devotes most attention to is not that of what emotions the state or the public should have, which would be the central issue for an expressive account of punishment that relies on motivational expression. His focus, rather, is on what emotions offenders should properly have.³⁸ Bennett believes that answering this question is important only because he happens to also hold the following view about symbolic adequacy: the only symbolically adequate response to a wrongdoing is to impose on the offender the burdens that he would impose on himself if he were properly motivated.³⁹ We should, therefore, take the central driving idea in Bennett’s expressive account of punishment to be symbolic expression rather than motivational expression.

5 Symbolic Expression

Bennett frequently claims that punishment is justified because it is the only symbolically adequate response to wrongdoings by responsible agents.⁴⁰ In order for punishment to be justified by symbolic expression, two claims need to be established. First, punishment is the only symbolically adequate response to a wrongdoing. Second, it is permissible to impose punitive burdens on an offender if that is necessary to make a symbolically adequate response to his wrongdoing. I believe Bennett has yet to offer a convincing justification for both claims and I will examine them separately below.

5.1 Symbolic Adequacy: Meaning

But before we can address the two normative questions, we need to first answer the conceptual question: what exactly does Bennett mean by “symbolic adequacy”? So far my characterization of the concept has been primarily negative. I have said that the symbolic adequacy of an act, on Bennett’s understanding, is not a conventional matter; it is not about communicating a proper message; nor is it simply about acting with the right attitude. But such a negative categorization, while sufficient for the task of distinguishing symbolic expression from the other two senses of expression, does not assist much in answering the two normative questions.

Bennett seems to take symbolic reasons to be a special type of non-consequentialist reasons.⁴¹ What is that special type of reasons? He offers the following actions as acts justified by their symbolic adequacy: “civil disobedience, the act of marriage, expressions of gratitude, acts of grieving and mourning.”⁴² What do these acts have in common? Bennett writes that symbolic reasons are not about “promotion and protection,” but about “marking and acknowledging” the significance of a situation.⁴³ The examples of symbolic act offered by Bennett fits this characterization. We should notice that, while reasons based on promotion are consequentialist, reasons based on protection need not be: deontological side-constraints are also protective principles. So, we can take symbolic reasons to be a special group of non-consequentialist reasons that are about marking and acknowledging, as differentiated from the wider class of non-consequentialist reasons in general (which includes protective deontological side-constraints).

5.2 Punishment as the Only Symbolically Adequate Response

We now know something more about what Bennett means by a symbolically adequate act: this is an act that marks or acknowledges a situation in a proper way. But it does not get us very far towards answering the question of why punishment is the only symbolically adequate response to a wrongdoing. I believe no general criteria for what makes one act a better marker of a situation than another has been provided by Bennett; but without such criteria, one may wonder why Bennett’s claim about the symbolic necessity of punishment should be accepted.⁴⁴ Why would it be inadequate, as the common objection goes, to mark wrongdoings by verbal censure or non-verbal measures that do not amount to punishment?⁴⁵

To be fair to Bennett, while he does not offer any general criteria for symbolic adequacy, he does make an original response to the common objection by arguing that punishment is necessary to mark the difference between wrongdoings by qualified members of the moral community and wrongdoings by non-qualified members adequately; but I will argue that his response is unpersuasive.

Let me begin my critique by introducing some ideas that feature frequently in Bennett’s argument in this paragraph. Some relationships, practices, or communities are intrinsically valuable to their members. Neighbourhoods and the community of teachers are given as examples of valuable communities.⁴⁶ According to Bennett, the moral community is also a form of valuable relationship.⁴⁷ Valuable relationships, however, do not only bring benefits; they are also constituted by their respective obligations.⁴⁸ Qualified members of a relationship can be expected to fulfil their obligations independently without supervision or guidance.⁴⁹ Registered nurses, for example, are qualified members in the practice of nursing care while non-nurses and trainee nurses are not. Offenders, unlike children or the insane, are qualified members of the moral community.⁵⁰

Why does Bennett believe that punishment is necessary to adequately mark wrongdoings by qualified members of the moral community? We can detect two arguments in Bennett’s work: the withdrawal by others argument, which focuses on the withdrawal of respect from the offender by others; and the withdrawal by self argument, which focuses on the withdrawal of respect from the offender by the offender himself.

Let us examine the withdrawal by others argument first. Bennett claims that, in order to adequately mark a person’s status as a qualified member of a valuable relationship, we must withdraw the respect specific to the relationship (i.e., the respect that is due only to members of the relationship but not to others) from him if he has committed an act that is inconsistent with the minimum demands of the relationship. To illustrate, consider an adult neighbour who has behaved in a way that is inconsistent with the minimum demands of the neighbourhood. Bennett claims that in order to adequately mark him as a qualified member of the neighbourhood (as opposed to a child),⁵¹ we must withdraw the respect normally due to neighbours from him, like refusing to greet him like a neighbour anymore, before he makes a proper apology.⁵² Similarly, Bennett claims that, in order to mark the offender’s status as a qualified member of the moral community adequately, we must withdraw the respect specific to the moral community from him, which entails punishment.⁵³

The withdrawal by others argument can work only if the following two claims are true:

• (1) The only symbolically adequate response to a wrongdoing committed by a qualified member of the moral community is to withdraw the respect specific to the moral community from him; and

• (2) Such a withdrawal leads to punishment.

Here, it is helpful to draw a distinction between two interpretations of the moral community. First, we can read it as the political community. Second, we can read it as the human community. Both interpretations of the moral community find some textual support in Bennett’s work.⁵⁴ But clearly claim (2) is false on the political interpretation of the moral community. As pointed out by Gabriel Mendlow, the right to be free from the kind of burdens imposed by punishment is a pre-political natural right: otherwise it would be permissible to impose such burdens on foreigners who have done nothing wrong.⁵⁵ As such, a withdrawal of the respect specific to the political community cannot lead to punishment.

Let me consider a rejoinder here. Can it be argued that the kind of punishment that Bennett tries to justify is much milder than normally conceived and only involves a withdrawal of political goods; accordingly, no withdrawal of pre-political natural rights is necessary for the kind of punishment that Bennett has in mind? There are two problems with this rejoinder. First, if Bennett takes this way out, then he fails to engage with most writings that challenge retributive justifications of punishment; he would simply have changed the question. Second, and more importantly, Bennett does not seem to take this way out. Bennett clearly thinks that community service is a justifiable form of punishment; but imposing community service on a foreigner would be a violation of his natural right to liberty if no justification is provided.⁵⁶ In what follows, therefore, I take the kind of punishment that Bennett tries to justify to involve not only a withdrawal of political goods, but also a prima facie infringement of the natural right to liberty.

Can the withdrawal by others argument work if we take the moral community to be the human community? One may wonder why the wrongdoer’s status as a qualified member of the human community cannot be adequately marked by purely verbal measures.⁵⁷ But, even if we accept Bennett’s claim that words are not enough and acts are necessary to adequately mark such a status,⁵⁸ and even if we further accept that a withdrawal of some of the respect specific to humans is necessary to mark such a status, it will not follow that punishment is necessary. We should observe that withdrawal of respect is a matter of degree: we can withhold some goods specific to a relationship but not others.⁵⁹ For example, we can suspend a teacher’s teaching license without suspending his right to use the teachers’ co-op store. One of the most basic goods, or valuable forms of treatment, that we owe to humans (but not to other animals) is respecting their liberty. There are, however, other goods, which are less basic, that we also owe only to humans but not to other animals. To take one example from T.M. Scanlon, one such less basic good is helping with their projects if the cost to us is trivial.⁶⁰ In order for claim (2) to be true, the withdrawal must be read in a robust sense that involves a withdrawal of the right to liberty. But it is unclear why we should accept claim (1) on this interpretation. Why is it insufficient to mark the offender’s status as a qualified member of the human community by withdrawing the less basic goods specific to the human community from him, like refusing to help with his pursuits in future even if the cost of helping him is trivial? After all, we should not withdraw such goods from non-qualified members of the human community, such as children, who have committed wrongs; we should continue to help children with their pursuits if the cost to us is trivial even if they have committed wrongs. As long as we withdraw such goods from sane, adult offenders, we treat wrongdoings by qualified members and non-qualified members of the human community differently.

Bennett may reply that milder forms of withdrawal, like the ones suggested by Scanlon, would not be a symbolically adequate response to serious wrongdoings because a symbolically adequate response needs to be proportionate to the gravity of the wrong.⁶¹ But it is unclear why someone who is not already a retributivist should regard such withdrawal as disproportionate. In the end, Bennett’s justification for the symbolic necessity of withdrawing the more basic goods from the offender seems to be nothing more than a brute appeal to his retributive intuitions.⁶²

Let us now move on to Bennett’s second argument for the claim that punishment is necessary to adequately mark the offender’s status as a qualified member of the moral community, the withdrawal by self argument. According to this argument, if a qualified member of a valuable relationship has committed a wrong—an act that is inconsistent with the minimum demands of the relationship—then he should withdraw the respect specific to the relationship from himself in order to give an adequate apology,⁶³ and this will necessarily lead to undertaking penance (i.e., “something arduous or demanding that could not normally be asked of [one] as a matter of duty”).⁶⁴ To illustrate, Bennett believes that only a qualified teacher, but not an apprentice teacher, has a duty to withdraw the respect specific to teachers from himself after committing a relevant wrong (say, negligence in teaching); and if a teacher withdraws the respect specific to teachers from himself, then, in addition to compensating the students who were harmed by his wrongdoing, he would perform other burdensome tasks, such as offering free extra classes.⁶⁵ Bennett believes that, similarly, an offender, who is a qualified member of the moral community, has a duty to undertake penance.⁶⁶ Bennett then claims that the only symbolically adequate response to a wrongdoing is to impose on the wrongdoer the burdens that he would impose on himself if he were properly motivated, that is, the burdens that he has a duty to undertake;⁶⁷ accordingly, the only symbolically adequate response to a wrongdoing by an offender is to impose penance on him, which can be seen as a form of punishment.

In reply, I do not want to challenge Bennett’s claim that offenders, unlike non-qualified members of the moral community who have committed wrongs, have a duty to undertake penance. But I wonder why the only symbolically adequate response to the existence of such a duty is to force the duty-holder comply with such a duty. It is crucial here to draw a distinction between duties enforceable by force and duties not enforceable by force, a distinction that needs to be recognized by any plausible moral theory: while we have a duty to be loyal to our spouse, the breach of which makes us liable to blame, most people believe that it is not permissible for anyone, including the spouse, to enforce such a duty by force.⁶⁸ If an offender’s duty to undertake penance is an enforceable duty, then it may be true that the only adequate way to mark its existence is to enforce it. But Bennett has not offered any defence for why the penance duty is an enforceable one. In fact, the case of teachers, used by Bennett himself, strongly suggests that the duty to undertake penance is not an enforceable one: the negligent teacher can choose not to teach free extra classes. Of course, the teacher’s association may revoke his license in such a case, but the crucial point is that we must leave the choice to undertake penance or not to him.⁶⁹ But, if the penance duty is not an enforceable duty, then it is unclear why the only adequate way of marking the existence of such a duty is to enforce it by imposing compulsory penance on the offender.

The above explains why I believe Bennett has not justified the claim that punishment is the uniquely adequate symbolic response to a wrongdoing. On this particular issue, perhaps a symbolic account of punishment that focuses on conventional symbols (let us call its advocates conventionalists) will fare better than Bennett’s account, which focuses on non-conventional symbols. Conventionalists may defend the unique symbolic adequacy of punishment by pointing to an empirical survey of what people actually take (reasonably or not) to be adequate symbols, an argumentative route that is unavailable to Bennett.⁷⁰ (Though a conventionalist will face other problems which are to my mind fatal: for example, it is unclear to me, first, why the fact that an act would violate a convention per se provides a reason against performing it and second, even assuming that there is a non-consequentialist reason for following a convention, that such a reason can legitimize deprivation of liberty).

5.3 Permissibility and Symbolic Adequacy

But, even if punishment were the only symbolically adequate response to a wrongdoing, why does that legitimize the coercive imposition of burdens that punishment involves? One of the objections against the communicative account of punishment I raised above is that, even if punishment is necessary to communicate a proper message, it is unclear why it is thereby legitimized. A similar challenge can be raised against an account of punishment that relies on symbolic expression: even if punishment were the only symbolically adequate response to a wrongdoing, why is punishment thereby legitimized?⁷¹

In fact, the acts that Bennett offers as paradigmatic examples of those justified by their symbolic adequacy (listed in Sect. 5.1 above) seem to be precisely the acts that we are not entitled to deprive other people of their liberty in order to perform. Suppose I cannot afford the only symbolically adequate gift I could give my wife unless I kidnap a rich man. Clearly, this would not turn kidnapping a rich man, which is otherwise an impermissible act, into a permissible act. Similar things can be said about mourning, etc. Why is punishment different?⁷² How does the fact that punishment is the only symbolically adequate response to a wrongdoing make an instance of deprivation of liberty, which is otherwise impermissible, permissible?⁷³

This objection, targeting the inference from the necessity of punishment as a means to expression to the permissibility of punishment, applies to most expressive accounts of punishment offered in the literature.⁷⁴ Disappointingly, I do not think many explicit responses to this objection have been made by expressive theorists in the literature. To Bennett’s credit, three arguments for establishing the permissibility of expressive punishment can be gathered from his writing.

First, Bennett acknowledges the challenge in a very recent paper, “Punishment as an Apology Ritual.” In this paper, he argues that if it is permissible for the state to set up a system of criminal law that is binding “in the relevant way,” i.e., the violation of which would be marked by a coercive condemnation, then the permissibility of expressive punishment can be established.⁷⁵ But so far this argument, which can be called the permissibility of setting up criminal law argument, has not advanced the debate. For, someone who rejects the permissibility of expressive punishment would also reject the idea that we should have a system of criminal law that is binding in the way suggested by Bennett.⁷⁶ Bennett acknowledges that more details need to be filled into justify why we should have such a system in the first place, but he does not provide such details in the paper.⁷⁷

Second, Bennett also discusses the relationship between symbolic expression and the permissibility of coercion in “Expressive Punishment and Political Authority,” another paper published after The Apology Ritual. Bennett accepts that the fact that depriving a person of his liberty is necessary for a body to issue a symbolically adequate response does not per se legitimize the body in depriving that person of his liberty. Yet, Bennett seems to believe that the deprivation of liberty is legitimized if two additional conditions are satisfied. ⁷⁸ First, the body is one that has a special kind of authority over the person, which can be called a directional authority. All authorities, qua authorities, can create content-independent reasons for action for their subjects through issuing directives.⁷⁹ But only directional authorities can create reasons that are directional, i.e., reasons that are owed to them, violation of which wrongs them.⁸⁰ Second, the person must have violated a directive issued by the authority, which forms the basis of the symbolic response. These two conditions explain, for example, why we are not entitled to deprive a rich man of his liberty in order to buy a symbolically adequate gift: we are not directional authorities over him. Let us call this argument for why we are entitled to deprive an offender of his liberty in order to issue a symbolically adequate response the authority argument.

The authority argument consists of two main claims. First, the state is a directional authority over us. Second, if a subject has violated the directive of a directional authority, then the authority can legitimately deprive him of his liberty in order to make a symbolically adequate response. I do not think Bennett has established either of these claims.

Concerning the first claim of the authority argument, I think it is fair to say that Bennett has not really argued for it. His main concern in “Expressive Punishment and Political Authority” is not really to justify this claim; rather, he is interested in a different issue. Let me summarize his aim and his moves in that paper extremely briefly, since we will soon see that they are not really relevant for our purpose. While Bennett believes that a directional authority may legitimately deprive of liberty a subject who has violated its directive in order to issue a symbolically adequate condemnation, he thinks that, in order to express the kind of condemnation that he is interested in, that is not enough; in addition, the state has to be a special kind of directional authority, namely, an adjudicative authority.⁸¹ He is worried that the concept of adjudicative authority seems to presuppose authority by expertise, which is inconsistent with political equality.⁸² Bennett’s main aim in the paper is to argue that the idea of adjudicative authority does not actually presuppose authority by expertise.⁸³ But, of course, accepting that the idea of adjudicative authority is consistent with political equality is compatible with rejecting that the state is a directional authority over us; and Bennett himself acknowledges that he has not done much to establish the latter claim.⁸⁴

Moreover, I think the second claim of the authority argument ought to be rejected, for two reasons. The first reason is that the idea of directional authority has two elements, direction and authority, and neither seems helpful in legitimizing the deprivation of liberty as a means of expressing a symbolically adequate condemnation. The breach of a directional duty towards a person by another, even if the duty is a content-independent one, does not obviously justify the former in depriving the latter of his liberty in order to issue a symbolically adequate condemnation. Suppose I breached my promise to you, which involves a violation of a content-independent duty I have towards you. It is hard to see why someone who is not already a retributivist would accept that this per se makes it permissible for you to deprive me of my liberty in order to express condemnation. Also, the idea of authority is unhelpful, as the question of whether a body has authority over a person and the question of whether it is permissible for the body to deprive that person of his liberty are concerned with quite different categories in the Hohfeldian scheme. Whether a body is an authority is a question about whether it has the normative power to affect the reasons applying to its subjects; but whether it is permissible for the authority to deprive a subject of his liberty is a question about what reasons apply to the body itself. ⁸⁵ So, it is hard for me to see why, if it is otherwise impermissible for a body to coerce a person, the fact that the body is an authority over the person would automatically turn this into a permissible act. Of course, one may argue that, while the ideas of direction and authority, considered individually, cannot work, the combination of the two concepts can do the work; but I am unable to see how this can be done.

The second reason for rejecting the second claim of the authority argument comes from considering an example given by Bennett himself, which concerns disciplinary actions taken by an academic department (which Bennett assumes to be a directional authority in the relevant domain; I will agree with him for the sake of argument) against its students.⁸⁶ Bennett seems to believe that this case supports state punishment because they are similar; but I think they are dissimilar in a relevant sense. I have already touched on a similar problem above in Sect. 5.2: an academic department cannot deprive a student of his liberty by imposing a compulsory sanction: academic sanctions are conditional in the sense that the student can choose not to comply by simply withdrawing from the course, while punishment is always compulsory. ⁸⁷ What the academic department example shows, at most, is that it is permissible for the authority of an institution to take away goods granted by the institution, such as the good of studying in the department, from its subjects. But the example does not show that it is permissible for the authority to deprive its subjects of other goods, such as their right to liberty; in fact, the example suggests otherwise. So much more needs to be said to explain why it is permissible for the authority of a political institution to take away not only the goods granted by the political institution, but also natural fundamental rights such as the right to liberty.

The third argument for claiming that it is permissible to deprive an offender of his liberty in order to issue a symbolically adequate condemnation does not appeal to the idea of authority; rather, it appeals to the Hegelian idea, endorsed by Bennett, that offenders have a “right to be punished.”⁸⁸ (I will accordingly call the argument the right to be punished argument.) Before I go on, let me observe that an offender’s right to be punished, strictly speaking, only entails that there is a duty to punish him but not that it is permissible to punish him: it is possible to have a duty to do something that is impermissible. Moral dilemmas provide an example of such. Bennett does not explicitly make use of the Hegelian idea to argue for the permissibility of punishment; so the right to be punished argument below may not be one that is endorsed by Bennett. That said, the issue of whether it is permissible for the state to punish and the issue of whether the state has a duty to punish are arguably connected, so it would be unfair to ignore Bennett’s discussion of the latter in exploring the former.

The argument runs as follows: Bennett believes that an offender has a right to be treated in a way that is symbolically adequate to his status as a qualified member of the moral community (in short, a responsible moral agent). Now, if punishment, as we are currently assuming, is the only symbolically adequate response to wrongdoings by responsible agents, then it may be thought that if we do not punish an offender then we fail to act in a way that does symbolic justice to his status as a responsible agent. This explains why Bennett believes that offenders have a right to be punished.

While Bennett does not explicitly make use of the idea of the right to be treated in a way that is symbolically adequate to one’s status as a responsible agent in the following manner, it may be argued that such a right is more important to a person than his right to liberty.⁸⁹ This point is significant because permissibility seems to be related to the idea of complaint: that no one can have a complaint towards an act seems to be sufficient for the act to be permissible.⁹⁰ If an offender’s right to be treated in a way that is symbolically adequate to his status as a responsible agent is more important to him than his right to liberty, then it may seem that he can have no complaint about the former right being respected over the latter when the two are in conflict. This Hegelian story also explains why, in contrast, we cannot legitimately use force on a rich man in order to buy a symbolically adequate gift: we do not compensate the deprivation of his liberty by respecting a more important right of his.

In response, I believe that the right to be punished argument, while interesting, leaves three questions unanswered. First, since a criminal wrongdoing consists of not one but two matters, namely, a wrong and a responsible agent, it is unclear how we move from the premise that punishment is the only symbolically adequate response to a wrongdoing by a responsible agent to the conclusion that if we do not punish, then we fail to respond to the responsible status of the offender in a symbolically adequate way. Why can’t the omission of punishment be seen as a failure to respond in a symbolically adequate way to the gravity of the wrong rather than a failure to respond in a symbolically adequate way to the responsible status of the offender? If this interpretation is available, then it seems to me that the offender might have cause for complaint if we take away his right to liberty rather than adopting a course (i.e., not punishing) that is not obviously a violation of his right to be treated symbolically as a responsible agent.

Second, even assuming that punishing the offender is the only way to do symbolic justice to his status as a responsible agent, it is unclear why the right to have his status as a responsible agent respected in a symbolic manner is more important to him than his right to liberty. There are two parts to this worry. First, given that having his status as a responsible agent respected symbolically does not enhance the offender’s welfare, it is unclear why doing so is of any importance to him. Second, even assuming that having his status as a responsible agent respected symbolically is of some importance to him, why is this more important to him than his right to liberty? This question is a legitimate one once we notice that the latter right is not derived from the former: children have a right to liberty even if they are not responsible agents; it is illegitimate to send them to prison for no reason.

Third, even assuming that the right to be treated in a way that is symbolically adequate to one’s status as a responsible agent is, objectively, more important to a person than his right to liberty, most offenders, I suspect, would prefer the latter; and it is paternalistic to claim that an offender can have no complaint if we ignore his actual wish in deciding which of his two incompatible rights should be satisfied. This question is particularly acute for Bennett, who wishes to remain within the tradition of liberal political thought;⁹¹ the idea of unwaivable rights, which Bennett accepts,⁹² seems particularly suited, to paraphrase Isaiah Berlin, to an illiberal regime where citizens are coerced in the name of respecting their rights.⁹³

6 Conclusion

In this paper, I have critically assessed Bennett’s justifications for retributive punishment based on the idea of expression. I first distinguished between three senses of expression: communicative, motivational, and symbolic. I then argued that the ideas of communicative expression and motivational expression cannot help in justifying punishment; neither does Bennett ultimately rest his case on these two senses of expression. Against the justification of punishment based on symbolic expression, I first argued that Bennett has not established that punishment is the only symbolically adequate response to a wrongdoing by considering and rejecting two arguments for the symbolic necessity of punishment, the withdrawal by others argument and the withdrawal by self argument. I then argued that Bennett owes us an argument for why punishment is permissible, assuming that it is the only symbolically adequate response to a wrongdoing. I considered three possible arguments, the permissibility of setting up criminal law argument, the authority argument, and the right to be punished argument, all of which I found wanting.