This thesis explores Seneca’s conceptualisation and valuation of the emotion we call ‘regret.’ Regret, as R. Jay Wallace points out in a recent wide-ranging philosophical analysis of the emotion, seems to us to be ‘utterly familiar and even natural’. Yet Stoic moral psychology has often been held to be insensitive to and dismissive of the complexities of emotions as we ordinarily experience them. Arguments proceeding from this assumption have, perhaps inevitably, maintained that the Stoic philosophy of emotions could accommodate neither a psychologically nuanced concept of regret, nor one that endows it with any of the potential for moral improvement with which the emotion is charged in the Christian tradition; Phillip Mitsis, for example, identifies in Stoic philosophy an ’emphatic rejection of the value of regret’, whilst Julia Annas argues that what ‘the ethical part of Stoic philosophy involves’ is for us ‘to accept counterintuitive conclusions such as that [‘] emotions like regret are all mistaken.’ Such approaches fail to account for the fact that, although Seneca remains rooted in Stoic philosophy, he nonetheless displays a sensitive and nuanced conception of this commonly felt and powerful emotion. By comparing Seneca’s configuration of ‘regret’ to that of the Old Stoa, early Christianity, and modernity, I seek to show that Seneca presents a reading of the emotion that builds on his Greek models whilst recognising the psychological complexities involved in the experience of regret. Moreover, as I shall argue at the close of this essay, Seneca provides an account of regret that not only remains true to the fundamental tenets of the Stoic ideal of the sapiens, but also makes regret integral to progress towards this ideal.
I begin in ??1 with a discussion of the language of regret and the precise connotations of Seneca’s paenitentia. I outline in ??2.1 the resonance of these nuances in the context of Stoic theory regarding agency and value-judgements, before considering in ??2.2 how conceptions of emotional passivity, particularly as presented in Seneca’s tragedies, interact with these theories, and potentially disrupt Seneca’s account of regret. In ??3 I discuss the relationship between regret and the sapiens, considering in ??3.1 the ideal of the supreme good as a means of obviating regret, and in ??3.2 the sage’s relationship with his own past actions. Finally in ??4 I examine how Seneca reconciles the Stoic ideal of the sapiens with his innovative advocacy of regret as having a morally corrective force.
1. Defining Regret
Close analysis of the way in which Seneca uses paenitentia, covering similar but not co-extensive ground to that which is addressed by ‘regret,’ belies the idea that Stoicism simply rejected out of hand the subtlety and value of this emotion. In recognising that the lexical labeling of emotion rarely translates precisely across languages, I hold that emotions are socially constructed, which is to say that ’emotions and their display are constituted, that is, formed and shaped, by the society in which they operate [‘] Emotions depend on language, cultural practices, expectations, and moral beliefs.’ Whilst some emotions have been shown to be universally recognisable across disparate cultures, this class is fairly limited, encompassing only ‘basic’ emotions such as happiness, anger, sadness and disgust. However, there are multiple close correspondences between the emotion referred to in Pseudo-Andronicus by the Greek ”””??, and in Seneca by the Latin paenitentia, which suggest a continuity not undermined by the linguistic shift. Nonetheless, the terms do not denote precisely the same conception of emotion, nor do either of these ‘regret’-related words correspond exactly to the connotations of the modern English ‘regret’. Accordingly, the following discussion will identify not only the points of correspondence between the Greek Stoic ”””??, the Senecan paenitentia, and our regret, but also the points of divergence.
Seneca does not offer a definition of what he means by paenitentia, but scrutiny of instances of regret in his oeuvre brings out certain common features of the way in which the word is used. These features correspond closely to the Stoic definition of regret provided by Pseudo- Andronicus, On Emotions 2 (= SVF 3.414), which states, ”””””’ ”””””””””””””””’??. This definition suggests:
(1) that regret is experienced as a distressing emotion;
(2) that regret is directed at an agent’s own actions, and that it is above all the agent’s role in bringing some event to pass that is now the cause of distress;
(3) that these actions are in the past;
(4) that they are now judged negatively.
Although modern philosophical and psychological approaches to regret are broader and less prescriptive than this, there is nonetheless a remarkable amount of continuity between this definition and what we instinctively think of as regret as we experience it in our everyday lives. The first component of this definition of ”””??, that regret is experienced as a painful emotion, can be taken uncontroversially as a necessary condition for any emotion to be categorised as regret. In both the Stoic and the modern conceptions, I take the pain entailed by regret to be both psychologically and physically manifested, and indeed there is again substantial continuity in the way in which the physical pain of regret is articulated: Rorty notes that ‘the standard feelings of regret are painful’, with the emotion ‘characteristically felt as a particular sort of painful feeling, a pang, a stab, waves of stabs’, whilst, as Graver notes, the Stoics conceived of distress (”??) ‘as an affective response’ that is, ‘one which involves a psychophysical ‘contraction’ of the psyche and is sensed as ‘biting’ or mental pain.’ Likewise, the fourth component, that regret entails a judgement that the object of regret is in some way negative, can again be taken as a necessary condition for an emotion to be categorised as regret. This does not have to be an unqualifiedly negative evaluation, since an agent can regret some aspect of their actions without regretting the consequences of those actions or, overall, the fact that they did what they did. Thus an agent could regret ending her marriage in a pique of rage, whilst being glad overall that she ended it, but she must judge her rage to have been the wrong way of going about things in order to regret it. The third component of the Pseudo-Andronican definition of regret, that the regretted actions are in the past, diverges from some modern conceptions, according to which regret is not exclusively retrospective. However, in both the philosophical literature and everyday usage, regret is most generally conceived of as concerning some past object. Therefore, the features of the definition of regret in Pseudo-Andronicus labelled here (1), (3) and (4), remain integral components of regret according to modern conceptualisations of the emotion.
However, the second component of the Pseudo-Andronican definition, that regret is directed at an agent’s own actions, covers an area less broad than that which is addressed by the modern conception of regret; as Thalberg notes, ‘I can regret events which are unrelated to my own actions; thus I might regret (the fact) that the U.S. Supreme Court is restricting First Amendment guarantees of free speech and assembly.’ The Stoic conception of regret therefore corresponds not to the entire class of feelings which are today classified as regret, but rather to that which Williams terms ‘agent-regret’. The definition of regret reported by Stobaeus in his discussion of Stoic philosophy (Ecl. 2.7.11i (102-3W)) reinforces the weight placed on the agent-relative nature of regret according to the Stoic conception. The Stobaean definition emphasises that the difference between general vexation or grief and regret is the agent’s role in having brought about the state of affairs that is regretted. Agent-regret is arguably the most powerful form of regret, and plausibly the purest. When an agent feels regret at the death of an acquaintance’s mother, that emotion will be mingled with feelings of sympathy, and possibly also other emotions, such as relief that it was the acquaintance’s mother who died and not her own. By contrast, if the agent herself had caused the death of her acquaintance’s mother, for example by running her over whilst she was driving, the regret she would feel would be significantly more persistent and all-encompassing than if she had had nothing to do with the death. This remains the case even if there was nothing she could have done to prevent it, for example if she was not drink-driving or speeding at the time of the accident, but driving carefully and fully aware of her surroundings.
Both in the scenario in which the agent was not at all involved in the death of her acquaintance’s mother, and the scenario in which she involuntarily killed her, there was nothing that she could reasonably have done to prevent it. Yet having run over her acquaintance’s mother, the agent might well find herself regretting the many factors that meant that she was in that particular place at that particular time: that she had stopped to buy a packet of crisps at the petrol station, or that the red light on the corner before the site of the accident had changed so quickly. In agent-regret, therefore, the emotion comes to encompass much more than the isolated immediate object of regret, and instead takes as its true object a causal sequence of events that are seen to have caused, or failed to prevent, the regretted occurrence. This will often correspond to the level to which the agent holds herself responsible. Having killed her acquaintance’s mother, the agent knows on some level that there was nothing that she could have done to prevent the death, and that her decision to buy the packet of crisps was causally relevant only coincidentally. Although her voluntary action in stopping at the petrol station was a causal factor in the death, the consequences of that action were neither voluntary nor forseeable. If, however, the agent had killed her acquaintance’s mother because she was drink-driving, although she still did not cause her death voluntarily, she can nonetheless be held responsible for it, since it was a foreseeable consequence of her voluntary decision to drive after drinking heavily. In this case, her regret will be correspondingly greater. Thus, having killed her acquaintance’s mother due to drink-driving, the agent might regret not only the drinks she had at the pub that night, but also the stress in her job that led her to feel the need to blow off steam at the pub, and so also the desire to earn a high wage that led her to take the high-stress job, and so also the entire chain of events and decisions in her life that led to her becoming the kind of person that desires a high wage. Therefore, although the Stoic conception of regret is not in this sense co-extensive with the modern account, it does apply to those instances of regret on the modern scheme which are most powerfully affective.
However, despite the large amount of overlap between modern conceptions of regret and the way in which ”””?? was articulated by the Stoics, the two accounts are not identical. Not only does the Stoic account make agent-responsibility a necessary condition of regret, but the modern conception has certain resonances not brought out by either Pseudo-Andronicus or Stobaeus. Modern accounts of regret tend to include a wish that things were other than they are, with Williams, for example, arguing that ‘Regret necessarily involves a wish that things had been otherwise, for instance that one had not had to act as one did’, although ‘it does not necessarily involve the wish, all things taken together, that one had acted otherwise.’ For example, an agent may regret her divorce, wishing that her relationship with her husband had not broken down, and that she had not felt the need to divorce him. Given that her relationship had broken down, however, she will not regret that she did divorce him, recognising that it was the right action to have taken in the circumstances, however much she might wish that the circumstances were otherwise.
This divergence also constitutes a difference between the old Stoic conception of regret discussed above, and Seneca’s paenitentia, which is partially semantic, and partially theoretical. Kaster links the wish that things were otherwise with the etymological root of paenitere, arguing that ‘Just as paene, ‘almost,’ is a relative notion [‘] so the displeasure of paenitere depends on a relative assessment, the assessment of what is as opposed to what ‘should be”. The wish that things were other than they actually are is not only rooted in the language of paenitentia, but also plays a key role in Seneca’s conception of the emotion. Thus at De Beneficiis 7.26.2, Seneca enjoins Liberalis to never regret having given a benefit, regardless of the ingratitude of the recipient, and characterises the one who regrets as crying out, vellem, non fecissem. Moreover, Seneca tells us that the sapiens will never feel regret because he will always recognise that he could not have acted in any way better than he did (De Beneficiis, 4.34.4). Seneca suggests, therefore, that without the wish that things had been otherwise, regret will not take hold: the counterfactual wish is an integral component, without which the emotion dissipates.
There is one final means of categorising regret that is relevant here, which is the distinction between ‘regret,’ ‘remorse’ and ‘repentance.’ The difference between regret and remorse is generally held to lie in distinctions of agency, morality and justice. To feel remorse, the agent must have acted voluntarily, and with full awareness of the consequences, so as to cause the regretted event. This is a necessary but not sufficient condition for an emotion to count as remorse, such that the regret that an agent expresses for the death of her acquaintance’s mother will not count as remorse, whereas the regret that an agent feels having killed her acquaintance’s mother through drink-driving may do, if it meets certain further criteria. Remorse must entail some recognition of interpersonal harm, whereas regret might concern solely intrapersonal harm. The drink-driving agent may regret killing her acquaintance’s mother because she will now have to go to prison, and for no other reason than the impact that this will have on her life. To feel remorse, the drink-driving agent would have to acknowledge the negative impact of her actions on others. Remorse, then, is concerned with social and ethical relations with others, with voluntarily doing things that are morally wrong and contrary to moral and societal norms, and it is on this basis that remorse is frequently classified as a moral emotion whereas regret is not.
Repentance is distinct from both regret and remorse in that it involves an urge to make amends, must be somehow performative, and is often ritualised. Within a Christian context, paenitentia is often conceptualised in terms that correspond more closely to remorse and repentance than regret; Tertullian, for example argues that ubi metus nullus, emendatio proinde nulla; ubi emendatio nulla, paenitentia necessario vana, quia caret fructu suo cui eam deus seuit, id est hominis salute (De Paenitentia 2.2). Tertullian denies the validity of the pre-Christian sense of paenitentia, and, as Strousma notes, ‘proposes to limit the meaning of paenitentia, by applying the term only to the rejection of evil actions. Doing so, he accomplishes an ethicization of the concept, based on the idea of God’s justice’. Moreover, Tertullian argues that paenitentia ought to be exhibited not only in the wrongdoer’s conscience, but also in the penitent’s ritual of exomologesis, which entailed fasting, weeping, wearing rags and ashes, and supplication (De Paen. 9.1). Tertullian was certainly correct in his assertion that the paenitentia of pagan Rome lacked the precise charge of the Christian conception; as Kaster notes, ‘a change of heart that leads one to seek purgation and forgiveness for sins … is not the paenitentia of pre-Christian Rome.’ However, the contention that regret, devoid of Christian theology, is necessarily vana is more problematic. For Tertullian, as for us, the potentially corrective value of regret is to be found in its manifestation as remorse and repentance, and the potential thereof to instigate a re-evaluation of our own actions in the wider terms of communal or universal justice. Repentance constitutes a ritualised means by which the wrongdoer can express his or her remorse and be reabsorbed into the social or religious community. Seneca, however, endows regret with a potentially corrective value that stands outside the concepts of remorse and repentance, and their concomitant valuations of interpersonal moral relationships and communal standing. Indeed, it is precisely the self-concerned, self-referential, and introspective nature of regret, distinctively opposed in this sense to remorse and repentance, that, for Seneca, means that paenitentia may constitute part of meaningful moral progress.
2. Stoic Theory
2.1 Agency and Assent
The Stoic conception of regret as specifically agent-regret clearly presupposes a concept of an agent. Given the moral charge of Seneca’s regret, this must be an agent who can be held morally responsible for his or her actions. Stoic psychological theory makes room, in a deterministic metaphysical system, for an agent’s personal responsibility for his or her actions through the theory of assent. The Stoics held the human mind to be unitary, made up of distinct faculties centralized in a single governing part. The governing part of all animals had two such faculties: impression, which is the sensation formed by an external, and impulse, which is the grasping of that impression and spur to action; the governing part of the adult human is distinguished from other animals by two further faculties, assent and reason, the latter of which renders the whole soul fully rational. Assent constitutes a mediating faculty between impression and impulse, consisting of the ability of a rational agent to evaluate the truth-value of an impression and accordingly determine whether or not to act upon it. Whilst we have no control over the impression we receive, assent, as Kahn notes, is ‘a rational action that is entirely ‘up to us’: as long as our behavior is controlled by the mechanism of rational assent, our behavior is in our own power.’ Since the Stoics held that assent to an impression was both a necessary and a sufficient condition for action, any action of a rational agent is potentially subject to moral evaluation as both indicative and formative of an agent’s moral character. Moreover, because assent is a matter of evaluation and judgement particular to a specific individual, it renders all aspects of an agent’s thought processes, opinions, and emotions open to moral judgement and constitutive of an agent’s moral character.
That voluntary action is, in Stoic moral psychology, contingent on assent, has led Charles Taylor to identify the Stoic theory of assent as ‘one source of the developing notion of the will’. Not only does assent emphasise an agent’s capacity for intentional action, but also makes it dependent on the capacity for self-reflexive mental processes. As such, the Stoic theory of assent corresponds closely to Frankfurt’s theory of the will, which holds the will to be an ‘effective desire ‘ one that moves (or will or would move) a person all the way to action,’ and identifies such a desire as a second-order volition, that is, a desire that the agent actively wants to be his will. As Inwood points out, the conceptual closeness of the Stoic account with Frankfurt’s is particularly striking in Seneca, whose focus on the self-reflexivity of the mind, ‘though hardly unique ‘ stands out for the frequency and explicitness of his interest.’ This conception of freedom of the will as having to do with a distinctive type of mental process means that regret is not eliminated by the Stoics’ deterministic philosophy. As noted above, paenitentia entails a wish that things had been other than they were. Yet superficially it seems as though such a feeling would be eliminated with determinism: if an agent’s actions are determined, then he could not have acted in any way other than he did, and hence for him to feel regret is unreasonable. However, the emphasis in both Frankfurt and Stoic philosophy on decision-making and thought processes enables both to argue for a compatibilist position in which someone may be responsible for an action even if he could not have done otherwise. For the Stoics, then, although events will necessarily happen as determined, an agent may be morally evaluated according to the way in which he submits himself to the course of things: ducunt volentem fata, nolentem trahunt (Ep. 107.10). The Stoics viewed the death of Socrates as exemplary in this respect: Socrates was fated to die, and so would have died whatever he had done to try and escape his fate. Yet since Socrates made no such attempt and instead went willingly to his death, cicuta magnum Socratem fecit (Ep. 13.14).
The example of Socrates encapsulates what it means to live in accordance with sapientia. Seneca tells Lucilius that the sapiens teaches us to not merely know, but to actually follow the gods, and to receive chance events as divine commands (Ep. 90.34), since to live in accordance with wisdom is not to live in accordance with one’s own reason, but rather in accordance with divine reason. Sapientia thus denotes the perfection of a human being’s reason, which entails complete assimilation with divine reason; for Seneca, to live in accordance with sapientia is the means to living a happy life (neminem posse beate vivere, ne tolerabiliter quidem sine sapientiae studio, Ep. 16.1). The happy life is dependent not on external factors, but on a particular type of mental activity, which Seneca conceives of in terms that closely reflect Frankfurt’s notion of the will:
sed hoc, quod liquet, firmandum et altius cotidiana meditatione figendum est; plus operis est in eo, ut proposita custodias quam ut honesta proponas. perseverandum est et adsiduo studio robur addendum, donec bona mens sit quod bona voluntas est.
It is through self-reflection that an agent can transform a first-order desire to live in accordance with reason into a second-order volition. In order to make moral progress, the agent must be unrelenting in his task, since inperfecta necesse est labent et modo prodeant, modo sublabantur aut succidant (Ep. 71.35). He must be vigilant in ensuring that the desire is operative in every decision: quotiens, quid fugiendum sit aut quid petendum, voles scire, ad summum bonum, propositum totius vitae tuae, respice (Ep. 71.2). None of the agent’s desires or actions can be conceived of as standing apart from the project of self-improvement. Only by constant self-scrutiny can an agent make all of his first-order desires conform to his second-order desire to live in accordance with reason.
Seneca conceives of regret as the result of an agent’s failure to live in accordance with reason, and the consequent mistaken attribution of value to things that are not the good. Thus at Ep. 27.2-3, Seneca characterises the objects of regret as pleasures which prove to be fleeting. The Stoic Epictetus, in his discussion of Euripides’ Medea, argues that the wish for things to be other than they are is the greatest cause of ””””””””? (2.17.18), and entreats Medea to wish for nothing but the will of Zeus (2.17.22). In wishing to change that which is outside her control, Medea is the very antithesis of the tranquillity of the Stoic sage, which Seneca characterises as a state of perfect equanimity: placido statu maneat nec adtollens se umquam nec deprimens (De Tranq. An. 2.4). Medea’s failure, therefore, is not only in judging externals (her husband, her home in Corinth) to be valuable, but also, as Epictetus notes, in not recognising ”””””””””?, ””” ””””’ ””” ””’ ”””””””’ ””””’? (2.17.21). The issue is not necessarily that externals are in themselves harmful ‘ regret is concomitant to the pursuit of them etiam si non nocent (Ep. 27.3) ‘ but rather the attribution of value to something other than the true good. According to this conception, therefore, regret is not solely concerned with the false attribution of value to externals but also with the attendant inconsistency in an agent’s system of valuation. This point is emphasised further in Ep. 90.34, which states that what the sage has brought to light (in lucem protaxerit) is the distinction between such pleasures and those things which truly merit an attribution of high value. Regret is presented here as the result of opiniones falsae, which lead the non-wise man to pursue not eternal goods, but mixtae paenitentia voluptates. Seneca therefore presents us with a conception of regret, which, although opposed to remorse in its self-centredness, is nonetheless imbued with a strong moral charge. Regret in Seneca concerns objects which are not truly the good, and so entails an ethical valuation of those things; it is directed at an agent’s false beliefs in pursuing those objects; and it is a mark of a man who is morally deficient, i.e. not yet a sapiens. Seneca’s conception of regret, therefore, is inextricably bound up with the moral worth of an agent.
2.2 Emotional Passivity
Hence, in order to avoid regret, all acts must be examined and evaluated by the mind, which must take itself as its own subject, cajoling and steering itself away from those things that will lead to regret: dicat sibi ipse: Voluptas fragilis est, brevis, fastidio obiecta, quo avidius hausta est citius in contrarium recidens, cuius subinde necesse est aut paeniteat aut pudeat (De Ben. 7.2.2). Seneca’s tragedies, emphatically focused on the psychologies of their characters, frequently dramatise agents coaxing and persuading their own minds in such a way. By contrast to the tragedies of fifth-century Athens, Seneca’s dramatic writing is centred on the extended soliloquies of its principal characters, speeches in which the characters dissect and evaluate their own plans and desires. However, these characters lack the cool evaluative capacity that Seneca in his philosophy suggests is cultivated by self-scrutiny, and as they articulate the inner workings of their minds, they vacillate, becoming increasingly conflicted: quid nunc moreris, anime? quid dubitas? Medea (988) asks herself, in strikingly similar terms to Clytemnestra at Ag. 108-9: quid, segnis anime, tuta consilia expetis? / quid fluctuaris? Seneca suggests that emotion can appear to denude the mind of its rational evaluative power:
ratio ipsa, cui freni traduntur, tam diu potens est quam diu diducta est ab adfectibus; si miscuit se illis et inquinavit, non potest continere quos summovere potuisset. commota enim semel et excussa mens ei servit quo impellitur.
De Ira 1.7.3
However, to interpret this passage as suggesting that reason can be ‘overcome’ by the passions in any strict sense is inadequate, since, as Seneca goes on to explain, for a passion to take hold the mind must admit it, giving it voluntas nostra (De Ira 1.8.1). Without rational assent, a passion cannot take hold, as non ‘ separatas ista sedes suas diductasque habent, sed affectus et ratio in melius peiusque mutatio animi est (De Ira 1.8.3). Since we have the means by which to avoid becoming subject to our passions, Seneca argues, we can be held morally responsible for the failure to do so: turpe est non ire, sed ferri et subito in medio turbine rerum stupentem quaerere: ‘huc ego quemadmodum veni’? (Ep. 37.4) Therefore, although the rational agent in the grip of a passion might feel buffeted about (commota’ et excussa), with his mind enslaved (servit), this sensation of passivity disguises the agent’s active responsibility for his own emotional experience.
Modern psychological discussions of emotion recognise the conflict noted by Seneca, that our emotions are instigated and controlled by us and yet we often feel as though an emotion has overwhelmed us by force. Averill argues that this disconnect can be explained by the potential for emotion to restrict an agent’s insight into himself and his own agency: ‘During stated emotional reactions, ego-boundaries are so narrowly drawn that the phenomenal self is perceived as the passive recipient (the Me) rather than the active initiator (the I) of the response.’ In his tragedies, Seneca depicts the destructive force wrought by emotion on the self. Seneca’s Medea presents her emotions as overpowering forces with their own agency: quo te igitur, ira, mittis, aut quae perfido / intendis hosti tela? (916-17) The militaristic language used here recalls De Ira 1.8.2, where Seneca compares the mind threatened by anger to a city under siege, and emphasises the necessity for the enemy (hostis) to be stopped at the gates, nam cum intravit et portis se intulit, modum a captivis non accipit. In the tragic figure of Medea, Seneca plays out the consequences of failure. Medea sees herself as caught between mutually exclusive identities of angry wife and loving mother, pulled this way and that: ira decessit loco / materque tota coniuge expulse redit (927-8). In the final moments of the play, Medea declares in matre si quod pignus etiamnunc latet, / scrutabor ense viscera et ferro extraham (1012-3), visualising the violence of her emotion made corporeally manifest. Medea’s remark bears resemblance to Jocasta’s declaration in Seneca’s Oedipus, as she goads herself on to commit suicide by stabbing herself in the womb: hunc, dextra, hunc pete / uterum capacem, qui virum et gnatos tulit (1038-9). Medea’s anger threatens to viscerally excise all vestiges of her identity as mother, at cost not only to the boundaries of her self, but even to the integrity of her very body. In Averill’s terms, under the influence of passion her ‘ego-boundaries are so narrowly drawn’ that she cannot occupy the middle ground: her anger fragments her identity, and as she is subjected to its force she vacillates, alternately identifying herself wholly with her emotion and seeing herself as ineffectual in the face its overwhelming power, a common victim, along with her sons, to the violent force of anger.
Both Seneca and Averill present such breakdowns of the boundaries of the self as only one script of (quasi-) emotional passivity. Averill notes that there are certain emotional experiences of which ‘I’ cannot be said to be an ‘active initiator’, that is, those reactions caused by ‘biological predispositions’, such as ‘startle to a sudden noise, feeling uneasy when looking over the edge of a tall building, and attacking the source of pain’. Seneca argues in remarkably similar terms:
omnes enim motus, qui non voluntate nostra fiunt, invicti et inevitabiles sunt, ut horror frigida aspersis, ad quosdam tactus aspernatio; ad peiores nuntios subriguntur pili et rubor ad improba verba suffunditur sequiturque vertigo praerupta cernentis.
De Ira 2.2.1
Seneca, however, does not allow that such reactions should be categorised as emotional, arguing that nihil ex his, quae animum fortuito impellunt, adfectus vocari debet; ista, ut ita dicam, patitur magis animus quam facit (De Ira 2.3.1). Yet, as Inwood points out, some of the reactions Seneca categorises as purely bodily ‘clearly involve considerable cognitive and social sophistication’. Recoiling from cold water differs from blushing at bad language in that the former is a physiological reaction to a tangible physical stimulus, whereas the latter is a physiological reaction to an intangible stimulus that must be processed mentally. Thus whereas a reaction such as recoiling from cold water could reasonably be assumed to be experienced universally, blushing at bad language is a socially constructed response and one which varies among different members of the same society’ as is often said, bad language has to be pretty bad to make a sailor blush. Seneca, unlike Averill, thus includes in this category emotion-type responses that cannot be attributed to our evolutionary history. The distinction Seneca draws between emotions and these reactions is not that the latter are wholly physical, in some way bypassing our cognitive processes, but rather that these are responses that we cannot control by our will (qui non voluntate nostra fiunt). Since emotions are always the result of the mind’s active assent, such reactions cannot be described as emotions, and as such, there is no truly emotional experience for which an agent cannot be held responsible. An agent always has the power to prevent an emotion from taking hold, and as such to maintain the rational evaluative capacity by which to assess his or her actions.
3. Regret and the Sapiens
3.1 The Ideal of the Supreme Good
The man who has succeeded in sublimating all of his desires to the supreme good will not act inconsistently, since his volitional principle remains constant. It is by living without reference to the supreme good that we come to regret our decisions, since antequam impleatur, incerta mentis volutatio est; cum vero perfectum est, inmota illa stabilitas est (Ep. 71.27). The sapiens will never wish that he had acted otherwise, because he has synthesised all of his actions and desires to a single unwavering principle. As Stobaeus points out (ecl, 2.7.11m (113W)), regret occurs when the agent comes to see some past action as having been in error, thus entailing false assent (”” ””””), of which the sage will never be guilty. Seneca argues:
non mutat sapiens consilium omnibus his manentibus, quae erant, cum sumeret; ideo numquam illum paenitentia subit, quia nihil melius illo tempore fieri potuit, quam quod factum est, nihil melius constitui, quam constitutum est; ceterum ad omnia cum exceptione venit: ‘si nihil inciderit, quod impediat.’
De Beneficiis, 4.34.4
Hence, regret on the Stoic account is less about the consequences of one’s actions, and more to do with the flawed deliberative processes that led to those actions. The Senecan notion that adherence to a single evaluative principle precludes regret bears notable resemblance to John Rawls’ argument that a rational agent will never regret having followed a rational plan, ‘at least in the sense that he later believes that at the time it would have been more rational to have done otherwise’, since ‘he does what seems best at the time, and if his beliefs later prove to be mistakes with untoward results, it is through no fault of his own.’ Both the Senecan and the Rawlsian account conceptualise regret as most fundamentally directed at an agent’s own rationality; comprised of the mind evaluating its own evaluations, regret consists in a highly cerebral interiority, directed, unlike remorse, not at externals but at the inner workings of an agent’s own mind.
However, to harmonise all of one’s decisions to a single deliberative principle will only preclude an agent later coming to regret those decisions if his perspective remains constant. Thus Williams argues that Rawls’ account of regret ‘implicitly ignores the obvious fact that what one does and the sort of life one leads condition one’s later desires and judgments. The standpoint of that retrospective judge who will be my later self will be the product of my earlier choices.’ That is, it assumes that there is ‘some currency of satisfactions,’ according to which we could refer all of our decisions quite neutrally and from an impartial perspective, which Williams argues implies ‘an external view of one’s own life, as something like a given rectangle to be filled in’. According to this argument, since we can never transcend our subjective viewpoint, and since our future selves are always conditioned by our present decisions, ‘the perspective of deliberative choice on one’s life is constitutively from here. Correspondingly the perspective of assessment with greater knowledge is necessarily from there,’ and hence ‘I cannot ultimately guarantee from what standpoint of assessment my major and most fundamental regrets will be.’ The Stoic account of the summum bonum attempts to provide precisely such an impartial perspective.
Seneca explicitly problematises the view ‘from here’, pointing out that, in the visual realm, tricks of perspective deceive the eyes:
nihil esse acie nostra fallacius non tantum in his a quibus subtiliter pervidendis illam locorum diversitas submovet, sed etiam in his quoque quae ad manum cernit. Remus tenui aqua tegitur et fracti speciem reddit; poma per vitrum aspicientibus multo maiora sunt; columnarum intervalla porticus longior iungit.
Naturales quaestiones, 1.3.9
Although Seneca’s point here is about the physical realm and its relationship to our subjective visual perspective, it would not be too great a leap to make these problems metaphoric of the corresponding problems with subjective moral perspective. Indeed, Socrates had made precisely this comparison in Plato’s Protagoras, arguing that heterogeneous motivational factors can be assessed by reference to a common feature according to which they differ only quantitatively. Incentives and disincentives for a certain course of action can be measured by a single scale, such that a unit of pain can be quantified as the negative equivalent to a unit of pleasure (Protagoras 356b-c). However, pleasures, like many externals, are potentially illusive: ”””””””’ ”””””””””, ”””’ ””?: ‘ ‘??; (Protagoras 356c) When an agent does wrong, it is precisely because he lacks the impartial perspective that Socrates’ science of measurement would afford him. The Socratic influence on Stoic thinking is well documented; Philodemus, for example, tells us that the early Stoics were perfectly happy to be referred to as Socratics. Thus, whilst there are obvious divergences with the Senecan point of view ‘ most significantly in Socrates’ advocacy of a hedonistic value-system ‘ there is substantial continuity with Seneca’s recognition of the illusiveness of perspective ‘from here’. Like Socrates’ scale, the thesis of the supreme good establishes a ‘currency’ to which an agent can refer all of his actions that is constant regardless of viewpoint. It is precisely this, that it is not subject to variation according to perspective, which makes the summum bonum unique, and as such can properly be said to obviate regret.
For Williams, such a conception of motivational factors would preclude there being individual agents with distinct characters, ‘in the sense of having projects and categorical desires with which that person is identified’. It is this quality of differentiated character, he argues, that ‘gives substance to the idea that individuals are not inter-substitutable,’ and thus is a key basis on which friendship, and other such attachments, rest. Moreover, Williams maintains that ‘such things as deep attachments to other persons will express themselves in the world in ways which cannot at the same time embody the impartial view ‘ yet unless such things exist, there will not be enough substance or conviction in a man’s life to compel his allegiance to life itself.’ Likewise, unless such things as projects and attachments exist, regret is precluded: an agent who has no attachment to her husband, and does not value the project of marriage, will not regret her divorce. Seneca does not suggest, however, that it is in this way that the sapiens will have no regrets, but explicitly states that even the sage will feel the blow of losing friends or children, nec enim lapidis illi duritiam ferrive adserimus (De Constantia 10.4). Rather, the sapiens will overcome such a blow. Thus Seneca offers an account of friendship that accommodates both an impartial perspective and an individual’s attachments to others. The sapiens is self-sufficient, but quamvis se ipso contentus sit, amicis illi opus est (Ep. 9.15). Seneca conceives of the good of friendship as something that is independent of external contingencies, since there is no true absence from a friend if he is present to the mind of the agent (Ep. 55.9-11). It has been argued that the villa at Vatia in Ep. 55 is a metaphor for the underworld. According to this interpretation, even the death of a friend will not impede the self-sufficiency of the sapiens, nor cause him to regret his friend’s death.
As Seneca explicates at De Ben. 4.34.4, the sapiens does everything with reservation, according to the condition si nihil inciderit, quod impediat. Thus Seneca explains that remembering his deceased friends brings him pleasure: habui enim illos tamquam amissurus, amisi tamquam habeam (Ep. 63.4-7). In his friendships, the sapiens recognises the transience of mortal life, and forms his attachments always bearing in mind that he may one day lose them. This does not entail that he does not form attachments, merely that he recognises the contingency of those attachments on fate, and thus avoids any conflict between his commitment to living in accordance with an impersonal rational principle and his attachments in the world. Thus the Senecan account of friendship demonstrates that the supreme good offers a perspective that is impartial, but not impersonal. An individual agent can have attachments particular to his individual character, just so long as he acknowledges the contingency of those attachments on fate, taking his attachments to be leviter, rather than valde, expetenda (Ep. 75.11), and recognises that the supreme good is wholly interior. Therefore regret, rooted in the mistaken valuation of externals, is precluded, but the fundamental tenets of social life remain intact.
3.2 Continuity of the Self
The sapiens, therefore, will not feel regret for what he has done, since his deliberative processes will always be in accordance with reason, which offers him an unchanging perspective by which to evaluate his actions. However, the sapiens will not always have been a sapiens, and prior to achieving this status, he will have done things on the basis of flawed deliberation. The emotional experience of agent-regret presupposes that the agent has a sense of self that endures through time: to feel regret for past actions, the present self must be able to recognise its causal responsibility for those actions. Thus John Locke argues that ‘Wherever a man finds what he calls himself, there, I think, another may say is the same person’ Whatever past actions it cannot reconcile or appropriate to that present self by consciousness, it can be no more concerned in that if they had never been done.’ (An Essay Concerning Human Understanding 2.27.26). As Sorabji notes, ‘Locke uses some distinctively Stoic terms in describing the acknowledgment of past actions as one’s own. His talk of our reconciling and appropriating certain past actions reminds us of the Stoic talk of oikei??sis’. Indeed, Locke’s account bears close resemblance to Seneca’s conception of a single self which endures through the discontinuous bodies an individual might be said to have over the course of a lifetime (Ep. 121.15-16). Seneca presents the life of each individual as unitary, emphasising the interconnectivity of different developmental stages, and the endurance of a singular and identifiable self.
In Williams’ terms, this individual agent can be said to have a distinct character on the basis of the projects and desires with which he is identified. Yet although the sapiens may retain some of the projects and desires of his former self, the reasoning processes by which he affirms them will be utterly changed. As discussed above, the Stoics held this process of assent to be constitutive of an agent’s character. In becoming a sage there is such a discontinuity of character that the sapiens can no longer be truly identified with the agent he was before. Hence, it would be nonsensical for the sapiens to feel regret for anything in his pre-sapiens life, since in this fundamental respect, he is no longer the person he was before. By contrast, the agent who has not undergone the disjunction in selfhood entailed by becoming a sage will have a sense of self that endures through time. Thus, for the vast majority of agents, past deliberative processes remain sufficiently identifiable as the agent’s own that they can ‘reconcile or appropriate’ them with their current selves, and so, regret them.
The conception, in both Locke and Seneca, of reconciling or appropriating something to oneself is highly self-reflexive, consisting of the self’s recognition of its own continuity, that ego tamen idem sum (Ep. Mor. 121.16). In Locke’s terms, a person consists of ‘what he calls himself’. This reflexivity situates the self’s auto-evaluative capacity centrally within the Senecan conception of subjectivity. In regret, the self is preoccupied with a negative evaluation of itself, and painfully so, since iniucunda est paenitendae rei recordatio (De Brev. Vit. 10.2). Thus, Seneca argues, any man who has in the past committed vices, necesse est memoriam suam timeat (De Brev. Vit. 10.4). The emotional experience of regret necessitates a consistent self, evaluating past actions against a normative ideal. Thus, when Medea declares paenitet feci, she immediately abjures her regret: misera? paeniteat licet, / feci. voluptas magna me invitam subit (989-91). To feel regret is painful; it focuses the mind on its own deliberative processes; it requires no declaration. Medea’s self-construction is so fractured and determinedly outward-facing that she is incapable of the requisite subtlety, self-reflexivity and interiority.
4. Seneca and the Value of Regret
That regret is necessarily a painful emotion constitutes a problem for orthodox Stoic philosophy, since, as Cicero reports, Chrysippus believed that there was something inherently inappropriate about distress, and that, making a lexical link between ”?? and ”?, aegritudinem ”’?? Chrysippus, quasi solutionem totius hominis appellatam putat (Tusculanae Disputationes 3.61). The problem that this causes for a Stoic conception of regret cannot be dismissed by arguing, as Graver does, that although the sapiens will never feel distress, since he or she will never judge an evil to be present, he or she ‘is not barred from believing, contrafactually, that it would be appropriate to be pained at a genuine evil if one were present.’ Rather, it is inconceivable that there could be a circumstance under which it would be appropriate for the sapiens to feel regret, since the Stoics characterised regret as a form of distress, and thus, as Sorabji argues, whilst ‘it can indeed be appropriate to the novice’s circumstances, if it acts as a spur to improvement ‘ The mistake, from the Stoic point of view, is to suppose that contraction or sinking is ever appropriate absolutely.’ Cicero recounts that, according to the Chrysippean account, even if the objects of distress are truly bad, any agitation of the soul is vitiosus, since the disturbing effect of distress is categorically opposed to the measured ideal of the sage (Tusc. Disp. 4.61). Sorabji argues on this basis that regret ‘must, then, remove the novice still further from being a sage.’
Seneca, however, diverges from the orthodox Stoic account in suggesting that although the attainment of wisdom will necessarily eliminate regret, paenitentia can be a positive force in the attainment of that wisdom:
non est tamen quod existimes ullam aetatem aptiorem esse ad bonam mentem quam quae se multis experimentis, longa ac frequenti rerum paenitentia edomuit, quae ad salutaria mitigatis adfectibus venit. hoc est huius boni tempus; quisquis senex ad sapientiam pervenit, annis pervenit.
Seneca’s suggestion that paenitentia is a means by which an agent can attain a healthy mental state has hitherto been under-interrogated as an example of emotionally intelligent Stoicism. That he describes this regret as longa ac frequenta demonstrates that Seneca conceives of regret as part of an ongoing and unrelenting process of self-evaluation, enabling an agent to utilise his past mistakes to bring all of his mental processes into alignment, and thus gain victory over his own mind. This reflects Seneca’s tendency to address his moral philosophy to the proficiens, rather than the sapiens: ad imperfectos et mediocres et male sanos hic meus sermo pertinet, non ad sapientem. huic non timide nec pedetemptim ambulandum est (De Tranq. An. 11.1). By contrast, the agent who is not yet a sage must proceed with caution, constantly evaluating his own actions and being prepared to recognise that they may have been wrong. This constitutes a fundamental difference in the ideal way of life of the sapiens and the proficiens, which entails that there are behaviours appropriate to one that are wholly inappropriate to the other. By drawing this distinction, Seneca is able to present a practical moral philosophy that recognises that the proficiens cannot merely imitate the life of the sapiens and hope thereby to become wise, but must live his life cognisant of the reality that he has not yet attained wisdom. Thus Seneca argues that regret, although inappropriate to the sage, nonetheless brings the proficiens not further from, but closer to, becoming a sapiens.
Seneca’s commitment to a positive valuation of regret is not, moreover, limited to his philosophical writings, but is also suggested in his tragedies, above all in the Agamemnon. From her first appearance onstage, Clytemnestra struggles to decide between the two courses of action she deems available to her, both of which she deplores: fluctibus variis agor, / ut, cum hinc profundum ventus, hinc aestus rapit, / incerta dubitat unda cui cedat malo (Ag. 138-40). Like Medea, Clytemnestra is torn between identification with two roles she sees as mutually exclusive, those of avenging mother (158-9; 162-73) and faithful wife (110-1; 239-41). Clytemnestra declares that her love for Agamemnon is enough for her to alter her plans, and recognises the potential for her to recover virtue through paenitentia:
sed nunc casta repetatur fides,
nam sera numquam est ad bonos mores via:
quem paenitet peccasse paene est innocens.
This is the first of a series of speeches Clytemnestra exchanges with Aegisthus in which the final lines are pithy sententiae that aphoristically summarise the argument and abstract from the specific dramatic situation to wider moral lessons. These lengthy speeches could almost be pared down to their final lines, in the manner of the stichomythic exchange between Clytemnestra and the nutrix at 145-54, which consists almost entirely of pairs of competing sententiae. The highly crafted, declamatory style of these exclamations is far from naturalistic, which together with their applicability beyond the specific circumstances of the play suggests that these axioms can to some extent be interpreted independently of the main dramatic action.
Moreover, Clytemnestra’s pronouncement on the value of regret at Ag. 243 adapts an aphorism from the explicitly didactic work of the elder Seneca (est quaedam proxima innocentiae verecundia, praebere se legibus, Controv. 7.8.6), suggesting that the axiom might fruitfully be interpreted as having an instructive moralising intention in its dramatic incarnation. Seneca’s modification of his father’s aphorism, although subtle, dramatically recasts the suggested means to quasi-innocence. Whereas the lesson to be learned from the elder Seneca’s instruction is, fittingly for a collection of exemplar legal cases, that exoneration may be attained by recourse to the social institution of the law, the younger Seneca suggests that it may be attained by the self-centred, introspective means of paenitentia. For the elder Seneca, by acknowledging his offence and accepting the law’s punishment, the wrongdoer may attain something morally akin to innocence. The elder Seneca characterises the emotional experience accompanying this behaviour as verecundia, an emotion closer to our shame than regret, and emphasises the sociality of the emotion: the wrongdoer recognises the impact of his wrong on another, and agrees to atone for this violation of the social contract by accepting punishment in accordance with social obligations. Thus, the elder Seneca advocates the wrongdoer repent; by submitting to legal process, he reasserts the commitment to the communal values that he disavowed by committing his crime.
By contrast, for the younger Seneca’s Clytemnestra, regret is not so much about offering recompense for her actions, or attempting to undo the harm she has caused, as it is about moral improvement, the recovery of the chastity and fidelity she abandoned in her adultery with Aegisthus (109-13). Whereas the elder Seneca advocates repentance through participation in legal process, Clytemnestra’s regret demands only reflection and the determination to self-improvement. Clytemnestra, at this point in the play, might still follow a path other than that taken by her Aeschylean predecessor. Thus Seneca presents retrospective regret as preventative of wrongdoing in the future, thanks not to the externally imposed machinations of justice, but the change of moral character that regret entails. Contrary to Tertullian’s accusation, this is no vana paenitentia; nor does Seneca reject the value of this emotion, as Stoicism has generally been taken to do. Rather, Seneca offers an account of regret that is fully integrated with the Stoic theory of assent, that abjures the idea of emotional passivity, and that is reconciled with the Stoic ideal of sapientia, and living in accordance with the supreme good. Seneca’s advocacy of regret as morally corrective, has, I believe, been insufficiently recognised for two reasons: firstly, the failure to separate the ideal behaviour of the sapiens from the behaviour appropriate to the proficiens, and hence the interpretation of the Stoic sapiens as a model, rather than an aspiration, for the proficiens; and secondly, the heavy colouring of our modern valuations of regret by Christian theology. Distinct from repentance, both through submission to legal process and Christian exomologesis, by its introspective nature, Seneca’s paenitentia nonetheless carries a promise of redemption, ut exeam melior (Ep. 68.14). To feel regret, on Seneca’s account, is to recognise your own moral shortcomings in such a way as to overcome them, and in this sense, quem paenitet peccasse paene est innocens.