2. The Granularity Challenge

Many philosophers working on moral responsibility accept what I'll call

Strawson's Thesis posits an extensional equivalence between facts on either side of the biconditional. One major dispute (or, in Shoemaker's terminology, a “faultline”) that has developed between so-called “response-independent” and “response-dependent” theories of moral responsibility is which side of the biconditional in Strawson's Thesis has explanatory priority (see Shoemaker, Reference Shoemaker2020, pp. 217–221). Response-independence about responsibility is popular among compatibilists in the canonical debate about free will and determinism. Advocates of this view argue that the left side of the biconditional has explanatory priority: Someone is an appropriate target of reactive attitudes if and only if, and because, they are morally responsible.Footnote ² To explain the left side of the biconditional in Strawson's Thesis, then, they need to provide a unified theory of the substantial conditions for being morally responsible. For many classical compatibilists, for instance, someone is an appropriate target of reactive attitudes if and only if, and because, their wrong action happened voluntarily, knowingly, and it was under their control.

Over the past 30 years, several commentators have argued for reversing the explanatory direction in Strawson's Thesis: Someone is morally responsible if and only if, and because, they are an appropriate target of reactive attitudes.Footnote ³ This general view is often called “response-dependence” about responsibility. Advocates of it take as their blueprint Strawson's (1962/Reference Strawson and Watson2003) argument for compatibilism. Central to his descriptive account of human moral psychology is two “commonplaces” about us as social beings.Footnote ⁴ First, he emphasizes how much we care about whether people's actions reflect attitudes of goodwill or disregard towards us and others. Second, this general concern is manifest in demands or expectations for some degree of goodwill or regard. These demands or expectations are what reactive attitudes express. “The making of the demand,” Strawson writes, “is the proneness to such attitudes” (Strawson, 1962/Reference Strawson and Watson2003, p. 90).

Strawson's account of moral psychology motivates two of his arguments for compatibilism. First, totally suppressing our reactive attitudes is psychologically impossible, since the reactive attitudes are so integral to human nature. Second, even if we could totally suppress them, doing so would be practically irrational due to the “gains and losses to human life” (Strawson, 1962/Reference Strawson and Watson2003, p. 83). This is why Strawson says his contemporaries over-intellectualize conceptual issues of the traditional free will debate in one way or another. As he writes: “The existence of the general framework of [reactive] attitudes itself is something we are given with the fact of human society. As a whole, it neither calls for, nor permits, an external ‘rational’ justification” (Strawson, 1962/Reference Strawson and Watson2003, p. 91). Some theoretical discovery like the truth or falsity of determinism neither could nor should undermine or justify our responsibility practices as a whole. Our practices aren't susceptible to that kind of wholesale rejection or justification because they are so integral to human sociality itself.

However, Strawson never provides an explicit argument for response-dependence about responsibility. He simply offers this descriptive account of human moral psychology. Thus, Shoemaker (Reference Shoemaker2017, Reference Shoemaker2020, Reference Shoemaker, Nelkin and Pereboom2022) has recently developed a positive argument for response-dependence. His “burden-shifting” strategy begins with an analogy between two views:

The corresponding biconditional here is: Something is funny if and only if it merits amusement. Dispositional views say something is funny if and only if, and because, people are disposed to be amused by it under standard conditions. Since this view is implausible for several reasons (see Shoemaker, Reference Shoemaker2017, p. 485), another option is response-independent theories. These theories give the left side of the biconditional explanatory priority. The goal is to develop an account of the various properties that the funny consists in without making any constitutive reference to our amused responses. According to two leading theories, for example, something merits amusement if and only if, and because, it either involves an incongruity between expectations and experience (Clark, Reference Clark and Morreall1987), or the benign violation of some norm (McGraw & Warren, Reference McGraw and Warren2010).

Fitting Response-Dependence about the Funny says that merited amusement has explanatory priority over the property of being funny. What makes something funny is somehow a function of what merits our amusement. Yet some things do or don't merit amusement, regardless of our actual dispositions to be amused or not. To capture this normative dimension left out by the dispositional view, “merit” is glossed here as the relation of fit. Roughly speaking, fittingness is a name for “the relation in which a response stands to a feature of the world when that feature merits, or is worthy of, that response” (Howard & Rowland, Reference Howard, Rowland, Howard and Rowland2022, p. 1). It is the relation in which fear stands to the fearsome, shame stands to the shameful, admiration stands to the admirable, and so on. Many advocates of “fitting-attitude” theories of value hold views consistent with Fitting Response-Dependence about the Funny. Facts about a joke being funny, they argue, are always fully grounded in facts that make amusement at that joke a merited response. And similar remarks are made about other emotions, such as fear, shame, admiration, and so on (see Howard, Reference Howard, Zalta and Nodelman2023a).

By way of analogy, then, Shoemaker argues that we should analyze merited anger as the claim that anger is fitting.Footnote ⁶ This is the view he calls Fitting Response-Dependence about the Blameworthy. What determines the contours of the blameworthy as an evaluative category is that our properly functioning sense of responsibility makes certain objective properties but not others the anger-makers — the properties it's appropriate to respond to with anger. Control, knowledge, voluntariness, and quality of will are these anger-makers because they are the properties that tend to trigger our (properly functioning) sense of blaming anger.

But it is unclear how Shoemaker understands the relationship between fittingness and correctness. There are two possibilities: Fittingness either consists in attitudes satisfying standards or norms of correctness that are internal to those attitudes, or it is a matter of attitudes accurately representing their objects (see Howard & Rowland, Reference Howard, Rowland, Howard and Rowland2022, pp. 12–13). Gideon Rosen (Reference Rosen, Clarke, McKenna and Smith2015) calls the latter the “alethic view.” Fitting fear, for example, accurately represents its object as being fearsome. Wang attributes the alethic view to Shoemaker. If this is right, moral anger would be fitting only when it accurately tracks properties that we tend to respond to with fitting anger. However, in a more recent article, Shoemaker appears to hold the former view. As he writes, “the relation between holding and being responsible is one of ‘fittingness’ — adhering to ‘distinctive norms for attitudes of that kind’” (Shoemaker, Reference Shoemaker, Nelkin and Pereboom2022, p. 309). Indeed, the first account of fittingness helps explain why attempts at wholesale justification or rejection of our responsibility practices appeal to the “wrong kind” of reasons.Footnote ⁷ And the alethic view might face its own challenges.Footnote ⁸ However, I won't take a stand on these issues here.

In a recent article, Wang (Reference Wang2022) rejects the claim that facts about moral responsibility are always fully grounded in facts about the appropriateness of reactive attitudes. Here is his argument for what he calls the “Granularity Challenge”:

1. (1) Facts about moral responsibility are fine-grained: We can be partially morally responsible to greater or lesser degrees.

2. (2) Facts about the appropriateness of reactive attitudes are coarse-grained: They don't have the granularity to register small differences in normativity relevant features of moral agents.

3. (3) (1) and (2) are true when the standard of granularity is held fixed.

4. (4) Conclusion: Response-dependence about responsibility is false. (Wang, Reference Wang2022, p. 281)

The motivation for (1) begins with two desiderata about blame: We can (a) be morally blameworthy to different degrees, depending on how much blame we deserve, and (b) someone can deserve more or less blame, even when we hold fixed the degree to which their action is morally wrong (Wang, Reference Wang2022, p. 278).Footnote ⁹ The truth of (a) and (b) is easily explained by (1). But it is unclear how (a) and (b) could be true when someone is either morally responsible or they aren't. Thus, (a) and (b) are inconsistent with the binary nature of the biconditional in Strawson's Thesis.

Wang's case for (2) begins with the following examples (Wang, Reference Wang2022, p. 276–279). Heru is slightly less competent than Henry, and they both steal from Ann, so we can hold fixed the wrongness of their actions. We want to say that Heru is slightly less responsible than Henry, given their comparatively similar, though slightly different, capacities for moral competence. However, this normatively relevant difference between them cannot be captured by differences in reactive attitudes. Due to the limits of human psychology, Ann's emotional states don't have the granularity to register this slight difference in moral competence between Heru and Henry. Now consider the difference between an adult who is fully reasons-responsive and someone who is less than fully reasons-responsive (such as young children or the mentally impaired). Other things being equal, we would resent the former more than the latter for the same wrong acts. Thus, reactive attitudes can only register different degrees of moral responsibility in the latter types of cases (at the “coarse-grained” level) but not in the former types of cases (at the “fine-grained” level). This is illustrated by what Wang calls the Thesis of Limited Information Registration: “human emotions […] only track potentially responsibility-relevant factors, such as moral competence [or reasons-responsiveness, or quality of will], in a fairly coarse-grained manner” (Wang, Reference Wang2022, p. 276). Response-dependence is much less attractive, then, unless patterns in reactive attitudes can reflect how these substantial conditions affect degrees of responsibility at both levels. Finally, (3) is required to explain this comparative mismatch between different measures of granularity.

Thus, response-dependent theories of responsibility must somehow be revised to pass this granularity test. In the remainder of this article, I propose one way of doing so.

3. How to Half-Suspend Reactive Attitudes

Above I mentioned two of Strawson's arguments for compatibilism. His third argument, often called the “argument from exculpation,” rests on a distinction between two kinds of circumstances where we tend to withdraw or suspend our reactive attitudes: excuses and exemptions. Section 2 concluded with an example of the latter. Exemptions happen either when someone acts under abnormal circumstances, or when someone's own psychological abnormality renders them incapable of participating in ordinary interpersonal relationships. Excuses happen when we find out that there was no degree of ill will or disregard on the agent's part. We come to find out that someone was pushed, that it was an accident, or that they were unaware of what they were doing. Excuses allow us to continue viewing the agent from the participant stance as an appropriate target of our blaming and praising reactive attitudes. Exemptions involve adopting what Strawson calls the “objective” stance by either temporarily or more permanently viewing the agent as an inappropriate target of our reactive attitudes.

Parental and therapeutic relationships are oft-cited examples of the second subgroup of exemptions used to draw a comparison between objective and participant stances. But there is a widespread tendency of emphasizing Strawson's remarks about how these two points of view tend to be profoundly opposed. In Section IV of “Freedom and Resentment,” Strawson says he “must deal here in crude dichotomies and ignore the ever-interesting and ever-illuminating varieties of case” (Strawson, 1962/Reference Strawson and Watson2003, p. 79). Indeed, in a recent article Neil Campbell and I (Reference Campbell and Carty2024) argue that the tendency to present the objective and participant stances as necessarily opposed fails to heed Strawson's own warning that his crude comparison between them is “as grossly crude as it is central” (Strawson, 1962/Reference Strawson and Watson2003, p. 88). Later on, in Section V, Strawson takes pains to mitigate these crude comparisons. He does so by describing these ever-interesting varieties of case in terms of a half-suspension of reactive attitudes, rather than a thoroughgoing retreat into objectivity of attitude. He also clearly acknowledges cases where we can occupy the participant and objective stances simultaneously by straddling both of them.

Interestingly, Strawson's two examples, which he reintroduces to make these observations, are parental and therapeutic relationships. Describing parental relationships, he writes:

When describing therapeutic relationships, he also mentions similar phenomena in terms of a “half-suspension” of reactive attitudes:

My goal isn't to develop an account of the half-suspension of reactive attitudes. However, there are two important insights that I will highlight here.Footnote ¹⁰ The first is that Strawson appears to think therapeutic and parental relationships involve gradually progressing away from some degree of objectivity towards fully participant or engaged attitudes. Neither young children nor some patients in therapy are viewed as entirely appropriate targets of the full weight and range of reactive attitudes. Yet the way parents treat maturing children, or the way therapists treat their patients, isn't completely objective either.

But, while parenting and therapy are both paradigm examples of half-suspension, Strawson clearly introduces them as importantly different. The straddling of attitudes in parental relationships involves a gradual progression from complete objectivity towards wholly participant attitudes. In therapeutic relationships, however, the straddling of attitudes is more localized. While a therapist's attitude is objective, their patient must be treated as a responsible agent who, with the proper guidance and cooperation, can modify their own behaviour. Jonathan Bennett (Reference Bennett and van Straaten1980) makes similar observations in his negative formulation of the reactive attitudes. He contrasts them with a form of teleological inquiry — an effort to understand how someone works — which is associated with the objective stance. Reactive attitudes are ways of viewing others that do not involve teleological inquiry. However, Bennett confesses to fail to see any conflict between teleological inquiry and reactive attitudes. For example, he says that therapists are engaged in teleological inquiry and yet remain personally engaged with their patient because they “treat […] [the] patient as a person — a person who needs help” (Bennett, Reference Bennett and van Straaten1980, p. 35).Footnote ¹¹

Children, by contrast, aren't morally responsible at all when they are born. But if we were to start holding very young children or the severely mentally ill responsible, then it appears they would be fully responsible. Some suggest that this leads to a reductio for Strawson's theory. It is more plausible, they argue, that our practices would somehow be mistaken.Footnote ¹² According to Audun Benjamin Bengston (Reference Bengston2019) and Daniel Coren (Reference Coren2023), however, there is a more plausible view of how children are held responsible. As Coren puts it, “moral agents […] gradually emerge from inappropriate objects of resentment and the full range of reactive attitudes into, eventually, appropriate objects of those attitudes” (Coren, Reference Coren2023, p. 57, my emphasis). From the time they are born, young children gradually emerge as partially morally responsible agents if and only if, and because, they gradually emerge as increasingly appropriate targets of reactive attitudes.Footnote ¹³ This is why a parent's “rehearsals” of holding them responsible must “insensibly modulate” from being wholly objective towards “true performances” involving the full weight and range of reactive attitudes.

The second important insight is that examples of half-suspension aren't limited to parental and therapeutic relationships. Indeed, after emphasizing the variety of interpersonal relationships we can share, Strawson says that the “range and intensity of our reactive attitudes towards goodwill, its absence or its opposite vary no less widely” (Strawson, 1962/Reference Strawson and Watson2003, pp. 76–77). It would be implausible to claim that all sorts of interpersonal relationships involve the same degree of emotional engagement, or lack of objectivity. Parental and therapeutic relationships are clearly two among many “ever-interesting and ever-illuminating varieties of case” Strawson wants to draw our attention to by initially comparing the objective and participant stances in an overly “crude” manner (Strawson, 1962/Reference Strawson and Watson2003, p. 79). It is more plausible that cases of half-suspension occupy various mid-points — or, as Strawson puts it, a “penumbral” territory (Strawson, 1962/Reference Strawson and Watson2003, p. 88) — between emotional responses that range from completely participant to completely objective.

4. Rescuing Response-Dependence

Drawing on these observations about the half-suspension of reactive attitudes, I'll now build my case for rejecting premise two of the Granularity Challenge. My argument proceeds as follows:

1. (1) Facts about moral responsibility are fine-grained: We can be partially morally responsible to greater or lesser degrees.

2. (2) Facts about our basic concerns for quality of will in different interpersonal relationships ground the facts about the “appropriateness” or “fittingness” of our reactive attitudes.

3. (3) Certain components of human emotions — like attention, bodily feeling, action tendencies, or (for reactive attitudes) basic concerns for quality of will — can come in fairly fine-grained degrees.

4. (4) Strawson's Degree Thesis: Someone is blameworthy (for some wrong act A) to degree D if and only if, and because, they are an appropriate target of reactive attitudes to degree D (regarding A).

5. (5) Corresponding degrees of intensity of reactive attitudes and of the basic concern that underlies them can be determined by the degree to which those reactive attitudes are half-suspended with an objective point of view.

6. (6) Conclusion: Human emotions are fine-grained enough to register small differences in normatively relevant features of moral agents, and parental relationships are an illustrative example of this.

I accept (1) for the same reasons as Wang. Additionally, I'll argue below that the biconditional Fitting Response-Dependence about Responsibility inherits from Strawson's Thesis should be modified into (4) in my argument. (2) is an uncontroversial premise, since this claim is widely accepted as Strawson's view. As Coren writes:

According to Strawson, for example, it would not follow from the truth of determinism that everyone is somehow incapable of participating in interpersonal relationships. In other words, determinism being true neither would, nor should, convince us to permanently exempt everyone by constantly viewing them objectively.Footnote ¹⁴

Wang draws on empirical evidence as indirect support for rejecting premise (3) in my argument. According to Lisa Barrett's (Reference Barrett2006) work on self-reports of emotions, our limited ability to describe our emotional experiences in precise terms indirectly motivates the claim that there are, at most, only coarse-grained differences between emotions. Indeed, as Wang acknowledges, whether a lack of fine-grained distinctions in self-reports of emotion is due to our limited vocabulary is a difficult question (Wang, Reference Wang2022, p. 278). But, even without a clear answer to that question, I still think that premise (3) enjoys indirect empirical support.

Laura Silva (Reference Silva2023) has recently defended a version of (3), though she does not extend its scope to reactive attitudes.Footnote ¹⁵ Silva argues that remarkable precision and differences in literary accounts of emotions suggest that we at least have a capacity to register fine-grained degrees in emotional experience with considerable detail. Fine-grained differences in emotions, she argues, consist in covarying fine-grained differences in their components — such as feeling, attention, representation, or action tendencies. To support this claim, she turns to the empirical work of Dominic A. Evans et al. (Reference Evans, Stempel, Vale and Branco2019). Their study shows that the properties of a threat (its size, speed, distance to safety, etc.) closely correlate with fine-grained differences in action tendencies for fear responses. Consider the contrast between instances of fear towards a large bear and a small yet venomous snake. If there were no such fine-grained differences, our fear responses would unreliably direct us to safety. So, the scale for degrees of intensity of emotions is more fine-grained than Wang suggests. Indeed, these observations are also compatible with the idea that, generally speaking, the fitting intensity of an emotion like fear, shame, or admiration covaries with how fearsome, how shameful, or how admirable its object is.Footnote ¹⁶

My case for extending (3) to reactive attitudes draws on the idea that there can be fine-grained degrees of expressions of basic concern. Coren (Reference Coren2023) has recently defended this claim in his account of parenting. He agrees that the biconditional in Strawson's Thesis should be modified into something like (4) Strawson's Degree Thesis. As Coren writes, “Strawson argues that the degree to which an agent is responsible must match the degree to which she is an appropriate target of blaming or praising attitudes” (Coren, Reference Coren2023, p. 59). This modified version of the biconditional, he argues, does a better job than Strawson's Thesis of explaining (3). One clear example is how children gradually emerge as morally responsible in the context of parental relationships. Over time, there are many fine-grained adjustments in exactly how much we are concerned about a child's quality of will towards us and others in the moral community. As Coren puts it:

As we saw above, the different degrees of intensity of reactive attitudes vary no less widely than the many interpersonal relationships we share. This provides further motivation for the claim that there are fine-grained degrees of basic concerns underlying those reactive attitudes. Having modified Strawson's Thesis into (4) Strawson's Degree Thesis, then, the resulting account does more justice to the Strawsonian compatibilist framework: Facts about the appropriateness of our reactive attitudes are grounded in facts about our basic concerns for quality of will, and furthermore, the former facts are what ground facts about moral responsibility. This explains why children gradually emerge as partially morally responsible agents if and only if, and because, they gradually emerge as increasingly appropriate targets of reactive attitudes and practices that express our basic concerns for their quality of will.

Finally, when parents hold children partially responsible by half-suspending their reactive attitudes, this involves fine-grained degrees of basic concern. Furthermore, the fitting intensity of reactive attitudes can correspond with the intensity of the fine-grained degrees of basic concern underlying those attitudes. Premise (5) explains these possibilities. The corresponding intensities can be explained by the degree to which those reactive attitudes and our basic concerns underling them are half-suspended with the objective stance. If this is right, then these cases are prima facie counterexamples for premise two of the Granularity Challenge. One might worry that fine-grained differences in action tendencies, basic concern, or other components of emotion don't always lead to corresponding intensities of reactive attitudes. But I haven't defended that claim. Instead, I've argued for the weaker claim that they sometimes can. Footnote ¹⁷

5. Conclusion

According to the Granularity Challenge, if we can be partially morally responsible to greater or lesser degrees, and if reactive attitudes are too coarse-grained to register small differences in normatively significant features of agents, then response-dependence about moral responsibility is false. My goal in this article was to build a case for rejecting premise two of the Granularity Challenge. Human emotions are fine-grained enough to register small differences in normatively relevant features of agents. One illustrative example of this, I've argued, is how children gradually emerge as partially morally responsible agents in the context of parenting. To be clear, I haven't suggested that Shoemaker's argument is the most promising way of developing response-dependence as a metanormative theory of responsibility.Footnote ¹⁸ Nonetheless, I hope this article inspires further discussion about how to rescue these theories from the Granularity Challenge.