The Starting-Points for Knowledge

Chrysippus on How to Acquire and Fortify Insecure Apprehension

In: Phronesis
Simon Shogry Brasenose College, University of Oxford 61466 Oxford UK

Search for other papers by Simon Shogry in
Current site
Google Scholar
Open Access


This paper examines some neglected Chrysippean fragments on insecure apprehension (κατάληψις). First, I present Chrysippus’ account of how non-Sages can begin to fortify their insecure apprehension and upgrade it into knowledge (ἐπιστήμη). Next, I reconstruct Chrysippus’ explanation of how sophisms and counter-arguments lead one to abandon one’s insecure apprehension. One such counter-argument originates in the sceptical Academy and targets the Stoic claim that insecure apprehension can be acquired on the basis of custom (συνήθεια). I show how Chrysippus could defend the possibility of custom-based apprehension, while also denying that there is custom-based knowledge.


This paper examines some neglected Chrysippean fragments on insecure apprehension (κατάληψις). First, I present Chrysippus’ account of how non-Sages can begin to fortify their insecure apprehension and upgrade it into knowledge (ἐπιστήμη). Next, I reconstruct Chrysippus’ explanation of how sophisms and counter-arguments lead one to abandon one’s insecure apprehension. One such counter-argument originates in the sceptical Academy and targets the Stoic claim that insecure apprehension can be acquired on the basis of custom (συνήθεια). I show how Chrysippus could defend the possibility of custom-based apprehension, while also denying that there is custom-based knowledge.

1 Introduction

In a series of rich fragments recorded by Plutarch (De Stoicorum Repugnantiis [St. Rep.] 1035F–1037A), Chrysippus recommends that students of Stoic philosophy hear not only arguments in favour of Stoic doctrine, but also those against. Indeed, he regards exposure to such contrary arguments as indispensable for the acquisition of the systematic knowledge (ἐπιστήμη) on which our happiness depends.

In section 2, I explore how Chrysippus’ account of knowledge leads him to offer this pedagogical proposal. With the Stoics more generally, he maintains that, if an agent knows that p, then her endorsement of the truth of p is unchangeable by argument (ἀμετάπτωτον ὑπὸ λόγου): the knower will never retract her commitment to p in the face of dialectical scrutiny, in part because she is able to diagnose the faults in any argument against p. So for those presently without knowledge that p but seeking to attain it, some consideration of the arguments on the other side will be necessary. How else could one come to be prepared to defend the truth of p against any possible challenge?

For Chrysippus, however, there is more to knowing p than being unable to be argued out of p: p must be true, of course, but also the knower must stand in a direct and unmediated relation to p, such that the clarity with which p is represented, and the causal process by which it enters the knower’s mind, ensure that p could not be false. This commitment is registered in the Stoic analysis of knowledge as a kind of grasp or apprehension (κατάληψις), and in their analysis of apprehension as an assent to an apprehension-prompting or kataleptic impression.1 An impression is kataleptic if it is true, clear, and ‘of such a kind that it could not have come from what is not the case’ (Sextus, M 7.248). So every apprehension involves assenting to a kataleptic impression, and every case of knowledge is also a case of apprehension.

Crucially, however, not every apprehension is a case of knowledge. In cases of what the Stoics call ‘weak’ or ‘insecure’ apprehension, the agent’s grasp is such as to be dislodged by contrary dialectical pressure. Chrysippus’ pedagogical proposal is therefore designed specifically to help fortify the student’s insecure apprehension of p. In being exposed to the arguments against p and coming to learn their flaws, the pupil approaches the robust condition of knowledge, in which her endorsement of p is so secure that no possible counter-argument could move her to disavow it.

This picture implies that a secure, knowledgeable grasp differs from an insecure one not in what is assented to, but rather in how or with what degree of rational strength the assent is given. Insecure apprehension is thus suitable to serve as the starting-point for knowledge (principium scientiae: Cicero, Academica [Acad.] 1.42). Weak grasps are commonplace, the Stoics think, whereas knowledge, given the extremely demanding dialectical skill it presupposes, is found only in the Sage. Famously, in the view of the Stoics’ critics, this wise person will be ‘rarer than the phoenix’ (Alexander, De Fato 199.18), but Chrysippus’ proposal sheds light on how the formidable knowledge of the Sage could begin to be acquired. By examining and disarming plausible counter-arguments against p, the non-Sage comes to fortify his insecure apprehension that p and upgrade it into the secure knowledge that p.2

Section 3 turns to the origins of insecure apprehension, with a view to better understanding why the Stoics assume it is easy for ordinary agents to obtain. Here I focus on a surprising but seldom discussed Chrysippean commitment, that insecure apprehension can be acquired ‘on the basis of custom’ (κατὰ τὴν συνήθειαν). This claim should strike us as odd. Given that the Stoics characterise apprehension as a special type of true belief—one which is in some sense guaranteed to be true and could not be false—how could it ever be acquired on the basis of custom? After all, custom is notoriously volatile, and so by accepting whatever is habitually held in one’s community, the student is just as likely to believe truths as falsehoods. How could the epistemic accomplishment that is apprehension ever have such a capricious source as this? Indeed, as I explain in section 3.1, there is ample evidence to suggest that the Stoics’ Academic opponents—beginning with Arcesilaus—criticised custom on these very grounds, and, moreover, that Chrysippus regarded this criticism as particularly forceful. So Chrysippus will have urgent need to defend the possibility of custom-based apprehension against this type of challenge—to preserve the coherence of his epistemological theory no less than the confidence of his students in the face of this particular Academic counter-argument.

In section 4, I reconstruct how Chrysippus could explain the compatibility of both assenting to p on the basis of custom and achieving apprehension of p. Here Chrysippus seems to have relied on a distinction between (i) the psychological reason for the agent’s assent to the kataleptic impression that p, and (ii) the features that make the resulting belief a case of apprehension. That p conforms to custom can be a case of (i)—the consideration that explains why the agent accepted p—but never (ii) an apprehension-maker. Rather, the apprehension-makers are restricted to the intrinsic qualities of the kataleptic impression: its clear phenomenology and unmediated causal connection with the world. So in assenting to the kataleptic impression that p on the basis of custom, the non-Sage fails to consult these intrinsic qualities in the course of forming his belief that p. Even so, this oversight does not undermine the criterial status of the impression to which assent is given, and so the resulting belief still counts as apprehension (albeit of the insecure and weak type). I show that this reconstruction is consistent with all the current interpretations of the kataleptic impression, although it does rule out certain internalist assumptions about the kataleptic impression’s justificatory role.

Finally, in section 5, I examine why Chrysippus denies the possibility of custom-based knowledge. Assenting to p on the basis of custom could never be a case of knowing p, since the fact that p conforms to common practice offers no considerations that the agent could appeal to in an attempt to rationally defend p. So while custom sometimes produces weak apprehension and is reliable to this extent, it could never support knowledge. The Sage, for this reason, will have replaced custom as the basis for her assent to the kataleptic impression that p with a richer appreciation of why p must be true in the circumstances. For this purpose, she will rely on her understanding of dialectic and logic and also of ethics and physics. I connect these doctrines with the contents of Chrysippus’ work Against Custom and explain how, as a long ancient tradition attests, the arguments of this treatise were later appropriated by Carneades and turned against the foundations of Stoic epistemology.

The significance of my project is both exegetical and philosophical. By unpacking Chrysippus’ verbatim remarks on how to acquire and fortify insecure apprehension, we gain an untainted vantage point on early Stoic epistemology at a critical stage of its development, when it is being systematised by its greatest theorist and no longer innocent of sceptical criticism. Chrysippus’ words throw light on several controversial and difficult points of Stoic theory: how the kataleptic impression serves as a criterion of truth; why insecure apprehension is easily acquired and why ordinary agents are prone to give it up; and, in general, how insecure apprehension falls short of knowledge but can be upgraded into it. What emerges is a clearer picture of how insecure apprehension functions as a starting-point for knowledge. Furthermore, as the shape of early Stoic epistemology comes into sharper focus, we can better appreciate what the Academic sceptics took themselves to be responding to, and what resources in Stoic theory they could exploit for their own destructive purposes. My aim here is not to prove that the Stoic account can withstand any possible line of sceptical attack (in the manner of the Sage himself!), but rather to fill in some gaps in our understanding of Chrysippus’ epistemology, so that the confrontation with the sceptics and other critics can be put into its proper context.

2 Contrary Arguments and Knowledge

We begin with a fragment of Chrysippus on the dialectical practice of arguing for contrary positions (τὸ πρὸς τἀναντία διαλέγεσθαι). Plutarch introduces the fragment as follows:

[T1] He [Chrysippus] says he does not entirely disapprove of arguing for contrary positions, but he recommends that it be used cautiously, just as those in the courts dissolve the plausibility [of the contrary arguments] without advocating for them. ‘For’, he says, ‘doing this is suitable for those who induce suspension of judgement on everything, and it contributes to their goal. But, on the contrary, for those who instil knowledge, on the basis of which we will live consistently, it is suitable to instruct pupils in the elements and to fortify them from beginning to end. In cases when it is opportune, it is suitable to mention the contrary arguments as well, destroying their plausibility just as is done in the courts’.3

PLUTARCH, St. Rep. 1035F–1036A

Here Chrysippus distinguishes two different ways of engaging with contrary arguments, rejecting one and approving the other for use in pedagogical contexts. This is because arguing for contrary positions is an activity that assumes different forms depending on the overall aim of the philosopher so arguing.

On the one hand, arguing for contrary positions is useful for a sceptic like Arcesilaus, whose goal is the universal suspension of judgement.4 By constructing two equally plausible arguments, one for p and another for p*, where p and p* cannot be true together, the sceptic provides his interlocutor with considerations of equal strength for contrary positions. Alternatively, in response to the interlocutor’s argument for p, the sceptic will propound an equally plausible counter-argument for p*. In either situation where this kind of argumentation is employed, the result is ἐποχή, the condition of the rational mind in which it is unable to make a determination or assertion concerning the truth of p (Galen, Best Way of Teaching 1.4–7).5

As distinct from this sceptical practice, Chrysippus juxtaposes a second way of engaging with contrary arguments, consisting in a teacher’s mentioning such arguments to students at the ‘opportune’ moment, and claims that such an exercise will be useful for philosophers aiming to ‘instil knowledge, on the basis of which we will live consistently’ (T1). These are of course Stoic philosophers, with Chrysippus referring here to his school’s standard formulation of the human goal (τέλος): living consistently or in agreement (τὸ ὁμολογουμένως ζῆν). According to Stoic doctrine, in achieving the human goal and living consistently, one is happy. This accomplishment depends solely on the possession of knowledge, the Stoics hold, since virtue is necessary and sufficient for happiness, and virtue either consists in or supervenes on knowledge.

But how exactly, on Chrysippus’ view, does the teacher’s cautious presentation of contrary arguments help the pupil to advance toward the all-important goal of knowledge? What kind of pedagogical exercise is Chrysippus recommending in T1, and how precisely does it differ from Academic argument in utramque partem? These questions will become more tractable once we set out the Stoic account of knowledge, captured in the following definition:

[T2] Knowledge is a grasp that is secure and unchangeable by argument.6 (Arius Didymus in Stobaeus, Eclogae [Arius in Stob., Ecl.] 2.73.19–21)

T2 defines knowledge as a certain kind of grasp (or apprehension, κατάληψις). A grasp, in turn, the Stoics understand as a special type of true belief—one which, because it consists in an assent to an impression that could not be false, is guaranteed to depict the world as it really is. To know something, then, is to grasp it—and thus assent to it and believe it—in the particularly firm way characteristic of the Sage. So whereas sceptical argument for and against p seeks to leave the mind at a standstill with respect to p, unable to affirm it as true because of considerations of equal plausibility on both sides, the Stoic practice adopts the opposite goal: not checking or suspending the mind’s assent but strengthening and fortifying its rational foundations.

I emphasise ‘rational’ here because it is not just any unchangeable grasp that will count as knowledge for the Stoics. It is specifically a grasp that is unchangeable by argument or by reasoning (ἀμετάπτωτον ὑπὸ λόγου). So if one were to grasp some fact p and continue to maintain it simply out of an intransigent refusal to consider any argument against p, clinging to p as an unchallenged article of faith, it would not be a case of knowing p (even assuming that p is true and the object of a grasp).7 The Stoics characterise knowledge as unchangeable in the face of active dialectical scrutiny, since it rests on the agent’s sensitivity to the considerations supporting p as well as her ability to diagnose the flaws in any argument against p.8

What seems to motivate Chrysippus’ proposal in T1, then, is the thought that, to acquire security and stability in one’s grasp of p, so that it becomes ‘unchangeable by argument’ and a piece of knowledge, it is necessary to examine not only arguments for p but also those against—and indeed not straw-men, but contrary arguments with some prima facie plausibility, for these are the arguments that could undermine one’s rational attachment to p. As we will see, it is possible for non-Sages who apprehend p to give up this grasp when confronted by an argument for p* (where p and p* are deemed incompatible) whose defects they cannot discern and whose plausibility they cannot dispel. Such a grasp is insecure and susceptible to being dislodged by contrary argument—an assent to p that is not a case of knowledge, but ignorance:

[T3] For ignorance is a changeable and weak assent.9 (Arius in Stob., Ecl. 2.111.20–1)

While the interpretation of T3 is controversial, it is clear that, at least in some contexts, the Stoics are prepared to classify the non-Sage’s grasps also as cases of ignorance (e.g. Cicero, Acad. 1.41–2).10 However, Chrysippus is optimistic that, with enough exposure to contrary arguments, coupled with an account of their defects, the student will learn how to defend what she has apprehended and help turn her ignorance into knowledge. Her grasp of p will stabilise, because initially plausible arguments against p will now be seen to be inadequate and no longer give her reason to revise her commitment to the truth of p. In this way, the examination of contrary arguments under appropriate pedagogical supervision—the exercise Chrysippus recommends in T1 and there distinguishes from the suspension-inducing argumentation of the Academic sceptic—is an activity that supports the role the Stoics attribute, more generally, to the logical part of philosophy, as the wall guarding the fruits of apprehension against the attacks of sophisms and plausible counter-arguments (Sextus, M 7.22–3; Philo, De Agricultura 15–6; Cicero, Fin. 3.72).

So then, for the Stoics, both Sages and non-Sages assent to kataleptic impressions and thereby achieve apprehension, but only the Sage’s assent is counter-factually robust enough to be maintained in the face of any possible line of contrary argument. With the aid of Chrysippus’ proposal in T1, Stoic pedagogy aims to replicate this condition in the student, by exposing the pupil to arguments not only for, but also against, the claims he presently grasps but, strictly speaking, is ignorant of. One implication is that a knowledgeable grasp differs from an ignorant one not in what is assented to, but rather in how or on what grounds the assent is given.11

Chrysippus further explains the vulnerabilities of insecure apprehension in his On Ways of Life, excerpted verbatim by Plutarch:

[T4] The contrary arguments and plausibilities in contrary views are to be exhibited not at random but with caution, lest those diverted by them lose hold of their apprehensions, since they cannot adequately understand the solutions and they apprehend in a way that is easily shaken.12

Plutarch, St. Rep. 1036D–E

T4 confirms that the teacher’s cautious presentation of arguments against p has a specific aim: to strengthen and fortify the weak grasp the pupil already has of p. According to Chrysippus, those who apprehend p ‘in a way that is easily shaken’ (εὐαποσείστως) are liable to give up their apprehension in the face of plausible arguments against p.13 This propensity to be diverted from their apprehensions will presumably be found among newcomers to Stoic philosophy and those without formal training, and is attributed to the fact that they ‘cannot adequately understand the solutions’ (T4): they cannot see how the contrary argument goes wrong and so cannot extricate themselves from the grip of its plausibility (cf. Cicero, Acad. 2.46). We can infer, then, that one way of ‘destroying the plausibility’ of contrary arguments—the task Chrysippus sets the teacher in T1—consists in showing the student where their faults lie and how they can be solved, so that these arguments no longer threaten his weak apprehension.

In this way, the contrary arguments ‘are to be exhibited not at random but with caution’ (T4): they are to be mentioned only in a context where the teacher can expose their deficiencies and explain how their conclusions can be resisted. This is what T1 calls the ‘opportune’ moment. Furthermore, we learn from Plutarch that Chrysippus recommends that contrary arguments should ‘always be presented with an indictment of their falsity and without being advocated for’ (St. Rep. 1036D).14 An argument is false, in Stoic parlance, when at least one of its premises is false, or if it is inconclusive (Diogenes Laertius, 7.79), and so ‘indicting’ an argument’s falsity will involve laying out why it has one or both of these features (why it is unsound or merely invalid, in our terminology).

To characterise the correct way of handling contrary arguments in the classroom, Chrysippus uses the language of the courtroom.15 For instance, in the passage from St. Rep. 1036D just cited, the teacher’s explanation of the falsity of a contrary argument is called an ἔνδειξις—a legal term for one type of ‘indictment’ filed against debtors and exiles who have trespassed into prohibited areas (Thür 2006). Witness also the reference here and in T1 to a kind of advocacy (συνηγορία) that Chrysippus considers out of place in the Stoic teacher’s presentation of a contrary argument. In Athenian trials, litigants are sometimes joined by a ‘supporting speaker’ (συνήγορος) to help plead their case and bolster their contentions to the jury.16 If the Stoic pedagogue is not to play the analogous role in the classroom, then he should not appear to endorse the contrary argument himself and, moreover, should forego any impassioned language or vehement appeals in its defence. In this way, the presentation of contrary arguments does not risk confusing the student and leaving them unsure about where the Stoics ultimately stand.17 On Chrysippus’ model, then, the Stoic teacher resembles the litigant himself—one who is particularly cool and level-headed in argument, mentioning the opposing case only when there is an opportunity to dispatch its plausibility and thereby shore up his own position.18

With the benefit of such exercises, in which contrary arguments are first exhibited and then shown to be false, the student also progresses toward the possession of the virtues the Stoics associate with dialectic.19

[T5] [The Stoics say] dialectic itself is necessary, and a virtue which includes others as species. Non-precipitancy is knowledge of when one should and should not assent. Uncarelessness is a strong rational principle against the likely, so as not to give in to it. Irrefutability is strength in argument, so as not to be carried away by argument into the contradictory.20

Diogenes Laertius, 7.46–7

The strength which constitutes irrefutability (ἀνελεγξία) is, as we will now expect, a kind of rational strength. The irrefutable person does not refuse to accept the conclusion of a contradictory argument out of obduracy or spite, but because she can spot its false-making features and so appreciate, on rational grounds, why it is inadequate. This virtuous condition of mind will be supported by uncarelessness (ἀνεικαιότης), ‘a strong rational principle against the likely’—i.e. against the plausible (cf. Diogenes Laertius, 7.75)—and informed by the global knowledge of when assent should and should not be given, which constitutes the virtue of non-precipitancy (ἀπροπτωσία). Such knowledgeable agents, according to another early Stoic text,

[T6] are unable to be refuted and are self-sufficient in apprehending propositions, since they are able to refute the obstructing argument and are strong when faced with the contrary. For it is necessary for them to be unmoved by cross-examination and to assent guardedly when faced with contrary positions.21

P.Herc. 1020, col. 110, lines 13–27

Again, virtuous people are not ‘unmoved by cross-examination’ because they refuse to play the dialectical game (as e.g. Thrasymachus does at the end of Republic 1). Rather their strength lies in a rational power to disarm the plausibility initially displayed by the contrary argument. In T1 and T4, then, Chrysippus tells us something about how the condition of the knowledgeable dialectician could begin to be acquired. Rational resistance to contrary arguments cannot come about without some prior exposure to these arguments and their defects.

We can now appreciate how the Chrysippean method of engaging with contrary arguments, aiming as it does at knowledge rather than universal suspension, differs in form from Academic practice. Chrysippus and his sceptical opponents agree that plausible arguments move the mind to assent, and consequently proceed on the assumption that to change what the mind affirms as true one must also change what strikes it as plausible. This presupposition guides the sceptic’s strategy for inducing ἐποχή. In response to an argument for p, the sceptic proceeds by constructing not just any argument for p*, but one that is equally plausible and forceful as the one just given for p, with no in-principle limit on the number of iterations, so that the inquiry could always continue. By contrast, this assumption of equipollence (to employ a Pyrrhonist term) is nowhere to be found in Stoic philosophy, and accordingly Chrysippus’ pedagogical proposal is not so unbounded. Unlike the sceptics, the Stoics have concluded that there are decisive arguments in favour of certain views—the tenets of Stoic doctrine—and that arguments to the contrary, although initially plausible, can eventually be shown to be flawed and silenced. To this end, Chrysippus stops short in T1 and T4 of proposing that students be trained in the construction of contrary arguments.22 Instead, Chrysippus thinks that, for the purpose of strengthening their insecure apprehensions, it is enough for pupils to be exposed to such arguments and to examine them cautiously, under the supervision of an experienced Stoic teacher who can immediately indict their false-making features and thus remove their prima facie plausibility. Chrysippus thus restricts the consideration of contrary arguments to contexts where they can be shown to be flawed, where the dangers they pose to weak apprehension can be safely removed.

However, there is no indication in T1 and T4 that Chrysippus regards this kind of engagement with contrary arguments as sufficient to instil knowledge. This is because the Sage’s knowledge that p does not rest merely upon her ability to diagnose the flaws in the arguments against p, but also on a systematic appreciation of how p fits into a larger body of facts, both those within its domain and in other parts of philosophical inquiry. Thus, just after T2, knowledge is characterised as a ‘system’ (σύστημα) of secure grasps, a term with deep resonance in Stoic theory, serving to indicate a unified complex structure.23 So further training, taking different forms, would seem to be required in order to integrate one’s grasps into such a structure.24 Even so, the examination of contrary arguments will still be necessary for instilling knowledge, since without the ability to disarm their plausibility, such arguments could divert one’s token grasps, in which case the systematic artifice could never be constructed.

3 How to Divert Insecure Apprehension Based on Custom

To better understand the motivations behind the Chrysippean proposal in T1 and T4, let us construct some concrete examples of contrary arguments that could lead the student to abandon her insecure apprehensions. Further explaining his remarks in T4, Chrysippus immediately goes on to say:

[T7] … since indeed those who apprehend on the basis of custom—both perceptual things and the other things that come from the senses—easily give up these apprehensions when diverted by both the questions of the Megarians and by others more numerous and more forceful.25

Plutarch, St. Rep. 1036E

Notice first that two things are said to be apprehended ‘on the basis of custom’ (κατὰ τὴν συνήθειαν): ‘both perceptual things and the other things that come from the senses’.26 Here Chrysippus seems to be invoking the well-attested Stoic distinction between perceptual and non-perceptual apprehension (Diogenes Laertius, 7.52). Given that propositions are the objects of apprehension (as we saw above in T6), we would expect that a ‘perceptual thing’ is a proposition grasped with the help of the sense-organs and the formation of a perceptual kataleptic impression, e.g. « This table is brown », when the perceiver is seated at a brown table and looking at it under normal perceptual conditions.27 Assenting to this kind of impression, and thereby achieving perceptual apprehension, is something that could be done by custom or habit (κατὰ τὴν συνήθειαν), according to Chrysippus, since he presupposes that we tend to affirm the deliverances of the senses by default, and that, barring abnormal perceptual circumstances, what the senses tell us about the world is true (cf. Sextus, M 7.424; Cicero, Acad. 1.42). Another example would be our grasp of « It is day » (Epictetus, Discourses 1.28.1–4). Through the ordinary exercise of the senses, we regularly entertain the perceptual kataleptic impression that « It is day » and assent to it by habit, more or less unreflectively. In both examples, the use of the sense-organs marks the resulting apprehension as perceptual, and the role of custom or habit in inducing assent justifies the claim that it arises κατὰ τὴν συνήθειαν.

In general, συνήθεια denotes a shared habit: a practice or way of thinking that has become habitual in a given community.28 Accordingly, I will sometimes render συνήθεια as ‘common practice’ in the course of discussing T7, but more often as ‘custom’ and ‘habit’. As we will see, both Chrysippus and his sceptical opponents assume that one συνήθεια common to human beings in general is the propensity to accept unreflectively the deliverances of the senses.

T7’s second kind of συνήθεια-based apprehension, of ‘the other things that come from the senses’, will be a wide category indeed, in light of the Stoics’ foundationalist epistemological assumptions. The propositions so grasped will likely include theological facts (as the Stoics see them), such as « The gods are providential » and « The gods exist ». The Stoics suppose that there is some genetic story linking our apprehension of these non-perceptual facts to our earlier perceptual encounters with the world, but the precise way this is supposed to work does not matter for present purposes.29 In any event, these theological truths are taken to be both suitable objects of apprehension and propositions which common practice will recommend. T7 suggests, then, that it is not only demonstration (ἀπόδειξις) but also custom that leads to non-perceptual apprehension.30 One could come to grasp « The gods are providential » from explicit Stoic instruction and argument, or simply from growing up in a pious community and accepting this fact as a matter of habit.

Chrysippus’ claim in T7 that custom is one mode of acquiring perceptual and non-perceptual apprehension will engage us repeatedly in what follows. But before taking a closer look at apprehension-by-custom and how it is dislodged, we first ought to clarify the more general matter of what it means to give assent ‘on the basis of’ (κατά) something.

According to Chrysippus, each act of assent is caused by the individual character of the agent’s mind (Cicero, De Fato 41–3; Aulus Gellius, Noctes Atticae 7.2.11–2). Thus he holds that the wider cognitive condition of the agent—including the beliefs she holds and the memories she has access to—controls which of her impressions she affirms as true.31 Moreover, in some cases at least, it is possible to identify which particular component of the agent’s psychological make-up induces her to accept an impression. This is the task we find Chrysippus engaged in when he speaks of apprehension ‘on the basis of custom’ in T7: here he intends to identify the element of the student’s psychological profile that is responsible for her assenting to the kataleptic impression that p. Following recent commentators, we can call this her reason for assenting to p, which in this case is simply that p conforms to custom.32 However, since the language of reasons is hotly contested and not always perspicuous, we should try to say something more about these reasons for assent found in the Chrysippean theory.

First, Chrysippus does not hold that such reasons will always be rehearsed explicitly by the agent—or spoken silently in an internal monologue—in the lead-up to assent. Nonetheless, the reasons for assent are still determined by the psychology of the agent and potentially recovered after the fact. For instance, if asked ‘why do you believe that the gods are providential?’, the student could answer, ‘because that is what I was taught to be true growing up in my community’. In this case, her answer would specify the consideration underpinning her judgement that « The gods are providential » is true, even if it was not consciously entertained prior to making that judgement. Of course, the subject could be deceived about what in fact has triggered her act of assent, and so bringing the reason to light could be more difficult than this simplified case suggests. Even so, if we peered into the subject’s mind, as it were, we could learn her reason for taking the impression that p to be true.33

Second, and as one may have already surmised from the case of this student, the agent’s reason for giving assent to p need not be the correct reason or the best reason for taking p to be true. Rather, it is the element of the agent’s psychological make-up that proves efficacious in inducing her to affirm p as true—more precisely, the aspect of her rational mind that is so efficacious, since, according to Chrysippus’ broader psychological theory, assent is an exercise of the adult human mind (διάνοια), which has no non-rational parts. So for the agent who is new to Stoic philosophy and not yet a Sage, that p conforms to custom will often be her reason for giving assent to the kataleptic impression that p, as she lacks any deeper understanding of why p is true in the circumstances and p is consistent with her habitual thought and practice. But, for Chrysippus, this is clearly not the best and most stable reason for taking p to be true, and so what the pupil must learn, ultimately, is the practice of the Sage, who never assents to p simply because p is dictated by custom. We will return to this issue in sections 4 and 5.

3.1 The ‘Questions of the Megarians’ and Other, ‘More Forceful’ Arguments

Now that we have a better handle on Chrysippus’ notion of apprehension-by-custom, we must address his claim in T7 that these apprehensions are insecure because they are liable to be abandoned under pressure from ‘the questions of the Megarians and others more numerous and more forceful’. This claim adds further precision to the remarks in T4, where Chrysippus has identified contrary arguments (ἐναντίοι λόγοι) in general as capable of dislodging weak apprehension. So, reading the reports together, it is clear that Chrysippus wishes to draw a distinction among contrary arguments (thus Babut 2004, 141), where some are ‘more forceful’ (δυναμικώτερος) than others, and the ‘forcefulness’ in question is simply the prima facie plausibility of the contrary argument: the power it has to persuade the student to revise their commitment to a proposition they have come to believe and grasp on the basis of custom. Megarian arguments—which were typically posed in a series of questions, with the interlocutor’s answers forming the premises and conclusion—display less of this power than numerous other arguments, Chrysippus says in T7, but they still have some. So how could these puzzles pose a threat to the student’s insecure, custom-based grasps?

The answer will depend on which particular Megarian sophisms Chrysippus has in mind in T7, and here scholars have proposed a number of possibilities. The Stoics characterise sophisms, in general, as false arguments or pseudo-arguments displaying plausibility, whose false-making and problematic features are not immediately obvious (cf. Sextus, PH 2.229; Galen, Affections and Errors of the Soul 5.72–3).34 Specifically, Chrysippus may be thinking in T7 of the Sorites, its variant the Bald Man, and / or the Horned Man, three sophisms standardly associated with the Megarian school.35 These arguments, if their plausibility is not destroyed as Chrysippus recommends in T1, could undermine the pupil’s rational attachment to facts he has previously been led by custom to apprehend, e.g. « One grain of wheat is not a heap », « This man is bald » (while pointing at the Dalai Lama) and « Humans lack horns ». Another possibility is Diodorus Cronus’ argument against loco-motion (Sextus, PH 2.242, 2.245, 3.71).36 If left unresolved, this sophistic argument could divert the untrained student from his grasp of any proposition relating to bodies changing place, and so from a wide swath of facts he has unreflectively affirmed on the basis of habit. Or perhaps Chrysippus has in mind the Megarian Stilpo, whom he harshly criticises elsewhere (Plutarch, St. Rep. 1036F). Stilpo is known to have propounded an argument against the possibility of predication that is reminiscent of the Late-Learner’s paradox in Plato’s Sophist (251a–c).37 So if its prima facie plausibility is not removed by the Stoic teacher’s indictment of its flaws, Stilpo’s argument could threaten effectively all of the pupil’s insecure, custom-based grasps, since even the simplest perceptual apprehension (e.g. of « This table is brown ») relies on predication.38

The state of our sources prevents us from determining exactly which, if any, of these possibilities Chrysippus is envisioning in T7, but we need not make such a determination for present purposes. My aim in laying out these possibilities is to provide a sense of the forcefulness of the Megarian puzzles and to illustrate the—varying, but generally limited—power they have to persuade the student to abandon his insecure, custom-based grasps. Having encountered e.g. the Horned Man or Stilpo’s anti-predication argument, only the most credulous pupil, it seems, would be moved to revise his beliefs and give up his weak apprehensions in the manner outlined above. Instead, we can imagine that for most students these contrary arguments would not succeed in dislodging their endorsement of « Humans lack horns » or « This table is brown », but would rather spark puzzlement concerning the sophism itself and how exactly it goes wrong.39 For instance, what makes the puzzle’s premises seem true? Do they admit of ambiguity? Does the conclusion really follow from these premises, or only seem to? The student may not have access to this logical vocabulary, but it seems fair to say that puzzlement of this kind, rather than the straightforward acceptance of the sophism’s counter-intuitive conclusion, is the more likely response. If this is right, then perhaps it is this feature which makes the Megarian puzzles less forceful in Chrysippus’ eyes than numerous other arguments (T7).

Even so, there will still be need for a teacher to explain these puzzles’ flaws and for students to learn their solutions, as Chrysippus recommends in T1 and T4.40 Even if the student is not so gullible as to be shaken from her custom-based apprehension of (e.g.) « One grain of wheat is not a heap » on the first time hearing the Sorites, nonetheless she must come to appreciate exactly what makes this puzzle fallacious—something which is far less obvious than the falsity of its conclusion—because in understanding its defects, the student advances toward the robust ‘strength in argument’ constituting the virtue of irrefutability (T5) and is closer to becoming ‘unmoved by cross-examination’ (T6) in any context where the interlocutor is employing sophistic reasoning.41

It now remains for us to consider the ‘more forceful’ arguments that Chrysippus contrasts with the Megarian puzzles in T7, whose prima facie plausibility poses a greater threat to the student’s custom-based grasps. Which arguments could these be? Building on existing work, I will offer two suggestions.

First, suppose that a student has apprehended « The gods are providential » on the basis of custom, that is, for the reason that it is commonly accepted in his community. Our poor student is then waylaid by a visiting Epicurean fanatic, who puts to him an argument for the contrary conclusion: « It is not the case that the gods are providential ». Roughly, the Epicurean’s premises could be that (1) according to our shared preconception the gods are blessed and immortal, and (2) these qualities are incompatible with providential care (cf. Cicero, Nat. 1.44–5; Diogenes Laertius, 10.76–7; and Epicurus, Principal Doctrines 1). Now, insofar as his endorsement of « The gods are providential » rests on custom, the student would have no rational resources with which to resist this contrary argument and so no way to destroy its plausibility. We can readily imagine, however, that with the benefit of a Stoic education he could come to learn (among other things) why (2) is false, and thus fortify his custom-based grasp against this particular Epicurean threat. Admittedly, this suggestion is somewhat speculative, as our sources do not directly attest to Chrysippus’ views on the plausibility of this contrary argument. But it fits the bill as one forceful enough to shake an unprepared pupil from her custom-based grasp of a non-perceptual, theological truth—unless, of course, she has already examined it in the classroom and been made aware of its false-making features.

A second possibility is better documented, and one I will be pursuing in the remainder of the paper.42 This is a contrary argument originating in Arcesilaus’ Academy, which targets the general reliability of custom and seeks to undermine the very idea of custom-based apprehension. As part of their efforts to refute the existence of the kataleptic impression, the Academics are said to attack custom as a whole (omnis consuetudo), ‘which it is their intention to render obscure’ (Cicero, Acad. 2.42).43

The basic idea here is the caprice and volatility of custom. For convenience, let us call this Academic argument the Volatility Argument. Here the Academics contend that if one forms the belief that p simply because p conforms to custom, then there is no guarantee that one will come away with a true belief rather than a false belief. This is because, they argue, the impressions recommended by custom for assent could always be false, and hence are not suitable to ground the epistemic achievement the Stoics call ‘apprehension’. For instance, since the senses sometimes lie, i.e. contribute to the formation of false perceptual impressions, the agent who is habitually guided by a blanket trust in the senses will be led to approve falsehoods on this basis. And when it comes to theological matters, it seems that following custom could induce a Hellenistic Greek to endorse « The gods engage in murder and theft »—a falsehood, by Stoic lights—just as easily as « The gods are providential », which the Stoics hold to be true. More generally, then, with the Volatility Argument, the Academics seek to establish that no impression recommended by custom for assent will count as kataleptic, since, as a class, these impressions could be false and so lack the requisite guarantee of truth. Consequently, there is no such thing as apprehension-by-custom.

What is particularly intriguing about the Academic Volatility Argument is its special interest to Chrysippus, who, remarkably, seems to have propounded a version of it himself. It is therefore likely that he would have regarded it among the ‘more forceful’ contrary arguments in circulation. But whereas the Academics infer from the volatility of custom that custom is incompatible with apprehension, Chrysippus draws a different conclusion. As we will see in section 5, Chrysippus authored a lengthy work Against Custom, which seems to have covered similar ground as the Academic critique and to have presented the dangers of relying on custom to form one’s beliefs. But the goal of this work, I will argue, is not to show the impossibility of custom-based apprehension, as the Academics conclude, but rather the impossibility of custom-based knowledge.

For Chrysippus, assenting to p on the basis of custom could never be a case of knowing p, since, as we have seen, the Stoic conception of knowledge imposes strict conditions on the character of the knower and, in particular, on the character of her assent. Everything that is known is assented to ‘securely and stably’ (ἀσφαλῶς καὶ βεβαίως: Arius in Stob., Ecl. 2.112.1–2) and grasped in a way that is ‘unchangeable by argument’ (T2). But assenting to p simply because p adheres to common practice leaves the agent with no rational resources to defend p when challenged. By contrast, one who knows p will always be able to dispel the plausibility of any argument against p and to explain why p is true. This accomplishment presupposes, then, that the knower has replaced custom as the basis for her acceptance of p with a richer, more systematic account.

We will examine these points in more detail in section 5, but first we must get clear on how Chrysippus could have defended the possibility of custom-based apprehension. After all, Chrysippus is explicitly committed to this possibility in T7, and it is clear that the Academics mean to target it with their Volatility Argument, as I have explained above: this argument is designed to show that no impression recommended by custom for assent could ever be kataleptic, and therefore that it is impossible to achieve apprehension-by-custom. Moreover, if left unrefuted, the Volatility Argument is capable of dislodging the student’s custom-based grasps, and, unlike many of the other contrary arguments we have discussed in this section, it is not obvious what form Chrysippus’ response would take. Indeed, scholars have not given much attention to the Stoic rejoinder to this particular Academic challenge, and it remains unexplored how Chrysippus could defend the use of custom as a route to apprehension, while simultaneously denying that it could ever support knowledge. Our task now is to reconstruct what leads Chrysippus to adopt this delicate position on the epistemic limitations of custom.

4 A Defence of Custom-Based Apprehension

It is curious, given the Stoic account of apprehension, that Chrysippus regards assenting to p on the basis of custom as compatible with apprehending p. But rather than dismissing Chrysippus’ remark in T7 as confused, I think we can learn much about Stoic epistemology by taking this commitment seriously. To defend the possibility of apprehension-by-custom, Chrysippus seems to be relying on a distinction between (i) the reasons leading the agent to assent to the impression that p and (ii) the factors which make the resulting belief that p a case of apprehension that p. For convenience, I will speak of (i) the assent-inducers or reasons for assent and (ii) the apprehension-makers, respectively.

For Chrysippus, I will argue, the apprehension-makers need not be the assent-inducers. A non-wise agent can achieve apprehension by many different psychological routes, including one in which their assent is not informed by the factors that render the underlying impression kataleptic. In such a case, though the apprehension-makers did not guide the agent’s assent, the belief so formed still enjoys a privileged connection with the world and so is fit to serve a foundational epistemological role.

Accordingly, we ought to understand the role of custom in T7 as the assent-inducer but not the apprehension-maker. Chrysippus accepts that apprehension could not fulfil its foundational role were it to be supported by custom alone. This is one reason why he wrote a work against custom (Diogenes Laertius, 7.198), and no Stoic ever dared to put forward custom as a criterion of truth.44 But Chrysippus is not committing himself to such a view in speaking in T7 of ‘apprehension on the basis of custom’. He is rather highlighting the reasons in the non-wise agent’s personal psychology that lead her to accept an impression that is in fact related to the world in a way that guarantees its truth. On Chrysippus’ theory, apprehension maintains its special connection with reality because the locus of the apprehension-maker is the kataleptic impression, namely, its clear phenomenology and direct causal origin in what it represents. By contrast, the reasons for assent are entirely psychological: they are the set of considerations and beliefs underpinning the subject’s evaluative stance in relation to a given impression (section 3). Since these reasons can fail to respond to or be informed by the intrinsic qualities of the kataleptic impression, Chrysippus means to reject the further claim that they determine whether the resulting belief is a case of apprehension. So if custom is understood as an assent-inducer—and not as an apprehension-maker or feature of the kataleptic impression—then Chrysippus is not reneging on the guarantee of truth promised in apprehension-by-custom.45

Once this basic distinction is in place, Arcesilaus’ Volatility Argument (section 3.1) can be argued to conflate the apprehension-makers with the assent-inducers. This is because the Volatility Argument turns on the claim that all the impressions which custom leads one to accept are on an epistemic par, with none more reliable than any other. But if the criterial status of an impression can be specified independently of the psychological basis for its acceptance, then this claim is unwarranted. For Chrysippus, what makes a belief a case of apprehension is the intrinsic qualities of the impression assented to, not how or on what basis the belief is formed. So even if Chrysippus concedes that custom is an unstable basis for assent, in that it induces one to form true beliefs on some occasions and false beliefs on others, this volatility does not undermine the reliability of the kataleptic impressions it sometimes leads one to accept.

Of course, there can be overlap in the assent-inducers and apprehension-makers, as we find in the apprehensions of the perfected Sage, i.e. in her pieces of knowledge (T2). She assents with an awareness of the apprehension-makers, i.e. the clarity and non-defective causal history of the kataleptic impression, and, as we will see in section 5, is positioned to defend the truth of these impressions if challenged. But this follows not from the fact that her belief is an apprehension but rather from the perfected condition of her belief-set as a whole, i.e. her wisdom, including her possession of the dialectical virtues, which enable her to lock on to the features of her impressions that render them kataleptic by ‘referring them to right reason’ (Diogenes Laertius, 7.47).46 By contrast, awareness of the apprehension-makers is not required for the non-Sage to end up with apprehension.

That apprehension can come about through custom is a feature, not a bug, of the Stoic theory, insofar as it helps to explain how apprehension functions as a starting-point for knowledge. Every case of knowledge is a case of apprehension (T2), but not vice versa. In order to be open to non-Sages (Cicero, Acad. 1.41–2; Sextus, M 7.152), and capable of fortification through the pedagogical exercises set out in T1 and T4, apprehension must be able to arise in the absence of the secure assent on which knowledge rests. Indeed, the assent-inducers in the case of knowledge cannot come apart from the features which make the assent a case of knowledge (from the ‘knowledge-makers’ if you will). One cannot acquire knowledge—as one can acquire apprehension—through just any psychological route, since one does not know p unless one assents to p in the secure and stable way that is unchangeable through rational means. But if this were a requirement for apprehension as well, then the Stoics could not analyse knowledge as an upgraded form of apprehension—as an existing apprehension that p, backed by the rational strength capable of disarming any consideration against the truth of p—since such rational strength would already be present in the initial apprehension.

I anticipate that, for readers versed in the scholarly debate over the kataleptic impression, aspects of my reconstruction may seem jarring, in that my proposal apparently rules out otherwise plausible accounts of the apprehension-maker. But in fact the picture I offer here is consistent with all the current interpretations of what the apprehension-makers are, i.e., with internalist, externalist, and hybrid interpretations of the kataleptic impression. For the internalist, all and only kataleptic impressions possess maximally clear and detailed phenomenology.47 For the externalist, kataleptic impressions are distinguished by their non-defective, direct causal origin in reality (but need not have any introspectible feature that marks off their privileged status).48 And for the hybrid theorist, kataleptic impressions have unblemished phenomenology because of their non-defective causal history.49 Note that all these interpretations share the assumption that a belief is an apprehension in virtue of the intrinsic character of the impression assented to, whether that be its phenomenology, causal history, or both, and not the psychological route by which the agent came to her act of assent. So they should all, in principle, accept the possibility that the apprehension-maker is not the assent-inducer, at least for the non-Sage.

Admittedly, Chrysippus’ notion of apprehension-by-custom seems on first glance to be most congenial to the externalist interpretation. For if the apprehension-maker is simply the kataleptic impression’s causal history, and this causal history is not necessarily accessible to the agent, as the externalist maintains, then it is easier to see how the assent-inducer can come apart from the apprehension-maker. However, even on the internalist and hybrid view, it is conceded that non-wise agents can fail to attend to the distinctive phenomenology of a kataleptic impression. For instance, the kataleptic impression that « It is day » could in fact possess unblemished clarity and detail, depicting day-time in all its glory, without the agent consulting such phenomenology in coming to assent to it: he could accept that « It is day », at least in part, out of habit. In such a case, it is open for the internalist and hybrid interpreter to take the view that the belief that « It is day » would still count as an apprehension because of the phenomenological features that the impression intrinsically has, even if they were not explicitly consulted in the run-up to assent.

Here one might object that my account fails to do justice to the status of the kataleptic impression as a criterion of truth. A point stressed especially by internalist interpreters against the externalist is that the Stoics understand the kataleptic impression as a ‘prodelic’ criterion, one that discloses facts to the agent immediately, through its own intrinsic features.50 Such interpreters infer from this claim that assent to p is justified only when it is based on conscious awareness of the intrinsic features of the kataleptic impression that p (on the apprehension-makers, in my jargon). Thus Casey Perin writes,

in its role as the criterion of truth the [kataleptic] impression p must provide me with a certain kind of reason for believing that p that I can recognize as such and on the basis of which I come to believe and sustain my belief that p. (2005, 389)

More tentatively, Tamer Nawar suggests that the clarity which attends every kataleptic impression is both a phenomenological mark and an ‘evidential or justificatory property’, i.e. a feature of the impression that gives ‘good grounds’ for assent to its content (2014, 10). But as I have construed them, the non-Sages in T7 achieve apprehension that p without consulting the intrinsic features of the underlying kataleptic impression that p. For these agents, the apprehension-makers are not the assent-inducers; rather, habit is. But since they do not rely on a prodelic criterion of truth to form their belief that p, it would seem that their assent to p is not justified.

As I read him, Chrysippus would offer a mixed assessment of the epistemological achievements of the non-Sages in T7. In one sense, they have shirked their epistemic duty by neglecting to rely on the apprehension-makers in the course of forming their belief. For indeed, if this policy were applied in all cases and never superseded, they would never obtain knowledge, as in the case of knowledge the assent-inducers must track the apprehension-makers (see above and section 5). But, for all that, Chrysippus need not accept the further suggestion that, in giving assent to a kataleptic impression on the basis of custom, the non-Sage is doing something entirely unjustified or wholly without ‘good grounds’. To be sure, the Academics assume that assent to p is justified only when it is maximally firm and secure (and consequently they suspend judgement universally).51 But Chrysippus need not agree. Some of the non-Sage’s acts of assent could be more justified than others, even though none of them is a case of knowledge.

To this end, Chrysippus could maintain that, when a non-Sage is induced by custom to assent to a kataleptic impression, the resulting belief is justified with respect to the goal of achieving apprehension. The basis for justification would be the objective features of the kataleptic impression assented to, which are really there and present in the impression, even if not considered in the lead-up to assent. After all, if we wish to invoke the notion of justification in this context, nothing prevents us from construing it in the manner of what is now called ‘propositional’ rather than ‘doxastic’ justification, and so derived from the intrinsic and objective quality of the underlying impression and not from the state of mind of the agent.52 So in grasping e.g. « It is day » on the basis of habit, the student has formed a belief that, because of the character of the kataleptic impression she has assented to, is related to the world in the right way to count as justified. Of course, more work would be needed to achieve knowledge of this fact—and different assent-inducers, to be sure—but for the sake of grasping it habit is a justified psychological route.

In sum, then, the caprice and volatility of custom does not undermine the objective qualities of the kataleptic impressions it sometimes leads one to accept. Custom and habit, at least for beginners to Stoic philosophy, will often be the path by which they affirm the truth of an impression whose phenomenological and / or causal features render it criterial. By drawing on this account, Chrysippus has available at least the rudiments of a reply to the Academic Volatility Argument against apprehension-by-custom.

5 Why the Sage Does Not Rely on Custom

Chrysippus maintains, then, that the belief that p counts as apprehension that p not because of the psychological reason for the assent to p. However, Chrysippus must also think that, for the purpose of acquiring knowledge (ἐπιστήμη), we ought not to adopt the general policy of believing what is customary. On the one hand, an unexamined deference to custom produces opinion, insofar as it sometimes leads one to accept non-kataleptic impressions, and it is a fundamental Stoic tenet that knowledge is incompatible with the possession of even a single opinion.53 On the other hand, apprehension-by-custom, although it consists in an assent to a kataleptic impression, is still a case of ignorance, insofar as it is a ‘a changeable and weak assent’ (T3; cf. Cicero, Acad. 1.41–2)—the kind of grasp that is insecure and liable to be abandoned under contrary dialectical pressure (T7). In this section, I will argue that the Sage’s knowledge is based, in part, on an awareness of the limitations of custom. For the Sage, custom is never an assent-inducer or reason for assent. Instead, she assents to the kataleptic impression that p in full appreciation of the apprehension-makers for p and with an understanding of how these guarantee the truth of p. I will develop this proposal with reference to the ancient evidence on Chrysippus’ books Against Custom (Κατὰ τῆς συνηθείας: Diogenes Laertius, 7.198). It is likely here that Chrysippus would have presented the anti-custom considerations that underpin the Sage’s knowledge, and explained how custom—including the widely-shared habit of unreflectively accepting the deliverances of the senses—brings about opinion and ignorance.

We learn something about the contents of these Chrysippean works from the hostile tradition that grew up around them. According to Plutarch, it is inconsistent for Chrysippus both to argue against custom and to fret about the dangers posed by the Megarian questions and other contrary arguments in T7:

[T8] You fear that these may divert certain people from apprehension, but that you yourself would disturb one of your readers—even though you wrote so many books against custom, where in your ambition to outdo Arcesilaus you added whatever you could think up—this you did not foresee? For indeed it is not merely the arguments against custom that [Chrysippus] employs either, but, as if being moved by emotion in a trial, he frequently exclaims that custom talks nonsense and is idle chatter.54

Plutarch, St. Rep. 1037A

Plutarch alleges that, more so than the arguments of the Stoics’ adversaries, Chrysippus’ own Against Custom is liable to dislodge the novice from their weak apprehension. Furthermore, Plutarch is not alone in T8 in attributing the tenor of this Chrysippean treatise to a personal connection with Arcesilaus:

[T9] And finally … when he was with Arcesilaus and Lacydes in the Academy, [Chrysippus] studied philosophy with them. For this reason, he argued both against custom and on behalf of it.55

Diogenes Laertius, 7.183–4

T9 reports that some kind of Academic influence inspired Chrysippus to author Against Custom and its counterpart work On Behalf of Custom (Diogenes Laertius, 7.198), but this need not imply a period of Chrysippus’ youth when he personally embraced Arcesilaus’ brand of scepticism.56 After all, as we saw in T1, the cautious examination of contrary views is a legitimate Stoic practice, different in form and aim from Academic argument in utramque partem. Moreover, Chrysippus needs to explain how Arcesilaus’ critique of custom goes too far in denying the possibility of custom-based apprehension. It is therefore tempting to suppose that On Behalf of Custom contained Chrysippus’ defence of this possibility, with a view to refuting Arcesilaus’ Volatility Argument (see sections 3.1 and 4), whereas his Against Custom explained the impossibility of custom-based knowledge.57 At any rate, since we have identified a plausible philosophical motivation for these books, we ought not to give much credence to Plutarch’s hypothesis of petty φιλοτιμία on Chrysippus’ part (T8).

But even if Against Custom is sincerely written in support of orthodox Stoic doctrine, it nevertheless proved useful for later sceptics such as Carneades, who appropriated Chrysippus’ arguments for the purpose of attacking Stoic epistemology. Plutarch reports that, though Chrysippus was admired by later Stoics, they also believed that:

[T10] Carneades says nothing original but, by using the arguments to the contrary which Chrysippus devised, seeks to attack Chrysippus’ own views … On the subject of his publications against custom, [these Stoics] go so far in their vainglory and boastfulness to assert that the arguments of all the Academics together rolled into one are not worth comparing with those that Chrysippus composed to discredit the senses.58

Plutarch, St. Rep. 1036B–C

Though unsympathetic and sensationalised, these texts plausibly characterise Chrysippus as eager to explain the epistemic dangers arising from improper reliance on custom and habit, including the habit of too easily trusting the senses. According to the early Stoics, we form non-kataleptic impressions while dreaming, drunk, or manic (Cicero, Acad. 2.47–54, 88–90; Sextus, M 7.247–8, 402–7) as well as in unintoxicated waking life, e.g. when trying to see something from far away (Sextus, M 7.258, 424; cf. Plutarch, St. Rep. 1057A–B). In some cases, the non-Sage will be able to suspend judgement on such impressions because their defect is obvious, but in others their unclarity and other flaws escape his notice, and, as a result, he gives assent to them precipitately (Plutarch, St. Rep. 1056E–F). Here he fails to monitor the conditions of his sense-organs, including their relation to the sense-object, and pays insufficient attention to the phenomenological quality and other representational features of the impressions so formed (Cicero, Acad. 2.51–3). Even so, there is still a reason why the non-Sage assents to such impressions, which Chrysippus identifies with habit, namely, the habit of unqualified acceptance of the deliverances of the senses. Not noticing their phenomenological tarnish and other deficiencies, the non-Sage relies on a long-standing and unquestioned deference to perception and accepts what these unreliable impressions are telling him. When this kind of habit is the assent-inducer, but the underlying perceptual impressions are not kataleptic, the resulting belief will be an opinion and so not one that the Sage would ever form.

According to the Stoics, the Sage never ‘mis-sees’ or ‘mis-hears’, i.e., never accepts a non-kataleptic perceptual impression (P.Herc. 1020, col. 105, 7–24; Arius in Stob., Ecl. 2.112.20–113.3). One reason for this is that she does not accept the perceptual impression that p simply out of a habitual and unexamined confidence that what the senses report is true. Instead, she assents to the perceptual impression that p only when she appreciates why in this circumstance, in relation to this object, at this time, and in this state of mind, the impression that p is kataleptic; and when she is unable to come to a determination on these matters, she suspends judgement on p (Cicero, Acad. 2.57; cf. Sextus, M 7.424). Here we may feel, with Cicero and other sceptical critics, that such systematic situational alertness is not humanly possible, and consequently that the only option for the Sage is universal suspension (Acad. 2.66–8, 90; Sextus, M 7.155–7). But it seems that Chrysippus’ Against Custom, in tandem with other epistemological works (cf. Plutarch, Comm. Not. 1059B), tried to bring this feat of alertness within reach, by enumerating the circumstances under which, and the reasons why, the deliverances of the senses must be true. Since her assent is guided by such considerations, rather than a blanket trust in the senses, the Sage upgrades the perceptual grasp that p to knowledge of p: such considerations would provide her with the rational resources necessary to defend the truth of p if challenged. ‘In this state of mind, in relation to this object, and under these perceptual conditions, what the senses report must be true,’ she could explain.

If this reconstruction is along the right lines, then we can appreciate why, as Plutarch and others report, Chrysippus’ ardour for searching out the limitations of custom and the senses could be appropriated by sceptics such as Carneades to undercut the possibility of apprehension. Chrysippus will have argued that in circumstances C, for reasons x, y, z, assent to the perceptual impression that p either should or should not be given, because p either is or is not guaranteed to be true in C. This is an attempt to specify the correct assent-inducers for one’s perceptual beliefs—the considerations on the basis of which one could attain perceptual knowledge. Carneades, however, could have repackaged these claims as part of a general attack on the Stoic idea of an impression that could not be false and hence on the Stoic idea of apprehension. If Chrysippus sees the need to explain under what circumstances the habitual confidence in the senses must be suspended, then he must assume that, in those circumstances at least, the impression itself does not indicate to the agent that it is non-kataleptic.59 But this result seems to open up the possibility of the general indistinguishability of kataleptic and non-kataleptic impressions, the perpetual battleground of Hellenistic epistemology. For the Stoics, the inability of the non-Sage to distinguish false from kataleptic impressions does not imply that there is no intrinsic difference between them (Cicero, Acad. 2.57–8). But the sceptic will reply that even the Sage could not attain the situational awareness necessary to rule out the possibility that, according to Chrysippus’ own guidelines, she is presently in circumstances when the senses are not reliable, since she could only reach such a determination by consulting her impressions.60

Furthermore, Chrysippus will have stressed the possibility of custom-based ignorance concerning non-perceptual matters. For instance, by adhering to what is widely accepted in his community, an ordinary Hellenistic Greek could come to acquire false beliefs about value, e.g. « Bodily pleasure is good » or « Death is bad ».61 Since these propositions are false, the Sage will never assent to them. In part, this is because she is able to demonstrate, on the basis of the principles or theorems (θεωρήματα) of Stoic axiology, that only virtue (and what participates in it) is good, and only vice (and what participates in it) is bad.62 For this reason, some Stoics go so far as to classify ethics itself as a virtue (Diogenes Laertius, 7.92), and so inter-entailed by those of dialectic: the Sage’s comprehensive understanding of value enables her to know when to give assent and when to withhold it when faced with propositions of ethical significance. Similarly, physics is also said to be a virtue (Diogenes Laertius, 7.92; Cicero, Fin. 3.72–3). The Sage affirms and knows that e.g. « The gods are providential » because she can see how this fact falls out of the real structure of the cosmos, and also why rival (e.g. Epicurean or traditional anthropomorphic) views of the gods are false. Indeed, the providentiality of the universe and related considerations will also help to establish why, in certain circumstances, what the senses report is guaranteed to be true, and so knowledge of physics will lend rational support to the Sage’s assents to perceptual impressions as well. So here we find clear confirmation of the systematic character of knowledge (section 2).

To attain knowledge, then, an assent-inducer other than custom must be present. Chrysippus sees custom as a ladder to be kicked off after serious study of Stoic philosophy has begun: it gets us to weak apprehension in some cases but to false and non-kataleptic beliefs in others, and therefore must be replaced by a more secure and stable basis for assent, if we want the ‘knowledge on the basis of which we will live consistently’ (T1). For this reason, Chrysippus takes up the task of criticising custom. He needs to set out why συνήθεια must be superseded as an assent-inducer by a more robust appreciation of why p is true in the circumstances. Of course, once this lesson is learned and the student has come to appreciate the epistemic limitations of custom, there is more work yet to be done. But without such an appreciation, knowledge will be forever out of reach.

6 Conclusion

From the fecund quotations preserved in Plutarch, we can recover a clearly articulated Chrysippean account of how the starting-points for knowledge are first acquired and then fortified. Insecure apprehensions are commonplace because kataleptic impressions are commonplace, and the special link such impressions have with the world is maintained regardless of the psychological route that leads the agent to accept them as true. Consequently, it is possible to obtain insecure apprehension on the basis of custom, and so the Academic challenge to this kind of epistemic success can be resisted. My reconstruction is consistent with existing interpretations of the kataleptic impression, but builds on previous studies by casting new light on Chrysippus’ understanding of the origins and nature of insecure apprehension. In particular, I have called attention to Chrysippus’ distinction between assent-inducers and apprehension-makers, and showed how it provides support for his analysis of knowledge as secure apprehension that is unchangeable by argument. One upgrades the weak, custom-based apprehension of p to knowledge of p by fortifying the rational grounds of one’s assent to p. This process will include consideration of anti-Stoic arguments and sophistic puzzles, because in coming to understand how these contrary views go wrong, one is no longer moved to give up what one has been led by custom to grasp. In this way, as Sextus playfully but accurately observes, Stoic pedagogy can be seen to ‘come to the aid of tottering common sense’ (PH 2.229).63


  • Alessandrelli, M. and Ranocchia, G., eds. (2017). Opera Incerta: PHerc. 1020, col. 104–112. Rome. [P.Herc. 1020]

  • Allen, J. (2019). Megara and Dialectic. In: Bénatouïl, T. and Ierodiakonou, K., eds., Dialectic after Plato and Aristotle, Cambridge, pp. 1746.

    • Search Google Scholar
    • Export Citation
  • Allen, J. (1994). Academic Probabilism and Stoic Epistemology. Classical Quarterly 44, pp. 85113.

  • Annas, J. (1990). Stoic Epistemology. In: Everson, S., ed., Epistemology, Cambridge, pp. 184203.

  • Arrighetti, G., ed. (1973). Ratae Sententiae. In: Epicuro: Opere. Turin. [Epicurus, Principal Doctrines].

  • Atherton, C. (1993). The Stoics on Ambiguity. Cambridge.

  • Ax, W., ed. (1938). M. Tullius Cicero, De Fato. Stuttgart.

  • Ax, W., ed. (1933). M. Tullius Cicero, De Natura Deorum. Berlin. [Cicero, Nat.].

  • Babut, D. (2004). Plutarque: Oeuvres morales, Tome XV (1re partie). Paris.

  • Baldassarri, M. (1976). Plutarco: Gli opuscoli contro gli Stoici, Volume I. Trent.

  • Barigazzi, A., ed. (1991). Galeni De optimo docendi genere. Berlin. [Galen, Best Way of Teaching]

  • Barnes, J. (1990). The Toils of Scepticism. Cambridge.

  • Bénatouïl, T. (2006). Faire usage: la pratique de Stoïcisme. Paris.

  • Bobzien, S. (2012). How to Give Someone Horns: Paradoxes of Presupposition in Antiquity. Logical Analysis and History of Philosophy 15, pp. 159184.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bobzien, S. (2005). The Stoics on the Fallacies of Equivocation. In: Inwood, B. and Frede D., eds., Language and Learning: Philosophy of Language in the Hellenistic Age, Cambridge, pp. 239273.

    • Search Google Scholar
    • Export Citation
  • Bobzien, S. (1998). Determinism and Freedom in Stoic Philosophy. Oxford.

  • de Boer, W., ed. (1937). Galeni De propriorum animi cuiuslibet affectuum et peccatorum dignotione et curatione. Leipzig. [Galen, Affections and Errors of the Soul].

    • Search Google Scholar
    • Export Citation
  • Boys-Stones, G. (1998). Plutarch on κοινὸς λόγος: Towards an Architecture of De Stoicorum Repugnantiis. Oxford Studies in Ancient Philosophy 16, pp. 299329.

    • Search Google Scholar
    • Export Citation
  • Brennan, T. (2005). The Stoic Life: Emotions, Duty, Fate. Oxford.

  • Brennan, T. (2003). Stoic Moral Psychology. In: Inwood, B., ed., The Cambridge Companion to the Stoics, Cambridge, pp. 257294.

  • Brittain, C. (2014). The Compulsions of Stoic Assent. In: Lee, M., ed., Strategies of Argument: Essays in Ancient Ethics, Epistemology, and Logic, Oxford, pp. 332354.

    • Search Google Scholar
    • Export Citation
  • Brittain, C. (2006). Cicero: On Academic Scepticism. Indianapolis.

  • Bruns, I., ed. (1887). De fato. In: Alexandri Aphrodisiensis praeter commentaria scripta minora ii. 2. Berlin.

  • Brunschwig, J. (1988). Sextus Empiricus on the Kritêrion: The Sceptic as Conceptual Legatee. In: Dillon, J. and Long, A.A., eds., The Question of ‘Eclecticism’, California, pp. 145175; reprinted in Brunschwig, J. (1994). Papers in Hellenistic Philosophy, Cambridge, pp. 224243 (page numbers to reprint).

    • Search Google Scholar
    • Export Citation
  • Butcher, S.H., ed. (1903–31). Demosthenis Orationes. Oxford.

  • Castagnoli, L. (2010). How Dialectical was Stoic Dialectic?. In: Nightingale, A. and Sedley, D., eds., Ancient Models of Mind, Cambridge, pp. 153179.

    • Search Google Scholar
    • Export Citation
  • Caston, V. (forthcoming). The Stoics on Mental Representation. In: Klein, J. and Powers, N., eds., The Oxford Handbook of Hellenistic Philosophy, Oxford.

    • Search Google Scholar
    • Export Citation
  • Cherniss, H. (1976). Plutarch: Moralia volume XIII part 2. Cambridge.

  • Cohn, L. and Wendland, P., eds. (1897). De agricultura. In: Philonis Alexandri opera quae supersunt ii. Berlin.

  • Coope, U. (2016). Rational Assent and Self-Reversion: A Neoplatonist Response to the Stoics. Oxford Studies in Ancient Philosophy 50, pp. 237288.

    • Search Google Scholar
    • Export Citation
  • Cooper, J. (2004). Arcesilaus: Socratic and Sceptic. In: Knowledge, Nature, and the Good, Princeton, pp. 81103.

  • Cooper, J. (1975). Reason and Human Good in Aristotle. Indianapolis.

  • Denyer, N. (2002). Neglected Evidence for Diodorus Cronus. Classical Quarterly 52 (2), pp. 597600.

  • Dorandi, T., ed. (2013). Diogenes Laertius, Lives of the Eminent Philosophers. Cambridge.

  • Döring, K. (1989). Gab es eine Dialektische Schule?. Phronesis 34, pp. 293310.

  • Duke, E., et al., eds. (1995). Sophista. In: Platonis Opera I. Oxford.

  • Ebert, T. (2008). In defence of the Dialectical school. In: Alesse, F., et al., eds., Anthropine Sophia, Naples, pp. 275293.

  • Frede, M. (1999). Stoic Epistemology. In: Algra, K., Barnes, J., Mansfeld, J. and Schofield, M., eds., The Cambridge History of Hellenistic Philosophy, Cambridge, pp. 295322.

    • Search Google Scholar
    • Export Citation
  • Frede, M. (1983). Stoics and Sceptics on Clear and Distinct Impressions. In Burnyeat, M., ed., The Skeptical Tradition, California, pp. 6593. Reprinted in Frede, M. (1987). Essays in Ancient Philosophy, Minnesota, pp. 151176 (page numbers to reprint).

    • Search Google Scholar
    • Export Citation
  • Gerson, L.P. (2009). Ancient Epistemology. Cambridge.

  • Gourinat, J.-B. (2019). Stoic Dialectic and Its Objects. In: Bénatouïl, T. and Ierodiakonou, K., eds., Dialectic after Plato and Aristotle, Cambridge, pp. 134167.

    • Search Google Scholar
    • Export Citation
  • Gourinat, J.-B. (2018). The Ontology and Syntax of Stoic Causes and Effects. Rhizomata 6 (1), pp. 87108.

  • Gourinat, J.-B. (2000). La dialectique des stoïciens. Paris.

  • Graver, M. (2007). Stoicism and Emotion. Chicago.

  • Hasan, A. and Fumerton, R. (2018). Foundationalist Theories of Epistemic Justification. In: Zalta, E., ed., The Stanford Encyclopedia of Philosophy. URL = <>.

    • Search Google Scholar
    • Export Citation
  • Hense, O., ed. (1905). C. Musonii Rufi reliquiae. Leipzig.

  • Inwood, B. and Gerson, L.P. (1997). Hellenistic Philosophy Introductory Readings, Second Edition. Indianapolis.

  • Kalbfleisch, C., ed. (1907). Simplicii in Aristotelis categorias commentarium. Berlin.

  • Kumaniecki, K.F., ed. (1969). M. Tullius Cicero, De oratore. Stuttgart.

  • Linder, C.G., ed. (1856). Hyperidis oratoris attici pro Euxenippo in Polyeuctum oratio. Upsala.

  • Long, A.A. (1978). Dialectic and the Stoic Sage. In: Rist, J., ed., The Stoics, California, pp. 101124. Reprinted in Long, A.A. (1996). Stoic Studies, California, pp. 85106 (page numbers to reprint).

    • Search Google Scholar
    • Export Citation
  • Long, A.A. and Sedley, D. (1987). The Hellenistic Philosophers. Cambridge. [LS].

  • Marrone, L., ed. (1997). Le Questioni Logici di Crisippo (PHerc 307). Cronache Ercolanesi 27, pp. 83100.

  • Marshall, P.K., ed. (1968). A. Gellii Noctes Atticae. Oxford.

  • Mau, J. and Mutschmann, H., eds. (1914–61). Adversus Mathematicos. In: Sexti Empirici Opera iiiii. Leipzig. [Sextus, M].

  • Mau, J. and Mutschmann, H., eds. (1914–61). Pyrrhoniae Hypotyposes. In: Sexti Empirici Opera i. Leipzig. [Sextus, PH].

  • Meinwald, C. (2005). Ignorance and Opinion in Stoic Epistemology. Phronesis 50, pp. 215231.

  • Menn, S. (1995). Physics as a Virtue. Proceedings of the Boston Area Colloquium in Ancient Philosophy 11, pp. 134.

  • Mignucci, M. (1999). Paradoxes. In: Algra, K., Barnes, J., Mansfeld, J. and Schofield, M., eds., The Cambridge History of Hellenistic Philosophy, Cambridge, pp. 157176.

    • Search Google Scholar
    • Export Citation
  • Moraux, P. (1968). La joute dialectique d’après le huitième livre des Topiques. In: Owen, G.E.L., ed., Aristotle on Dialectic: The Topics. Proceedings of the Third Symposium Aristotelicum, Oxford, pp. 277311.

    • Search Google Scholar
    • Export Citation
  • Moss, J. and Schwab, W. (2019). The Birth of Belief. Journal of the History of Philosophy 57 (1), pp. 132.

  • Nawar, T. (2014). The Stoic Account of Apprehension. Philosophers’ Imprint 14 (29), pp. 121.

  • Perin, C. (2005). Stoic Epistemology and the Limits of Externalism. Ancient Philosophy 25, pp. 383402.

    • Crossref