Doing Emotion History

As Sir Geoffrey Lloyd observes, “very considerable confusion still exists … on what emotion is and how it should be defined.”¹ The modern English term “emotion” has its own particular history,² and its international dominance, like that of English-language categories of particular emotions, is a regular source of complaint among those who study comparable phenomena in other cultures.³ English “emotion” is by no means co-extensive with its nearest equivalents in Greek or Latin. Pathos in Greek (from paschein, to undergo) has a wide range of senses in fields from logic to medicine.⁴ And although Greek writers from the 4th century bce onwards do refer regularly to the pathē of the soul (psychē), not even that category is co-extensive with English “emotion.” The Romans, for their part, deploy a range of different terms, such as animi motus, adfectiones/affectiones, adfectus/affectus, passiones, and perturbationes, for what the Greeks call the pathē of the soul.

The issue of definition is a live one in emotion research.⁵ For James A. Russell, “Emotion researchers face a scandal. We have no agreed upon definition of the term—emotion—that defines our field.”⁶ But Russell’s own work reveals that this is less of a scandal than it might seem. Though influential modern approaches have sought to pin down precisely which element in any scenario is “the emotion,”⁷ Russell is one of many to have demonstrated that such a project is misplaced. Emotions are fundamentally event-like in structure, a feature often described in terms of “scripts.”⁸ A script is a more or less prototypical scenario, not an essence:⁹ neither the category of emotion nor the categories of particular emotions (anger, fear, shame, etc.) are “classical” ones, in which membership depends on fulfilment of a limited set of necessary and sufficient conditions, but “fuzzy” ones that manifest indeterminate boundaries, overlap with other categories, display ambiguities regarding membership, and have an internal structure that encompasses peripheral as well as prototypical cases.¹⁰ The labels that are applied to these categories cover a range of cases with a variety of characteristics.

Though there may be a prototypical script or set of scripts for each emotion category, their elements are flexibly combinable, and no element is essential. Prototypical emotions involve appraisals of states of affairs, but there are less prototypical cases which do not—as when we are angry or afraid, but do not really know why. Emotion episodes regularly do include visible expressions and typical patterns of behaviour, but they need not. And even the feeling of arousal that we regard as characteristic of emotion is not absolutely necessary: fear of snakes can be dispositional as well as occurrent, so that we take care never to encounter one. And when we say things like, “I’m afraid you’re mistaken,” we are drawing on aspects of the category of fear without commenting on an occurrent experience.

Yet the occurrent emotional state, involving both valence (whether the stimulus is pleasant or unpleasant, good or bad, for the agent) and arousal (how calm or agitated the agent feels), is the prototypical case. Emotion research focuses overwhelmingly on episodes that involve some form of phenomenally salient affective state. Equally typically, states of that sort involve perceiving and appraising an object in certain terms. Emotions typically have both a markedly phenomenal aspect (what it feels like to experience one) and an intentional aspect (in the technical sense of being about something, having some object in mind). Thus emotions involve perceptions of ourselves as experiencers of the world as well as perceptions of the world. This combination of phenomenal salience with intentionality in the occurrent affective states that we, in English, prototypically regard as emotions is found also in the analogous Greek and Latin categories.¹¹ Typical examples highlight forms of affective arousal that present themselves as things that happen to us rather than as things that we do. This is no doubt a prototype that classical concepts themselves have played a significant role in shaping. But though the combination of intentionality, phenomenality, and passivity may account for the “best examples” of the categories in question, it is not difficult, in English, Greek, or Latin, to find examples of emotion (pathos, adfectus, etc.) that do not exhibit all these features. Much modern research suggests that phenomenal passivity is a feature only of folk concepts of emotion.¹² Old antitheses between reason and passion, cognition and affectivity, break down in contemporary views of affective experience as a continuum that underpins all conscious psychological processes,¹³ including those most associated with rational choice.¹⁴ The episodes that contemporary English describes as “emotions” are just the peaks and troughs of an affective continuum.¹⁵

The perception that emotions involve ways of perceiving the world and of perceiving ourselves as perceivers of the world is not a thin, passive experience, but a cognitively rich and conceptually permeated one, involving active, “top-down” aspects deriving from knowledge and experience as well as “bottom-up” sensory data.¹⁶ As Russell puts it,¹⁷

Perceiving oneself as having an emotion is no different in kind from other perceptions. Percepts are often compelling, but they are not simple. Nor are they infallible. Like other acts of perception, an emotional experience is not entirely “bottom-up,” not entirely data driven. A percept is the end product of a complex process involving raw data, concepts, learning, and context.

Thus emotions involve perceptions that depend on the concepts and categories of the cultures we grow up in. These are not entirely subsumed in language (they can be active in the non-verbal performance of emotion), but language is a central aspect of both the conceptualization and the experience of emotion.¹⁸ Attention to historical and cross-cultural variation in the sense and reference of emotion concepts is not about focusing on emotional discourse or “signs of emotion” rather than on “emotions themselves,”¹⁹ but a fundamental aspect of emotion research, an enterprise in which historical, linguistic, and literary disciplines such as Classics have much to contribute.

As events in themselves and as aspects of larger event-structures, emotions lend themselves to representation in narrative terms.²⁰ Narratives furnish some of the best evidence for the study of emotion in any culture. Though there is a genuine sense in which the prototypical case of emotion is one that is being performed, by oneself or another, right here, right now, the fact that such cases are understood intersubjectively in terms of shared concepts, with their associated scripts and narratives, means that people can talk, in narrative terms, about the present and past emotions of themselves and others. Historical and literary sources are rich in such narratives.

Narrative entails contexts in which events proceed from cause to effect in ways that reflect the character and motivation of agents. When Thucydides (1.23.6) identifies the Spartans’ fear (phobos) of the Athenians’ growing power as the “truest cause” of the Peloponnesian War, it does not matter whether he thinks or means readers to think that any individual Spartan or group of Spartans experienced an occurrent episode of emotion on any particular occasion. The Spartans’ fear represents their attitude towards and behaviour in the light of developments as they construed them. That construal may have been one to which historical Spartans gave voice; but Thucydides’ identification of that emotion as a cause of war is adequately justified by his understanding of the policies and actions of the Spartan state. It is a regular feature of emotion scenarios that the behavioural and evaluative aspects of the script may be much more prominent than the affective, phenomenological elements.²¹ Emotions involve the outward as well as the inward aspects of the body, interaction between the individual and the physical and social environments, embodied practice in roles, relationships, and institutions, as well as the shared norms and concepts that give them their structure.²² The narratives that encompass such elements have their own cultural and historical contexts, and so the contextual factors one takes into account in assessing the presentation of emotion in a given narrative may need to encompass various aspects of genre and intertextuality.²³ One cannot separate emotion as subjective experience from the contexts, expression, behaviour, norms, and discourse of emotion.

Thus Classicists are in a fortunate position. Many of the discipline’s skills—the understanding that translation is about ideas, not lexical tokens; the knowledge that ideas must be understood in contextually situated terms; the sense that terms and concepts change over time; the sensitivity to narrative cues about the motives of characters in the diegetic, mimetic, and mixed genres of ancient literature—turn out to be essential in using textual sources as material for the history of emotions. In so far as emotions are event-like in nature, readers have access to the emotions of individuals represented in any text that has even a minimal narrative form—from the epics of Homer to the sketches of the conditions, evaluations, and behaviour patterns associated with individual emotions in Aristotle’s Rhetoric.

Ancient Theories of Emotion

Ancient theories of emotion vary in their emphases, but generally agree with mainstream modern theories in seeing emotions as psychophysical events that feel a certain way, motivate action, and depend on appraisals of contexts and their significance for the agent.²⁴ Ancient theorizing about emotion begins with Aristotle, but earlier thinkers had also shown an interest in relevant phenomena. Gorgias’s Encomium of Helen reflects a concern with the emotional effects of poetic speech that is rooted in the implicit aesthetics of earlier Greek (especially epic) poetry and anticipates later discussions in works such as Plato’s Ion and Aristotle’s Poetics.²⁵ Plato never attempts to develop a systematic category of emotional pathē, but emotions in our sense do feature prominently in his work.²⁶ A notable feature for Plato is their ambivalence: in the Philebus they are introduced as mixtures of pleasure and pain in the soul (Phlb. 47e–50e). In the Republic and the Phaedrus, emotions such as anger and shame are crucial to the account of the “middle” element of the tripartite soul, the thymoeides or “spirited part.” The potential of the thymoeides to support and to conflict with reason suggests a combination of cognition and affectivity in the emotions with which it is associated. This is a feature of thymos also in the Homeric poems, an important source of inspiration for Plato’s own conception,²⁷ where it is associated with a wide range of cognitive, desiderative, and affective processes. Though the Republic does recognize educable, morally useful emotions, public emotional display, and in particular the effects that representations of emotion in epic and tragedy have on the emotional character of audiences, are also central to the dialogue’s case (in Book 10) against the performance of those genres. In later dialogues, such as Timaeus (69c–d) and Laws (644c–d), fear, anger, confidence, hope, and the like are typically represented as unreliable motives.

Aristotle discusses emotions especially in the Rhetoric and the De anima. In the former, he defines the pathē as the things that cause people to change and differ in respect of their judgements (2.1, 1378a19–22), specifying them as “anger, pity, fear, and all other things of that kind, as well as their opposites” (cf. Eth. Nic. 2.5, 1105b21–23; Eth. Eud. 2.2, 1220b12–14). In Rh. 2.2–2.17, he discusses twelve sets of emotion scripts, in terms of the dispositions of their patients, their targets, and the conditions and circumstances in which they are elicited, concluding with a series of character sketches outlining the differences in people’s emotional dispositions according to age and circumstance. As befits its focus on persuasion, the Rhetoric’s emphasis is on emotions as evaluative responses, especially on the part of audiences.²⁸ This focus on audience emotion becomes a general orientation of ancient rhetorical theory (cf. e.g., Cic. De or. 2.185–216; Quint. 6.1–2). But the scenarios discussed also provide material for narratives of the emotional behaviour of others (e.g., one’s opponents), and the speaker’s representation of his own emotions is also important (e.g., 3.7, 1408a9–23). Throughout, the social aspect of emotion is to the fore: the scripts that Aristotle adumbrates reflect his chosen emotions’ role in social and political interaction and the dynamic interplay between emotion and social norms, especially in terms of the relation between emotion and social status.

But the Rhetoric also specifies that emotions are accompanied by pleasure and pain (2.1, 1378a20–21). The occasional addition of the term “disturbance” (tarachē) suggests that pleasure and pain here encompass not only what modern psychologists call “valence” (see section 1), but also the subjective phenomenology of physiological arousal. In the De anima discussion the physical aspect is much more prominent. In the opening chapter, Aristotle includes examples of what one might call emotions (anger, confidence, desire, 403a7; thymos, mildness, fear, pity, confidence again, joy, loving, hating, 403a17–18) alongside processes such as “perceiving in general” (403a7) in discussing whether all pathē of the psychē are shared by the body (403a16–b19). These pathē encompass all that the psychē undergoes in the most general sense. The emotions are chosen as prototypical examples mainly because they best illustrate the point that all or almost all pathē of the soul are “with body.” The De anima’s overall position is that soul and body may be distinct in definition, but are inextricable as aspects of the living organism (1.1, 403a10–16; 2.1, 412a1–413a10). The emotions and the other pathē of the soul reflect this conception in so far as they are “enmattered accounts” (logoi enyloi, 403a25).²⁹ A dialectician and a natural scientist may offer different (formal versus material) definitions of (e.g.,) anger (as “desire to cause pain in retaliation” and as “boiling of the blood and hot stuff around the heart,” 403a30–b1), but neither is correct in itself. The true natural scientist’s definition of an “affection of the soul”—as “a certain movement of a body of a given kind or a part of it or a capacity of it produced by such and such and for the sake of such and such” (403a26–27)—requires both. Emotion is embodied not only in the sense that all experiences of embodied individuals are embodied, but also in that an emotion requires a specific physiological process (403a18–19, 403a31–403b1). Both conditions, formal and material, are necessary, but neither is sufficient: one can be faced with powerful reasons for anger or fear, but not feel the emotion because one is not in the requisite physical condition, but equally, if one is in that condition, the emotion may result even when the external circumstances do not warrant the appraisal that normally elicits it (403a19–24; cf. Rh. 2.2, 1379a10–22). This means not that physical conditions can determine the occurrence of an emotion without any appraisal at all, but that an appraisal of external circumstances which would not otherwise elicit the emotion can result from the physical condition of the patient. This emphasis on the physiological aspects of emotion recurs in Aristotle’s other works.³⁰

Emotions are prominent elsewhere in the Aristotelian corpus. In the Nicomachean and Eudemian Ethics, the proper development of emotional capacities forms the basis of the virtues, which involve feeling emotion as reason prescribes in a given situation (Eth. Nic. 2.6, 1106b18–23). Emotion is thus essential for the developed forms of disposition and judgement that enable one to practise the virtues and to live a good life. This is not a matter either of repressing emotion or of moderating it in all circumstances, but of the development of states of character that allow one to feel emotion whenever and however it is appropriate to do so. The utility of emotion and its alignment with moral judgement also emerge in the Poetics, where pity (eleos) and fear (phobos) are the means by which tragedy effects the famous yet enigmatic katharsis (purification) of such emotions (Poet. 6, 1449b27–28). In one kind of tragic plot (the best, one of the two best, or the second best, depending on one’s interpretation),³¹ Aristotle continues (Poet. 13, 1452b28–1453a12), pity and fear are elicited by the change of fortune (from good to bad) of a high-status individual who is neither outstandingly virtuous nor thoroughly wicked, but who fails because of some error (hamartia). Though a passage in the De anima (3.3, 427b17–24) suggests that people are unaffected when they contemplate emotion-eliciting scenarios in works of art,³² the position of the Poetics is much more in keeping with ancient aesthetic and rhetorical theory in emphasizing that audiences experience real emotion, as one would in comparable real-life situations, in response to representations in art, narrative, and drama.

The approach to emotion in the main Hellenistic philosophical schools bears comparison with Aristotle’s account to the extent that both Stoics and Epicureans would agree on the indivisible unity of emotions as cognitive and physiological experiences: for both, all mental events are at once experiences of the body, including its no less material psychē. Mind and body affect each other mutually, and both judgements and physiological changes are necessary for emotion.³³ But whereas for Aristotle everyday emotions can be integrated into a life that fulfils human potential for virtue and flourishing, for the Stoics and Epicureans, though the virtuous life is by no means affect-free, common or garden emotions are by and large regarded as material for therapy.³⁴ This can be oversimplified as a distinction between the allegedly Aristotelian pursuit of moderation in emotion (metriopatheia) and a Hellenistic therapeutic approach that takes emotion, at least as commonly understood, to be a bad thing (e.g., Cic. Tusc. 4.38–57; Sen. De ira 1.6–21).

Of the two schools, the Epicureans make more concessions to everyday experience, but their approach can also involve a critical attitude towards many ordinary emotions, as obstacles to the goal of psychological stability (ataraxia). Substantial reflections of Epicurean approaches to emotion survive in the fragments of Philodemus’s De ira and Lucretius’s influential De rerum natura, 3.258–322.³⁵ From Philodemus, it appears that Epicureans classified emotions, like desires, as natural and necessary; natural but not necessary; or neither (“empty”).³⁶ The tendency to anger is natural (De ira, columns XXXVII.16–XLIV.35 Indelli), but not necessary (fragments 10.14–15, 12 Indelli). It is found even in the wise (XXXVI.17–20, XLVI.40–XLVII.18, XLVII.41–L.8), who may experience its pain when friends are wronged or when a friend does wrong (XLI.17–25); but such anger will be brief and mild (XXXIV.16–XXXV.5, XLII.4–14, 21–40, XLIV.26–8, XLVII.29–41), stopping far short of anything that could be described as “rage” or “fury” (thymos, XLIII.41–XLVI.14). The error of treating anger as necessary, by contrast, and taking pleasure in revenge or punishment (XLII.21–34, XLIV.28–35), makes it “empty” (XXXVIII.1–6, XXXIX.7–8). For Lucretius, anger is a matter of an individual’s physical constitution: the degree to which one is disposed to it will depend on the element of heat in the soul (3.288–306). Reason cannot completely remodel such dispositions—there will be residual traces (3.310–22); but it would in any case not be right to eradicate anger, as it is sometimes appropriate (3.312–13).

The Stoics go further in arguing that all pathē involve mistaken judgements and so have no place in the virtuous life. For them, pathē involve an impression (phantasia) that leads to arousal, but also an assessment that a stimulus of a certain type is present, an assent to that assessment, and a judgement that it is appropriate to be moved on that basis.³⁷ They are thus functionally rational but normatively irrational:³⁸ though they depend on rational assent, they are irrational in being mistaken about the good, the product of defective psychological states, and (after assent) difficult to control.

The Stoics distinguish four general categories of pathē: two that focus on the present (pleasure and pain) and two on the future (desire and fear).³⁹ But Stoic definitions also make summary reference to physical changes (e.g., “contraction” in the case of pain and “elation” in that of pleasure).⁴⁰ Emotions are not only judgements but also alterations in the pneuma (a mixture of fire and air) that is concentrated in the control centre or hēgemonikon in the heart (Chrysippus in Galen, PHP 3.7.2–4).⁴¹ But pneuma also suffuses the entire body, and the Stoics are attentive to a wide variety of physical symptoms beyond the changes that occur in the hēgemonikon. Soul and body affect each other mutually (by sympatheia). All these processes entail physical changes of an ensouled body.⁴² Thus, though the Stoics are famous for their claim that emotions are judgements, these are judgements of a particular sort: both the judgement (with its own physiological underpinnings) and the accompanying feelings and symptoms are necessary for an experience to count as an emotion.

Since emotions are not simply “cool” judgements, and since, once the subject has given assent, they are difficult to control,⁴³ motivational conflict is possible in Stoic theory.⁴⁴ This, however, is a matter not of the conflict of rational versus irrational desires, but of temporal vacillation between successive judgements of value.⁴⁵ The fact that emotions involve assent makes one responsible for them.⁴⁶

The Stoic ideal is the eradication of such pathē; but this is not a demand that human beings should be immune to all forms of affective experience. Seneca (De ira 2.2.5) writes of “preliminary beginnings of emotion” (principia proludentia adfectibus), and earlier Stoics had a category of “pre-emotions” (propatheiai).⁴⁷ These are described in terms (such as “bites” or “pangs”) very like those used of the pathē in ordinary language. Even the Stoic sage will experience pre-emotions (e.g., Sen. De ira 2.2.2), though she will not assent to them.⁴⁸ The sage will also experience “good emotions” (eupatheiai) informed by virtue.⁴⁹ To the four categories of pathē there correspond three of eupatheiai: joy (at being virtuous), wish (to be virtuous), and caution (regarding anything that detracts from virtue). There is no virtuous analogue to pain (lypē), because nothing bad is present to the sage.⁵⁰

Ancient philosophical and psychological theories of emotion thus have a history, one that is bound up with the norms and values of specific societies and the psychological and ethical theories of the major thinkers and schools. There is a strong family resemblance between ancient and modern categories and theories, especially with regard to emotions as irruptive, phenomenologically passive states.⁵¹ Though the rationality of pathē (their role in a virtuous life) is a recurrent issue and forms of reason versus passion dualism are found, there are no hard and fast antitheses between mind and body, cognition and affectivity.

Approaches to Emotion in Classics

A strong stimulus to contemporary research on ancient emotion was William Fortenbaugh’s 1975 book on Aristotle, which is fully informed by the cognitive-evaluative approach to emotion prominent in the 1950s and 1960s.⁵² That approach is central also to the spate of monographs and edited collections on emotion that have appeared since the 1990s.⁵³ These have been joined by a large number of studies on particular emotions—anger,⁵⁴ disgust,⁵⁵ envy and jealousy,⁵⁶ fear,⁵⁷ forgiveness,⁵⁸ grief,⁵⁹ hope,⁶⁰ joy,⁶¹ love,⁶² pity,⁶³ remorse,⁶⁴ shame,⁶⁵ and wonder.⁶⁶ The central strength of these works has been their focus on the ancient emotional lexicon and the construction, conceptualization, and valorization of emotion.

Douglas Cairns’s study of the Greek concept of aidōs (shame, respect, embarrassment, modesty, sense of honour) focused especially on strong evidence for internalization of and commitment to values of the honourable and the shameful that refute easy antitheses between shame and guilt. In a landmark study of Roman emotion, Robert Kaster concentrated on the social emotions verecundia, pudor, paenitentia, invidia, and fastidium as families of scripts for performance and self-presentation that involve delicate forms of balance between self-assessment, attention to others’ assessments, other-regarding behaviour, and the observance of social norms. David Konstan’s extensive body of work on emotions spans both Greek and Roman cultures and has contributed in particular to our understanding of ancient theories of emotion (particularly those of Aristotle and the Stoics, which offer fruitful opportunities for dialogue with modern cognitive-evaluative approaches),⁶⁷ to the semantics and history of ancient emotional concepts,⁶⁸ and to the sharpening of our appreciation of salient differences between ancient emotional lexica and our own.

One of the central emphases of this work has been the interaction between emotion and moral and social norms. This needs to be seen in terms not only of the embeddedness of ancient emotions, emotion concepts, and theories of emotion in social interaction and cultural normativity, but also of the fundamentally affective character of ancient moral, social, and legal values.⁶⁹ These features also constitute major topics in other disciplines.⁷⁰

Another important element of this current of research concerns the ways in which the conceptualization, labelling, and valorization of emotions shifts over time and varies from culture to culture. The fact that aidōs focuses on the honour of both self and others is a significant indication of the inclusive character of Greek notions of honour. This in turn provides a template for comparative analysis of analogous concepts in Latin.⁷¹ Though the Greek and Latin concepts map the emotional landscape in different ways, at a more general level both cultures illustrate the interplay of the esteem one seeks from others, one’s own image of oneself, and the respect for others’ claims to esteem that social interaction requires, prompting conclusions that undermine crude antitheses between competition and cooperation, heteronomy and autonomy. These conclusions reflect wider truths about the fundamental importance of mutual recognition in social interaction, in ethics and politics, and in the constitution of identity.⁷² In historical terms, a significant development may be observed in the way that both aidōs and its near-synonym, aischynē, begin (c. 430 bce) to focus not only on the future, on things one is ashamed to do, but also on the past, on things one is ashamed to have done (Eur. Heracl. 813–816; Hipp. 244).

Shifts can be observed in the usage of other emotion terms: in Homer, nemesis refers to righteous indignation at breaches of social norms, but it comes to be associated with grudging responses to others’ good fortune, a function exercised in the classical period by phthonos, normally malicious envy or jealousy, but sometimes also a morally responsive form of indignation.⁷³ Emotion terms also come into existence: the late 5th-century interest in the retrospective, guilty conscience is reflected in the coinage of the term syneidēsis (calqued in Latin as conscientia), which facilitates new ways of talking about what English-speakers call “guilt.”⁷⁴ In the 4th century bce, Greeks begin to use the term zēlotypia for forms of jealousy that might hitherto have been spoken of in other terms.⁷⁵

Literature has been central to these studies, because literary sources provide rich evidence for the complex dynamics of emotional episodes in multifaceted depictions of more or less realistic forms of social interaction. Genres such as epic and drama provide manifold perspectives on characters’ motivation and substantial information on the eliciting conditions of their emotions. A wide range of other genres, from elegiac poetry and historiography to forensic oratory and biography, rely similarly on narrative constructions of characteristic affective scenarios as contexts for their representations of emotion. Representations of the dynamics of emotional agency in mimetic, diegetic, or mixed literary genres provide some of the best evidence of affectivity in action in ancient cultures.⁷⁶ Genres can also be characterized by the thematic salience of particular emotions: the first word of the Iliad highlights the fact that (as Martin Mueller has put it) “Anger is to heroic what sex is to Victorian morality.”⁷⁷ But the dominant note as the poem ends is the pity (eleos) that Achilles presents as a fundamental response to the vulnerability to vicissitude that all human beings share.⁷⁸ As Virgil’s Aeneas says, weeping as he contemplates Trojan sufferings visually depicted at Carthage, sunt lacrimae rerum et mentem mortalia tangent (Aen. 1.462); yet, where the Iliad ends with pity, the Aeneid ends with a resurgence of anger. The mutability of fortune that makes all human beings vulnerable is both part of the script for pity in Aristotle’s Rhetoric and a recurrent feature of both literature and literary theory in antiquity, from Gorgias of Leontini onwards.⁷⁹ If one is to believe Socrates in Plato’s Philebus, however, it is pleasure in others’ misfortune that defines comedy (Phlb. 48a–50d). Forms of anger and indignation, meanwhile, can be seen as characteristic of Juvenal’s persona as satirist,⁸⁰ while Latin love elegy, to take another example, abounds in scenarios of erotic jealousy.⁸¹

Historiography, for its part, reflects not only the history of emotion concepts and the power of narrative to represent and to elicit emotion, but also the role that emotions play as historical causes. Western historiography begins with Herodotus’s narrative of offence and retaliation between Greeks and non-Greeks and the motif of reciprocity and retaliation structures the work on several levels.⁸² For Thucydides, as noted above, the “truest cause” of the Peloponnesian War is Sparta’s fear (phobos) of Athenian expansion (1.23.6). Herodotus’s interest in what one might call the “emotional regimes” of the peoples whose customs he reports is evident throughout;⁸³ but differences in affective style between Athenians and Spartans are just as important in Thucydides.⁸⁴ The ethnography of emotion is a recurrent feature of ancient historiography,⁸⁵ often in conjunction with various forms of environmental determinism.⁸⁶

Emotions also drive historical narrative: Tacitus claims to write sine ira et studio, but few believe him. Herodotus is explicit that wonder (thōma, i.e., thauma), both on his own part and as expected in his readers, is a major criterion behind the selection and presentation of his material. His own emotional attitudes to the events he narrates, for example pride at Greek resistance to Persian aggression or censure when that resistance is undermined, are not difficult to discern, with inevitable implications for audience emotion. When Thucydides writes that Nicias, of all his contemporaries, least deserved his fate (7.86.5), he not only pays tribute to his virtue (aretē), but also classes him as worthy of pity. In invoking that script, Thucydides is also addressing his readers’ emotions. Direct authorial intervention is just one of the ways in which historical texts do this. Historiography is thoroughly influenced by the emotion-eliciting practices of other literary genres (especially epic and tragedy) and by the theorization of those practices, especially in rhetoric. Though there could be controversy over how a historian might seek to move his audiences, as in Polybius’s criticism of Phylarchus for using sensationalist and dishonest methods to elicit sympathy for people who allegedly did not deserve it (2.56–63),⁸⁷ Polybius is fully in step with his fellow historians in regarding readers’ emotions (specifically pity and anger, 2.56.13) as essential elements in their understanding of historical processes. In a case such as Tacitus’s account of the death of Vitellius (Histories Book 3), a complex set of emotional responses is steered by the portrayal of Vitellius’s own actions and feelings, the deployment of typically emotion-inducing scenarios (such as the dramatic reversal in Vitellius’s fortunes, 3.68.1), authorial spin, and the reactions of internal audiences.⁸⁸ As a victim of the mutability of fortune, Vitellius is a potential candidate for sympathy (misericordia), until the abject terror in which he attempts to escape his fate and the unseemliness of his end make him “a disgusting spectacle,” object of the mob’s abuse rather than their tears (3.84).

Such passages highlight the techniques of literary narrative, elements of theatricality, and the vividness (enargeia) so prized by literary and rhetorical theory.⁸⁹ The primary aim of literary and rhetorical enargeia is audience emotion ([Longinus] Subl. 15; Quint. 6.2.29–32); and large among the emotions thus elicited looms pity.⁹⁰ The mutability of fortune is as typical a scenario for pity in ancient historiography and biography as it is in poetic and dramatic sources.⁹¹ Polybius (1.1.2) regards the use of others’ reversals to instruct people how to bear their own misfortunes as fundamental to the historian’s craft.⁹² The presentation of exemplary figures as worthy of emulation (Sall. Iug. 4.5) is as central to historiography as it is to biography.⁹³

Early 21st-century approaches to the history of ancient emotions, fostered especially by a series of volumes edited by Angelos Chaniotis,⁹⁴ widen the field, its focus, and its source base, with greater attention to sources for the expression, performance, and context of emotion in non-literary texts and in material culture. This represents a move away from elite and culturally authoritative texts to other forms of textual evidence—for example letters, wills, and petitions;⁹⁵ inscriptions set up by private individuals;⁹⁶ and inscriptions, religious and secular, commissioned by communities of various kinds.⁹⁷ But the broadening of the source base also encompasses a shift of focus to the non-textual evidence of visual and material culture. Visual culture is an area in which great opportunities exist, but also considerable obstacles. In principle, sources such as vase-painting and sculpture might be thought to afford direct access to the physical expression of emotion in gesture and body language.⁹⁸ But scholars have no unmediated access to the non-verbal expression of ancient emotions: to link the depiction of non-verbal behaviour in ancient art to ancient concepts of emotion one typically requires warrant from linguistic and especially narrative sources, along with as much contextual information as can reasonably be obtained, as well as a thorough understanding of the iconography of the wider corpus to which the depiction belongs.⁹⁹ Recent work is now meeting these challenges.¹⁰⁰

Chaniotis’s study of the multiple ways in which the dedication of a statue provides evidence for aspects of ancient affectivity indicates another fruitful approach in this connection, one that considers the products of the visual arts not just in their own right, as evidence for the depiction of emotion, but in their wider context, as functional objects in specific physical and cultural settings: statues not only represent emotional experience, but also express emotional commitment and elicit emotional responses.¹⁰¹ This complements earlier work on the emotional dimensions of sanctuaries and other locations for ritual performance.¹⁰² Epigraphic texts, dedications, religious architecture, and the configuration of the site more generally all contribute to the creation of a shared space for emotional experience and emotional performance, a locus for the enactment of the emotions—awe, fear, wonder, respect, hope, gratitude, and the rest—on which religious experience depends.¹⁰³

Consideration of the embodied and embedded aspects of ancient emotion need not restrict itself to material evidence. Literary texts are also informative about the physicality of ancient emotions as aspects of the ways in which embodied human beings interact with the world. Most obviously, they are rich in representations of the objects, artefacts, spaces, symptoms, movements, postures, and gestures through which emotions can be expressed, symbolized, constructed, and elicited.¹⁰⁴ But they are also rich in metonymies and metaphors drawn from the embodied experience of emotion itself, from other aspects of embodied experience, and from interaction with the physical and social environments more generally. Not only do these play a fundamental role in the formation and extension of emotion concepts, but they also afford at least a degree of access to a linguistic community’s shared representations of the phenomenology of emotion.¹⁰⁵ To say “I shudder” rather than “I am afraid” is to give a more vivid and immediate sense of the emotion as a holistic, embodied experience; to present the onset of grief as the feeling of being suddenly enveloped in a cloud or a garment presents an individual’s emotion in terms of a shared cultural model of what that emotion is supposed to feel like and links it to its visible expression in body language and dress.¹⁰⁶ Similarities, where they exist, between these schemas and those found in other societies will at least partly reflect the constraints that actual physiology, symptomatology, and other features of human embodiment place on metaphors and metonymies that depend on embodiment and embodied interaction with the physical and social environments.