Is Anger a Hostile Emotion?

In this article I argue that characterizations of anger as a hostile emotion may be mistaken. My project is empirically informed and is partly descriptive, partly diagnostic. It is descriptive in that I am concerned with what anger is, and how it tends to manifest, rather than with what anger should be or how moral anger is manifested. The orthodox view on anger takes it to be, descriptively, an emotion that aims for retribution. This view fits well with anger being a hostile emotion, as retribution is punitive. I will argue that a different view of anger deserves our attention. On this alternative view, anger aims for recognition of harms done, rather than for the punishment of those who have committed them. I argue that we have reason to favour a strong view that excludes retribution from anger’s main aims. In addition, I offer a diagnosis of the reasons that led the retributive view of anger to become, and remain, orthodoxy. This diagnosis provides indirect reason to give my descriptive proposal serious consideration, for it highlights that the orthodox view has dominated folk and philosophical conceptions of anger for reasons that do not speak in favour of the view’s veracity. The view that anger is a hostile emotion will therefore emerge as in need of serious scrutiny.


Introduction
In this paper I explore the prospects of a controversial thesis regarding anger: I argue that anger has been unduly characterized as a hostile emotion in philosophy. Anger is taken to be a paradigmatically hostile emotion, at least in part, because it is traditionally characterized as a retributive emotion. The orthodox view on anger takes desires for retribution to be definitional of anger. I follow others in using the terms retribution, payback and revenge interchangeably throughout (Pettigrove 2012;Nussbaum 2016). Although retribution can arguably be sought in non-hostile ways, if anger is inherently retributive it seems fitting that anger be a paradigmatic example of a hostile emotion, as agents will be disposed towards punitive actions.
I will argue that we have reason to take seriously the view that anger is essentially recognitional, that is, that it aims for recognition of harms done, as opposed to aiming for the punishment of its targets, and that punitive aims are either uncommon in anger, or entirely instrumental towards attaining recognitional goals. I will rely on empirical work from experimental and social psychology as well as work from the history of emotions to inform my argument. My project in this paper is primarily descriptive, in that it investigates what anger is and what its behavioural effects are, rather than what anger should be, or how moral anger should unfold. My project is also partly diagnostic, in that I will uncover reasons for why the view that anger is inherently retributive has remained the orthodox view on the emotion. These, we will see, will be reasons of the wrong sort in that they do not count in favour of the view's veracity. The outcome will be that we have strong reasons to give a view of anger that excludes retribution from the emotion's main aims serious consideration. Such an argument is, to my knowledge, currently lacking in the literature.
If retribution is excluded from anger's main aims, then characterizations of anger as a hostile emotion are likely mistaken. I will follow a number of conventions in the philosophy of emotion literature regarding what I mean by anger. As with most contemporary philosophical discussions of the emotions, I focus on occurrent states of anger that have distinctive phenomenology, rather than dispositional states. I focus also only on cases of anger that have intentional objects. It is common to distinguish between the formal and particular objects of emotions. 1 Formal objects are typically construed as evaluative properties that emotions of the same type attribute their particular objects, offence for anger and danger for fear, for example. 2 All instances of anger share a formal object, as this is partly what individuates the emotion type. 3 Anger's particular objects are objects that anger is felt towards. I will be assuming that anger, like most emotions, admits of a variety of particular objects, which include individuals, groups, institutions and states of affair. I use the terms 1 On the formal objects of emotions see Kenny (1963) and Teroni (2007). 2 Although the formal object of anger is typically taken to be 'offence', a broad understanding of these terms is typically employed in order to account for everyday cases of anger. Someone cutting in front of you in a queue may not strike you as a personal offence but may still make you angry because such an action is an offence in the norm-violation sense. The formal object of anger is therefore this broad sense of 'offence' that includes wrongs and unfair actions or states of affair, whether or not the angry individual is personally insulted. 3 The formal objects of emotions are typically also thought to supply the correctness conditions of emotions, such that anger is said to be appropriate or justified when its formal object is instantiated in or by the particular object anger is felt towards, i.e. anger is appropriate when an offence has occurred. It is important to note that assessments of appropriateness often equivocate over at least two distinct types of normative assessment (D'Arms and Jacobson 2000). An emotion can be appropriate in the sense that it is correct or fitting, with respect to an evaluative state of affairs, but the very same emotion can be inappropriate in a moral sense (for example, it might be inappropriate to laugh at a funeral, even though your friend's whispered comment was funny). Here I am concerned with appropriateness in the fittingness, or correctness, sense. Emotions are appropriate in this sense when they represent an evaluative state of affairs in a manner that matches how things really stand in the world.
'object' and 'target' interchangeably to refer to the entity anger is felt towards. In line with much philosophical writing on anger (Bell 2009;Pettigrove 2012;Cogley 2014;Nussbaum 2016;Srinivasan 2018), I take the term 'anger' to encompass a range of related affective phenomena, such as rage, resentment and indignation. Lastly, in line with common thinking, I take anger to be typically triggered by offences, injustices, or other disruptions to one's goals that one perceives as unfair.
I will argue that we should take seriously the view that the orthodox account of anger is not only guilty of oversimplifying anger, but that it gets the emotion's nature seriously wrong. I will begin by outlining the orthodox view on anger as inherently retributive in section 2. I will then introduce recognition as a separate desire that has been observed to be central to anger, in section 3. Despite mentions of recognitional goals in anger, systematic challenges to the orthodox view of anger in light of the centrality of this desire are lacking (Silva (2021a) is an exception). The rest of section 3 is dedicated to arguing that if there is a definitional desire to anger, it is an open possibility that this be the desire for recognition and not retribution. We will see that empirical work suggests that retribution is typically sought in scenarios where recognitional aims are impeded or have been exhausted, suggesting that recognition is primary in anger. Additionally, empirical work on the nature of retribution will suggest that purely retributive aims are neither necessary nor sufficient to make anger subside. This will strengthen the view that recognition, and not retribution, is anger's main aim. Note that the empirical work provides only indirect and suggestive evidence, but it fits nonetheless with a picture of anger that departs significantly from orthodoxy, and, I will argue, deserves serious consideration. With my proposal on the table, I will then move in section 5 onto diagnosing why the orthodox view has been so prevalent. This will serve to bolster how seriously we should take my proposal, for we will see that unfair contextual features may have historically led anger to be seen as retributive and hostile. This stands in line with Griffiths' (1997) insight cited in the epigraph above: that to presume our vernacular emotion concepts aptly describe the relevant emotional phenomena, is to disregard the role social factors, which are often unjust, may have played in bringing these folk understandings about. I conclude that, in light of my argument, anger may be mistakenly characterized as a hostile emotion.

The Orthodoxy on Anger
The Orthodox view of anger takes the emotion to be, at its core, retributive. This view can be traced to ancient Greek philosophy, in particular the writings of Aristotle. For Aristotle, anger is 'a desire accompanied by pain for conspicuous revenge caused by a perceived slight' (Rhet. 1378a31-33). 4 Anger conceptually involves a desire for returning pain (DA 403a31), and it ceases when the offender 'pays back for the offense; for revenge stops anger' (EN 1126a21-22). 5 This retributive view is evident in the later Roman writings of Seneca, anger's most famous critic, for whom anger 'consists entirely in aroused assault. Raging with an inhuman desire to inflict pain' (1.1). Seneca's anger is 'hungry for a vengeance that will bring down the avenger too' (1.1). Retributive conceptions of anger are alive in contemporary philosophy as well. Deonna and Teroni (2012), for example, take anger to involve 'feeling the body's readiness to act so as to retaliate' (81), and Pettigrove (2012) takes anger to 'involve the desire to lash out at its object or to see that object hurt' (358). For Nussbaum 'anger involves, conceptually, a wish for things to go badly, somehow, for the offender in a way that is envisaged, somehow, however vaguely, as a payback for the offense' (Nussbaum 2015: 46). In a similar vein, Ben-Ze'ev (2000) writes that 'the urge to attack is essential to anger, even if it is expressed in a nonstandard aggressive act' (384 emphasis my own). 6 Most recently, Callard (2020) takes anger untied to vengeance to be a 'philosopher's fiction' (10).
I take the retributive construal of anger to be the orthodox view on the emotion, and use the labels 'retributive' and 'orthodox view' interchangeably. Indeed, Nussbaum calls the retributive view the 'traditional' view of anger (Nussbaum 2015: 41). In psychology too, revenge or punition has often been taken to be a definitional component of anger (see Berkowitz and Harmon-Jones 2004;Frijda 1994). The retributive construal of anger is more often assumed, many times implicitly, than defended. The view is clearest in those who stipulate such a desire to be constitutive of anger, but is also evident in any view of anger that singles out a strong link to a desire for retribution. I take views of the latter sort to involve a causal commitment that anger prototypically triggers desires of this sort. On a causal reading, desires for retribution are not actual parts of anger, but are nonetheless the primary causal effect of the emotion. The orthodox view can therefore come in the stronger, constitutive, or the weaker, causal, form. Nussbaum (2015:46), Ben-Ze'ev (2000) and Callard (2020) make explicit that they take the punitive aim of anger to be constitutive of the emotion, rather than being merely causally related to it. The desire for the perpetrator's suffering becomes a conceptual, or essential, part of anger for them. Similarly, in Aristotle, a constitutive view is endorsed as anger is equated to the desire for revenge, accompanied by pain. In Tappolet (2016), a causal commitment seems clear, for she takes specific desires to be 'regular causal effect(s)' of emotions, rather than 'essential ingredients or parts of emotions' (66), and characterizes anger as bearing such a causal relation to desires for 'revenge or punishment' (72). In other cases, however, such as in Pettigrove (2012), it will be ambiguous whether the author endorses the constitutive or the causal claim regarding anger's retributive nature. I think a commitment to at least the weaker causal claim to be widespread in our scholarly, as well as every day, conception of anger. I will refer to both views under the label 'orthodox' or 'retributive' view of anger, as they both link anger strongly to one desire, a desire for retribution, and are both therefore in-keeping with a view that takes anger to be a hostile emotion.
What is this desire for retribution? Proponents of the retributive view of anger have not said much on what conception of desire they employ. I therefore make use of a thin and intuitive notion of desire throughout. I take desires to dispose one to act in ways that aim to achieve the desire's aim, and take desires to be typically satisfied when the actual state of affairs in the world matches the desire's aim. The orthodox view of anger, in stipulating a desire for retribution as the key conative component of the emotion, holds that retribution is anger's central goal or aim. It is clear that the desire for retribution is a desire for the offending party to suffer for what they have done. The centrality of pain is clear in Aristotle, for example, as the angry person is one that desires to inflict pain (Rhet. 1382a8), and in contemporary theories that see anger as involving a desire to lash out, retaliate, or wish for things to go badly for the target of anger, the emotion aims for some form of pain or suffering (Pettigrove 2012;Nussbaum 2016). The retributive view of anger fits well with anger being a hostile emotion, for if in anger we desire to make our targets suffer for what they have done, hostility will be a common causal effect of the emotion. Of course, anger need not always motivate hostile behaviour, but by bearing a strong link to desires for retribution the emotion has punitive, and therefore arguably hostile, aims. I therefore take the orthodox view to be a key motivation for the classification of anger as a hostile emotion.

Recognition
Despite orthodoxy, a distinct conative component has at times been proposed of anger: a desire for recognition of harms done. Srinivasan (2018) takes anger to often aim to get a perpetrator to recognize the pain she has caused another, the wrong she has done another, rather than to make the target of anger suffer. Similarly, Cogley (2014) writes that, in anger, an agent 'confronts the target of her anger in an attempt to bring the target's attention to her cause for anger. She then asks after or demands an explanation or justification' (211).
The view that anger is more concerned with issuing demands for respect and recognition, than punishing its targets, can be found also in Adam Smith who writes that anger's aim is not so much to cause its target to 'feel pain in his turn... as to make him sensible, that the person whom he injured did not deserve to be treated in that manner' (Smith 1976: 95-6). Similarly, Darwall (2013) takes the objects of anger to be 'also their addressees', anger comes 'with an implicit RSVP' on whether the object of anger acknowledges that a harm has been done (89). Angry attitudes, therefore, do 'not seek to diminish, humiliate, insult, or damage their objects. The acknowledgment they seek is of a reciprocal standing' (89). Cherry and Flanagan (2017) have listed 'recognition respect anger' as one of seven forms anger often takes and characterize it as aiming not for payback, but for recognition of selfworth. There is, therefore, a trend to see one of anger's main aims as recognitional, rather than punitive.
What exactly is the desire for recognition? I will take the desire for recognition to be a desire for epistemic changes in the targets of anger, such that they come to evaluate their actions or omissions as wrongful in some way, and additionally that this epistemic change be communicated to the harmed party. I take genuine apologies to be common cases where anger's recognitional aims are met. Genuine apologies give one reason to revise one's anger, not because the wrongs that triggered it have been excused or erased, but because one's desire for the wrong to be recognized as a wrong has been satisfied. 7 Note that it is a common view that apology is among the set of things that can cause anger to terminate (Hieronymi 2001;Callard 2018; Na'aman 2019). My claim is that this is so because apology satisfies recognitional aims in anger. This claim is only incompatible with accounts of apology that take apology to involve the lowering of the status of the apologizer, as such 'status-lowering' could be construed as a form of retribution (Bovens 2008). These are quite unpopular accounts of apology however (see Helmreich 2015). On most accounts, apology serves to communicate the acknowledgement of having committed a wrong (Hieronymi 2001) and commitment to treating the offended party better (Martin 2010;Hieronymi 2001), and is often explicitly contrasted to retribution (Hieronymi 2001;Callard 2018; Na'aman 2019). In any case it is clear that the conception of recognition invoked by those who think that it is a central desire in anger is a non-punitive conception. As seen in the citations above, recognition is often invoked in explicit contrast to inflicted suffering and payback. 8 It is a move open to orthodox theorists to try to argue that all cases where anger seems to have purely recognitional aims actually involve desires for psychological, subtle or symbolic forms of punition. Such a move risks extending the boundaries of what counts as retributive far beyond the term's normal usage however, and indeed beyond the term's explanatory utility (cases where anger at a friend motivates you to seek their company to talk things out would count as cases where retribution is sought, as would all cases of peaceful communicative forms anger-motivated political activism). 9 I will move forward with the assumption that recognitional aims are non-punitive aims, as this most faithfully reflects the intended meaning of the term in the work of those who have invoked it.
The appealing idea that anger sometimes has purely recognitional aims has not dislodged orthodox thinking on anger as distinctively retributive however. Those that take the desire for recognition to be characteristic of anger have not launched sustained attacks on the orthodox view in light of this, and rarely make clear how prevalent they take recognitional aims in anger to be. The view that anger is a hostile and punitive emotion remains dominant, and the intuition that anger involves genuine aims for recognition remains unaccounted for or outright denied by many contemporary philosophical accounts of the emotion (Nussbaum 2016;Callard 2020). For Callard recognitional anger is a 'fiction'. For Nussbaum (2015Nussbaum ( , 2016, there seems to be room for something called 'transition-anger' that is non-punitive. Nussbaum characterizes transition-anger as anger that is not retributive, and which focuses on 'brotherhood', 'justice', 'reconciliation and shared effort' instead, typically motivating constructive actions (Nussbaum 2015: 53-54). Nussbaum takes such anger to be a 'borderline case' that is 'rare and exceptional' and only present in individuals with superior 'selfdiscipline' (54). In other work, I have argued that far from being 'exceptional', nonpunitive recognitional desires are common and robust in anger. Anger often aims for its targets to recognize that a harm has been committed rather than aiming for the suffering 7 Where these reasons for anger are the initial wrongs that caused the anger (Callard 2018; Na'aman 2019), or the fact that the offender could harm the offended again in the future (Hieronymi 2001). On views of the latter sort, anger ceases when an apology is issued because the apology removes the current threat. 8 An exception is Christensen (2016), who proposes an Aristotelean account of anger where revenge involves punitive recognitional aims, such as causing experiences of remorse and regret in wrongdoers. 9 See Silva (2021a) for further consideration of the moves the orthodox theorist could make in response to arguments for the centrality of desires for recognition in anger. of wrongdoers. This is in-keeping with an attractive pluralist view where anger sometimes has retributive aims and other times recognitional aims (both aims may also be at play simultaneously) (see Silva 2021a). In this article I will argue for something stronger than a pluralist view, I will argue that we should take seriously the proposal that the orthodox view gets things exactly in reverse. That is, I will argue that there are good reasons to think that anger definitionally involves desires for recognition, and only exceptionally or instrumentally involves desires for retribution. I will rely on empirical and conceptual considerations to inform my argument. First, I will argue that phenomenological and empirical work supports a robust desire for recognition in anger. That this be the case doesn't itself challenge the construal of anger as a hostile emotion, for we can seek recognition in aggressive and destructive manners. We will see however, that empirical work suggests that hostile actions are typically pursued only when recognitional aims are either blocked or have been exhausted. 10 This gives us reason to think that non-punitive means of seeking recognition are primary in anger. We will see that empirical work suggests, further, that when retributive or hostile actions are pursued, they only satisfy the angry agent when punition is accompanied by important epistemic changes. Taken together, these considerations suggest that the desire central to anger is recognition as opposed to retribution.

Anger Robustly Aims for Recognition
Often, in anger, we wish the target would acknowledge the gravity of what they have done. In the first instance we might wish for a justification of the harm we have suffered, such as to potentially excuse it (I missed our dinner plans because I was ill, not because I don't care), but if none can be given, we wish for accountability, where the offender judges their own act, or omission, as wrong. We don't aim to harm the offender, in a literal or symbolic way, we want them to share our evaluation of the harm as unacceptable.
Imagine that you are angry at your best friend for not being there for you throughout your divorce. When you needed their support the most, your friend decided to go on a spontaneous three-month long holiday abroad. Your anger at your friend would be justified, but it is unlikely that you would wish to make them suffer. You would not want to make your friend suffer physically, or hope for things to go badly for them in the future. Nor would you seek to ensure their social exclusion, or defame their character. Your anger's goal does not seem to be that your friend suffer, but rather, that they understand what they have done. In so far as retributive actions might be taken against your friend, they seem to have the aim of making your friend understand that they have committed a wrong rather than having the final aim of making them suffer as payback. Any aim for retribution would therefore seem instrumental in relation to the desire for recognition. My point is not that one never feels vindictive anger towards loved ones, but rather that we should be cautious to attribute such vindictiveness to anger's nature.
I think desires for recognition extend beyond cases where one has an underlying interest in the wellbeing of the target of one's anger, as is the case in the friendship example above. Indeed, most everyday cases of anger can be seen as aimed at recognition. When you get angry at someone for cutting in front of you in line at the bank, you arguably want them to acknowledge that what they have done is not ok, rather than make them suffer for it. Similarly, if you are angry at a group or company for an unethical action, your central desire is plausibly for them to acknowledge what they have done, in a manner that indicates that they will change their ways. Empirical work stands in line with a robust desire for recognition in anger. First, self-reports on everyday anger show the emotion to motivate a far greater proportion of non-punitive actions than punitive ones (Averill 1983). Interpersonal harms often motivated angry agents to approach offenders for a discussion, and the targets of anger often offer apologies and social support that solidify relationships (Yoo et al. 2011). In a number of studies on collective action, anger was observed to significantly motivate communicative actions (such as petition writing and peaceful-protesting), as opposed to punitive actions (ranging from destructive actions such as property damage to merely inconvenient ones such as blocking roads or buildings) (Tausch et al. 2011;van Zomeren et al. 2004van Zomeren et al. , 2012. 11 In other studies, anger has been observed to motivate both punitive and non-punitive communicative actions (Halperin 2008), giving us at least reason to think recognition might be amongst anger's aims.
It is important to note that these experimental studies fit into a broader shift in how the psychological literature has construed anger and its effects. There has been a a gradual yet radical change that philosophers working on anger should heed. The early influential frustration-aggression hypothesis posited 'that the occurrence of aggressive behavior always presupposes the existence of frustration and, contrariwise, that the existence of frustration always leads to some form of aggression' (Dollard et al. 1939). The theory then posited a necessary and sufficient relation between goal frustration (a paradigmatic cause of anger) and aggression (Dennen 2005). Over the years the theory was modified to weaken and restrict this strong claim. Goal frustration was soon posited to motivate both aggressive and non-aggressive behaviour. By the late 70s, research provided mounting evidence against the initial tenant of the frustration-aggression hypothesis, such that Zillmann (1979) held that goal frustration would likely lead to the expression of a negative emotion such as anger, but would not, in and of itself, generally produce interpersonal aggression or hostility. The blockage of a goal would, instead, likely invoke behaviour aimed at terminating or redressing this blockage, and would only tend to be aggressive if the instrumental value of aggressive action exceeded that of non-aggressive alternatives (Zillmann 1979).
In line with this, views on the very value of aggression have shifted, from originally being seen as destructive, typically dysfunctional and not instrumentally valuable, aggressive actions are now understood through a functionalist lens, whereby they are means to tackle goal obstructions, offences and injustices and can often be very effective, as well as constructive, in addressing them (see Silva 2021b). In interpersonal negations for example, angry communications from a disadvantaged party prompts dominant subjects to compensate angry agents (van Kleef and Côté 2007), while 11 Which actions count as punitive or non-punitive is not set in stone. The empirical work employs an intuitive understanding of the distinction, where actions that do not straightforwardly aim to harm the targets of anger are considered non-punitive. Section 3.2 and 3.3. provide further evidence for the centrality of non-punitive desires in anger that go beyond what is canvassed in section 3.1. communications of anger from oppressed groups have been observed to provoke empathy and support, as opposed to retaliation, from dominant groups (de Vos et al. 2013(de Vos et al. , 2016Tagar et al. 2011). There is therefore evidence that anger, plausibly even when communicated aggressively, can have constructive instrumental effects in confronting goal frustrations such as offences and injustices. 12 A parallel shift has occurred in work on crowd behaviour. Initially thought of as inherently violent (Le Bon 1895), researchers now take violent crowd action in response to injustices to be far more contingent on specific background and circumstantial conditions (Drury et al. 2020). This highlights the importance of context in determining how anger unfolds. Indeed psychological research has long moved beyond categorical claims regarding anger, and is more interested in uncovering those contextual features that favour or disfavor aggressive vs non-aggressive action in the first place, as well as those contextual features that might lead to making even aggression instrumentally valuable. I have argued elsewhere that this fits well with a pluralist view of anger, where orthodox claims are abandoned, and the focus is on theorizing distinct desires in anger and the relevance of those conditions that favour each (Silva 2021a). I take this to be a long overdue shift in philosophical thinking that will adequately mirror the shift that psychological work has undergone. In what follows I will be concerned with a stronger thesis than that endorsed by a pluralist view of anger however. The pluralist takes anger to often involve a desire for retribution or punition, while holding that the emotion also often involves desires for non-punitive recognition of harms done. Here I will argue that we have good reason to go even further, and consider the prospects of a view of anger whereby retribution is excluded from the emotion's non-instrumental aims. 13 For now, I take myself to have outlined that there is much psychological evidence that fits well with a view that takes, as we saw many philosophers to, desires for recognition to be central to anger. The robust correlation between anger and the pursual of communicative actions fits well with such a view, as does the general shift in psychological thinking from claims that anger is inherently tied to aggression, to views where non-punitive behaviours are significantly correlated with the emotion, and where contextual features determine how anger unfolds. Similarly, evidence that anger communications invoke empathy and support in their targets is at least in keeping with a view where anger aims for epistemic changes in, as opposed to the suffering of, its targets.

Aggression in Nothing-to-Lose Scenarios
In the contemporary psychological literature then, acceptance of anger's behavioural pluripotency, i.e. that anger is significantly related to different types of actions, is widespread (Spring et al. 2018). The question becomes not whether anger triggers punitive or non-punitive actions but, rather, when and why anger displays its different motivational effects. Experimental work suggests that key factors moderate the effects of anger. Moderators are crucial to determining when certain effects hold. They are typically contextual variables that influence which effects are observed. Contextual 12 In previous work I have argued that this evidence gives us reason to abandon what I call an 'inefficacy claim' that is prevalent among those who condemn anger specifically in fights for social justice (Silva 2021b). 13 I thank an anonymous reviewer for prompting me to consider the broader shift in psychological thinking on anger and its effects. moderators are likely crucial to determining whether anger will motivate punitive actions or not. A key moderator over anger's behavioural effects seems to be the perceived changeability of the target of anger, that is, how likely one's actions are to bring about a change in their attitudes and actions. Many studies have observed that aggressive actions take place in what have been called 'nothing-to-lose' scenarios (Bandura 2000;Scheepers et al. 2006;Tausch et al. 2011). These are scenarios where agents judge themselves to have low ability to change the unfair situation. In collective action studies, destructive actions were predicted not by levels of anger but, in line with the nothing-to-lose hypothesis, when agents judged that other measures were unlikely to be effective at bringing about any change (Tausch et al. 2011). 14 These studies suggest that anger at groups and institutions robustly motivates constructive collective action and that it is only when communicative actions have systematically failed, or are blocked, that groups seem to pursue punitive actions. Perceptions of the targets of anger as unchangeable are characteristic of nothing-to-lose scenarios. Similar results have been observed in studies on interpersonal psychology, where anger is felt by one individual for another. Desires for vengeance were observed to be highest when agents believed that the targets of their anger (such as bullies) have an unchangeable character, again suggesting that retribution is pursued under conditions where change is perceived as improbable (Yeager et al. 2011). It may be only in cases where such recognition is impossible, or unlikely to be achieved by non-punitive means, that anger motivates hostile or retributive actions. We saw above that even modifications to the early frustration-aggression hypothesis highlighted that aggressive actions were typically only pursued when the value of alternative non-aggressive actions was perceived to be low (Zillmann 1979). Further evidence suggests that aggression is far more typical of anger that occurs in otherwise malevolent relationships, than ones that have cooperation as a background condition (Zeineddine et al. 2015;Parkinson et al. 2005;Tiedens and Leach 2004). Section 3 and 3.1 gave us reason to think that desires for recognition are robust desires in anger. This is compatible with a pluralist view that takes desires for retribution and recognition to both be central in anger (see Silva 2021a). Such a pluralist view would still challenge the orthodox view by making a desire other than retribution central to anger. Here we have seen reason to take seriously a stronger proposal, that is, the proposal that desires for recognition and retribution in anger do not stand on equal footing. Recognition might be the emotion's primary desire, where retribution is sought when recognition is blocked or hard to attain, such as in nothing-to-lose scenarios, 14 In the Tausch et al. 2011 study, contempt, as opposed to anger, was significantly related to the motivation of destructive actions. This fits with the result that low efficacy was significantly related to destructive actions as well, as contempt is thought to represent its objects as less changeable than anger. While anger is a response to wrongdoings, contempt is thought to involve evaluations of its targets as more globally 'bad' (Bell 2013). Anger evaluates actions or omissions as unjust or offensive, while contempt evaluates the target of those actions as bad or despicable. Anger and contempt will often co-occur of course, but contempt is likely to grow out of repeated occasions for anger as it signals a more global evaluation of its target as inherently bad. Contempt will therefore be more expected in cases where the objects of anger are perceived as unwilling to change or address injustices, as these will be one's where a global assessment of these objects as despicable or bad is apt. In cases where targets are more flexible and willing to make changes and seek amends, anger as opposed to contempt will be more typical. In line with this, interpersonal studies on marriage show contempt, and not anger, to be a strong predictor of divorce (Gottman 1993). or ones where similar contextual features (the unchangeability of anger's targets for example) are present. At this point then, we should take seriously the possibility that recognition is anger's most central desire. Genuine desires for retribution are still permitted on this view, primarily when recognition has been blocked or denied. Indeed, having been systematically denied recognition of a harm seems like arguably rational grounds for desiring vengeance. 15 Retribution still seems to be a robust desire in anger then, albeit an arguably secondary one. We will see however, that once we scrutinize what vengeance amounts to, recognition will emerge as potentially the only genuine desire in anger.

When Is Retribution Satisfied?
What counts as retribution differs depending on the orthodox view in question. On Aristotle's view, for revenge to be enacted the offender must know by whose hand, as well as for what reason, he suffers (1380b22-25). For Nussbaum (2015: 46), on the other hand, anger involves the desire 'for things to go badly somehow' for the offender, and she takes this to include cases where one is not at all causally related to the suffering of the offender. There is a desire, on Nussbaum's account, for a form of fateful harm to fall on one's offender. This could involve a desire for the offender to suffer an offense of the same type as the one they have caused the angry partyan eye for an eye, but it need not. At its most permissive, the claim is that in anger we wish for the offender to suffer, even if in the distant future, and even if in a form that might be unrelated, causally or conceptually, to the wrong committed against the offended party.
Psychological work on revenge has investigated what sorts of conditions satisfy justifiably angry agents. In social psychology, there are competing views on whether the goal of revenge is suffering, such that comparative amounts of pain are equaled between offended and offender (Frijda 1994), or understanding, such that the offender come to understand the wrong they have committed. These have been called the 'comparative suffering' and the 'understanding' hypotheses of revenge, respectively (Gollwitzer and Denzler 2009). The understanding hypothesis has received most empirical support (Gollwitzer et al. 2011). Studies have shown that seeing the offender suffer from 'fateful harm', i.e. harm conceptually and causally unrelated to the harm they caused, did not lead to a reduction in anger, or to an increase in satisfaction on behalf of the offended party. Satisfaction was only observed when the offender expressed understanding of why retribution was being sought against him (Gollwitzer and Denzler 2009;Gollwitzer et al. 2011). This empirical work suggests that revenge is only satisfying when accompanied by epistemic elements. 15 Note that such a proposal has considerable capacity to cope with cases where angry agents act aggressively in the first instance, for the claim is not that retributive actions are only pursued when attempts at recognition have indeed failed, but that punitive actions will be pursued under conditions where recognition is unlikely. Evaluations that recognition is unlikely can occur at a conscious deliberative level over time, from observing failed attempts, but they can arguably also occur almost instantly, at the appraisal level involved in the emotion itself. This means that cases where angry agents react to harms in an immediately aggressive manner will not necessarily count as counterexamples to the relevance of the perceived changeability of anger's targets as a key moderator.
Even when revenge is sought then, it seems to be the epistemic changes in the culpable agent that drive satisfaction, as opposed to suffering itself. This suggests that punition, on its own, might not be amongst the satisfaction conditions of a desire central to anger. If this is the case, then it seems that punishment is unlikely to be a sufficient satisfaction condition for a desire central to anger. If we think there are also non-punitive desires central to anger, as those who claim anger often aims for recognition hold, then punishment is not a necessary satisfaction condition for a desire central to anger either. We therefore have reason to take seriously the proposal that recognition is the only necessary and sufficient satisfaction condition for a desire central to anger. Now we have reason not only to think recognition is a robust desire in anger, and that recognition is the primary desire in anger (as retribution is sought under conditions where recognition is hard to attain), but further, we have reason to think that retributive aims might be instrumental towards attaining recognitional aims. 16 If punishment is neither necessary nor sufficient for the satisfaction of a desire in anger, then it is at least worth considering a view on which retribution may only ever be instrumentally desired in anger, as a way of enforcing recognition on the emotion's targets. Indeed, this fits not only with evidence regarding what makes revenge satisfying just canvassed, but also with evidence surveyed in the previous section (3.2) regarding when revenge is sought in the first place. Recall that retribution was typically sought in nothing-to-lose scenarios, where recognition had been denied. This evidence is consistent with angry agents shifting from an intrinsic or non-instrumental desire for recognition to an intrinsic or non-instrumental desire for retribution in nothing-to-lose scenarios, as I suggested above, but it is also consistent with a view that takes nothing-to-lose scenarios to motivate instrumental desires for retribution that are aimed at enforcing recognition on the emotion's targets. In nothing-to-lose scenarios then, perhaps what angry agents intrinsically desire is still recognition, but they just seek it through retributive means given that nonpunitive means have failed or been blocked. This suggests that recognition might be the single desire that should be taken to be constitutive of anger, and that the orthodox view got things not only a bit wrong, by failing to grant robust recognitional aims, but drastically wrong, by making constitutive of anger retributive desires that on a plausible account may only be instrumental towards attaining intrinsic recognitional goals.

Summing up
I have argued that empirical and conceptual considerations give us reason to give serious consideration to the proposal that the desire central to anger is recognition as opposed to retribution. This can be cashed out in a stronger constitutive claim or a weaker causal claim that mirror the constitutive and causal versions of the orthodox view I highlighted above. My argument for the centrality of desires for recognition is consistent with both a constitutive account, whereby anger necessarily involves only a desire for recognition and any other desires, including desires for retribution, play merely instrumental roles towards the attainment of recognitional goals. Alternatively, a causal recognitional view of anger would hold that desires for recognition are anger's most prototypical causal effect, making desires for retribution secondary, albeit allowing them to be sought for their own sake, that is, non-instrumentally. On both the causal and constitutive readings, desires for recognition are arguably definitional of anger (prototypical causal effects are typically relevant to providing definitions of psychological phenomena). On both the stronger and weaker readings, the orthodox view is abandoned, and the status of anger as a 'hostile' emotion comes into question. Developing a systematic account of anger as tied to recognition, of the constitutive or causal variety, is a topic for future work. Here I have endeavored to argue that an account of this sort deserves our attention. This is an important outcome as views of anger that privilege recognitional aims have, so far, been lacking. I have not argued that there are no rebuttals open to the orthodox view, 17 instead of devoting the rest of this paper to considering the form such rebuttals might take and assessing their force, 18 I want to do something different in the remainder of this piece. I think there are important considerations pertaining to why the retributive view of anger became, and has remained, orthodoxy, that are worth bringing to light. The remainder of this piece will be diagnostic in this sense. Uncovering reasons for the prevalence of the orthodox view will act as further, indirect, reason to take the proposed view seriously, as the reasons for which the retributive view remains the orthodoxy will be revealed to be bad reasons to maintain adherence to it.

The Making of Orthodoxy
Put simply, the retributive view of anger may have become, and remained, orthodoxy because it was in the interest of those in power that it be this way. Evidence from psychology as well as the history of emotions provides significant support for the view that our vernacular concept of anger may have been influenced by unjust social arrangements. Throughout most of the history of the Latin west, the main governing social powers were small prosperous ruling classes, and the increasingly wealthy Christian church. It is important to note that the writings of perhaps the most famous, and harshest, of anger's critics, Seneca, had a deep and pervasive influence on the fathers of the Christian church (Freedman 1998). For Seneca (2010), anger is 'hungry for a vengeance that will bring down the avenger too' (1.1), making the emotion often counterproductive to the emoting agent. Indeed, 'angry people curse their children with death, themselves with poverty, their households with ruin, and they deny they're angry just as madmen deny they're insane' (2.36.5). In likening anger to insanity, anger is cast at complete odds to reason 'distinct symptoms that 17 Note that evolutionary considerations are unlikely to challenge my proposal, for it would be a mistake to assume that anger's aim for recognition emerges on the time scale of human learning from child-rearing. The types of conditions that would have made anger aimed at recognition adaptive are not a feature of modern society alone, and may have influenced the emotion several thousands of years ago (Sterelny 2016). 18 See Silva (2021a) on this. mark madmen...the signs of angry men too, are the same' (1.1.3). As a stoic, Seneca viewed all passions as essentially defective beliefs that are to be curtailed so as to live a virtuous life. Seneca's De Ira casts anger as the worst amongst the emotions, 'see the foundations of the most celebrated cities hardly now discerned; they were ruined by anger' (1.2.1), 'we shouldn't control anger, but destroy it entirely-for what "control" is there for a thing that's fundamentally wicked? ' (3.42). For Seneca, 'no pestilence has been more costly for the human race' (1.2) than anger.
Seneca gives us three reasons related reasons to condemn anger. First, anger is responsible for violent and destructive actions against others. Second, in causing destructive actions, anger is often counterproductive to the aims of the angry agent themselves. Actions done out of anger can compromise one's long term aims, as well as provoke opponents to retaliate against oneself. Lastly, by desiring the suffering of another, anger is plausibly 'fundamentally wicked', whether or not it causes destructive actions. Seneca's De Ira is a dialogue that records a debate between a Stoic and an Aristotelian view on anger. Although Aristotle and Seneca both adhere to the descriptive orthodox view that anger is inherently retributive, Aristotle has a more optimistic view of anger. First, for Aristotle, anger often follows reason (EN 1149b), such that the stark opposition Seneca advocates between anger and reason is not present on an Aristotelian account. Secondly, and crucially, for Aristotle virtue can demand anger, making the emotion morally justified when it is felt for the right reasons, in the right proportion and for the right length of time (1125b32-33). In De Ira Seneca is engaged in refuting the claim that anger can be morally justified (see Cherry and Flanagan 2017). Seneca says that his Aristotelian opponent can raise the following objection to his total condemnation of anger: 'Just as virtue is kindly disposed to honorable behavior, so should it greet disgraceful behavior with anger' (2.6), to which Seneca replies that 'virtue will never make the mistake of imitating vices... It's a natural property of virtue to be glad and joyful' (2.6.2). Seneca rules out the possibility of anger ever being called for, or justified, even if its negative consequences are curtailed. Anger, for Seneca, is completely at odds with virtuous life.
It is significant that Seneca, and not Aristotle, who believed in justified and moral anger that often follows reason, was the writer who most influenced the Church. This arguably led to the popularization of an extremely negative view on anger's nature, and, significantly, one that holds that injustices should not be met with anger. Seneca's influence on the Christian church is most evident in the writings of Martin de Braga, who popularized a self-authored book entitled De Ira, made up almost entirely of verbatim quotations from Seneca's work by the same title. This early medieval work resurrected the strong anti-anger sentiment of the original text and spread its message throughout the growing Christian Kingdom. By the late middle ages, monastery rules prohibited anger of monks, and religious advisors to European monarchs counselled against anger, highlighting its incompatibility with a virtuous leader (see Freedman 1998). Note that Seneca's influence is not limited to the Christian west. Indeed, the religion most faithful to Stoic teachings is arguably Buddhism. The key Buddhist text, The Way of the Bodhisattva, dedicates an entire chapter to anger, arguing that it 'has no other function than that of causing harm' (6.8) (Śāntideva 1997).
Once inscribed in religious teaching, the view that anger is inherently retributive will have easily dominated folk psychology. This genealogy of the concept of retributive anger of course does not itself undermine the orthodox view, but it does suggest that the folk concept of anger could have been otherwise, had the Christian church not been so influenced by Seneca's writings. Importantly, if anger for many of us, including Aristotle, is sometimes called for, or appropriate, particularly in response to injustice, it is at least relevant to note that the view of anger that became popularized in the middle ages was one that made little room for justified anger in response to injustice.
Another important historical datum is found in the exceptions made of anger's necessarily sinful nature. In Christian scripture, God can become enraged, and justifiably so. As supremely virtuous, God's anger does not threaten his divinity and instead anger is made acceptable and expected amongst the almighty. This had its counterpart in the earthly realm, where ira regis, the anger of Kings, was accepted as rational and virtuous, likely through a divinely inherited legitimacy that the Christian church granted the monarch. Kings, therefore, through their proximity to God, and their divine right, were also entitled to anger. Virtuous anger soon became the purview of the noble alone, extending beyond the King but excluding the lower classes, 'anger in the latin west was a sin, but a sin that could be turned into a virtue, monopolized by an aristocracy' (Rosenwein 1998: 5). Indeed, the anger of peasants has been historically either ridiculed for being mistaken, or vilified for being violent and animal like; 'comic or murderous, peasant anger was quintessentially irrational' (Freedman 1998: 171). As anger was more than often about defending one's honour, the anger of peasants seemed inherently misplaced, as they were not allowed any honour that could be in need of defense or recognition. Indeed, a recurring satirical depiction of peasants in the middle ages was that of a foolish peasant attempting to pass for a knight, but giving himself away in his failure to act chivalrous or defend his honour in battle (Freedman 1998). This suggests that anger is allowed recognitional aims only of those in social positions that warrant them. The anger of peasants doesn't seek recognition in the eyes of those with power, because there is nothing that can be recognized if the subject harmed is not properly a subject at all.
Anger is conceived differently depending on who is getting angry at who. The anger of agents or groups with lower social status is perceived as less rational, as not actually tracking any wrongs. This is an insight that has long been made in the feminist philosophy literature, where the anger of the oppressed is observed to be systematically denied uptake by dominant groups (Frye 1983;Scheman 1980;Campbell 1994). That is, the anger of the oppressed is not perceived to be a response to injustices; their anger is not perceived to communicate that a wrong has occurred, while the anger of dominant groups is. Often things are more fine-grained than this. The anger of the oppressed is often taken seriously as a claim that a wrong has occurred only when this anger is expressed within domains over which the oppressed group is perceived to have agency (the anger of women is taken seriously in the kitchen, but less so in the political sphere or the bedroom, for example) (see Frye 1983). Oppressive social relations therefore influence whose anger is seen as a legitimate claim about a wrong having occurred, and whose aren't.
Conceptions of anger as inherently retributive compliment social practices aimed at disregarding the anger of oppressed groups. If anger is about punishment rather than recognition, then anger doesn't necessarily communicate that an injustice has occurred, it communicates that the angry agent wants to make targets of anger suffer. If anger is about punishment rather than recognition it can easily be discounted as excessive, for it always risks becoming morally problematic in disposing agents towards retributive actions. If anger is punitive rather than recognitional it needs to be controlled, abolished and managed, and its epistemic value is obscured. A punitive conception of anger serves those who stand to lose the most as it dismisses and/or occludes anger's recognitional aims. I am not suggesting that the retributive conception of anger was necessarily endorsed for these reasons, but rather noting that this conception of anger serves those in power and who are reluctant to give up their power. This is not an arbitrary state of affairs, even if the view was not maliciously endorsed by dominant groups. As Srinivasan (2018) notes, we talk about anger 'in the way we do because it serves those whom anger most stands to threaten, and that this is no mistake at all.' Experimental work in psychology stands in line with the observations just made. First, it is relevant to note that social psychologists have made similar claims regarding the influence of ideology on psychological theorizing. Drury et al. (2020) for example argues that early accounts of crowd psychology, that tied the angry crowd to immanent violence, 'achieved popular acclaim and were dominant for so long not because of the strength of their evidence, but rather because they provided a "scientific" gloss on a prevalent anti-collectivist ideology' (176). Secondly, the perceived appropriateness of anger has been observed to moderate how the targets of anger respond to the anger of oppressed groups (de Vos et al. 2016). When anger is perceived as appropriate the targets of anger experience empathy towards the oppressed, while when they perceive anger to be inappropriate the dominant groups responds by retaliating against the angry group. This stands in line with how important perceptions of appropriateness are likely to be for anger's recognitional aims to me met. Most importantly, recall that retribution was observed to be sought in nothing-to-lose scenarios, and other scenarios where the targets of anger are perceived to be unwilling to change and/or address injustices. These are scenarios where agents believe there is no other avenue available to confront the injustice they suffer, and therefore where they have little to lose in acting in risky and punitive manners. Real world cases of social injustice are likely to often be quite similar to these nothing-to-lose scenarios. These are cases where dominant groups are often reluctant to give up power, where the anger of oppressed groups is caricaturized as excessive, and where it is often not perceived as rational or appropriate. These are cases where the purely recognitional aims I have argued to be central to anger will be hard to achieve, and therefore situations that the empirical work predicts will result in retributive actions. If real life cases of social injustice are like nothing-to-lose scenarios, then they involve key moderators that the experimental work has highlighted to play a determinate role in leading angry agents to act retributively: they involve targets of anger that are unwilling to change and that are likely to perceive the anger of the oppressed as inappropriate. Anger will therefore, in fact, likely be more retributive in cases of social injustice than in cases where the targets of anger are more receptive to anger's recognitional aims. This means that features of social injustices may turn anger hostile. Without diagnosing the features that lead this to be the case, anger may seem, as the orthodox view claims, inherently retributive and hostile. The orthodox view is guilty of reading into the very nature of anger, what may only be contingent features of anger in specific contexts particular to nothing-to-lose, and related, scenarios. In so far as the orthodox view might capture retributive forms of anger successfully, it does so by obscuring the social structures on which they may depend. Importantly, the orthodox view may contribute to its own veracity, as a society in which this orthodoxy dominates folk psychology will be one more inclined to disregard anger as a hostile caprice, rather than an epistemically important state that seeks recognition of wrongs.

Conclusion
I have argued that a view that excludes retribution from anger's central aims is worth serious consideration. This gives us reason to question whether anger is aptly construed as a hostile emotion. My argument suggests that attention to empirical evidence, both experimental and historical, stands in line with a view far removed from the orthodoxy that anger is a retributive and hostile emotion. This stands as a methodological bid that future work on anger not ignore such bodies of research. That being said, the evidence relied upon is indirect and suggestive and I have not purported to provide a knockdown argument in favour of the view that anger aims only for recognition. I do however take myself to have argued that such a view is worth serious consideration; consideration which has, until now, been lacking.
I argued that not only is recognition a robust desire in anger but that recognition may be the only necessary and sufficient condition on the satisfaction of a desire central to anger. Furthermore, retributive action tendencies may only be pursued under conditions where non-punitive recognition is an implausible or unavailable goal. These considerations are compatible with both a constitutive reading on which anger is necessarily tied to recognition and only instrumentally aims for retribution, as well as a weaker causal view where desires for recognition are still primary in anger, but intrinsic retributive desires are sometimes allowed (perhaps in nothing-to-lose type scenarios). These considerations are also compatible with a view whereby two distinct types of anger, one retributive and one recognitional, exist, but the data suggests that the latter sort of desires are primary, supporting a view stronger than a pluralist one that I have previously argued for (Silva 2021a). My argument therefore suggests that the orthodox conception of anger is not merely an oversimplifying of anger phenomena, but may be seriously mistaken. This in turn suggests that anger's characterization as a hostile emotion is ill-fitting. This project has been primarily descriptive, such that the normative features and consequences of a view that takes anger to aim for recognition remain to be explored. It is not implausible, however, that as the desire for retribution becomes less central to anger, a more optimistic account of the emotion's moral status emerges.

Code Availability n/a
Funding Open Access funding provided by Université de Genève. Leverhulme Trust.

Conflict of Interest n/a
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.