SciELO - Scientific Electronic Library Online

vol.42 issue3THOMAS KUHN’S THEORY OF RATIONALITYFrom modal fallacies to a new argument for fatalism author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links



Print version ISSN 0100-6045On-line version ISSN 2317-630X

Manuscrito vol.42 no.3 Campinas July/Sept. 2019  Epub Oct 21, 2019 



1Autonomous University of Barcelona Department of Philosophy Bellaterra, Cerdanyola de Vallès Barcelona


The five modes are a list of tools used by ancient sceptics to guide dogmatic people towards suspending their judgement. Attributed to Agrippa (of uncertain date) and used extensively by Sextus Empiricus (2nd or 3rd century CE), these modes are still widely discussed today by epistemologists and specialists in ancient philosophy. Scholars disagree, however, on how to understand the way the five modes are used together and what the logical form of the sceptical strategy behind their deployment is. This paper offers a reconstruction of the system of the five modes that avoids these problems. In specific, unlike previous reconstructions, (a) it includes a non-trivial version of the mode of relativity, (b) avoids committing the sceptic to a normative principle and (c) follows the textual evidence more closely. Moreover, I argue that the system can be better understood as a list of steps in a process, whose underlying logic can be expressed by a single algorithm.

Keywords: Skepticism; Pyrrhonian; scepticism; Agrippa's trilemma; Ancient; philosophy


The five modes -the mode of disagreement, regress, relativity, hypothesis, and reciprocity- are a list of tools used by ancient sceptics to guide dogmatic people towards suspending their judgement.1 Attributed to Agrippa (of uncertain date) and used extensively by Sextus Empiricus (2nd or 3rd century CE), these modes are still widely discussed today by epistemologists and specialists in ancient philosophy. Scholars disagree, however, on how to understand the way the five modes are used together and what the logical form of the sceptical strategy behind their deployment is. Reconstructions run the risk of oversimplifying, misconstruing or weakening the sceptical challenge posed by the five modes.

This paper offers a reconstruction of the system of the five modes for the suspension of judgement that aims to avoid these problems. In specific, unlike previous reconstructions, (a) it includes a non-trivial version of the mode of relativity, (b) avoids committing the sceptic to a normative principle and (c) follows the textual evidence more closely. The paper has three sections. Firstly, I show that previous reconstructions are unable to express important features of the workings of the system. Secondly, I revise the sources, show that it is problematic to mix and match the two surviving versions of the modes and assess how much evidence of systematic use of the modes we have. Finally, I propose a new reconstruction of the system. I argue that the system can be better understood as a list of steps in a process, whose underlying logic can be expressed by a single algorithm.


How should we understand the five modes for the suspension of judgement? Many people agree that they are meant to work as a dialectical system.2 However, the textual evidence of how these modes may fit together is, at best, scarce and insufficient. Barnes (1990a, 1990b), for example, offers a couple of basic diagrams that are meant to capture the systematic aspect of the modes, but he admits that they are inventions with no historical actuality (1990b, 215). Recent scholars are often too ready to attribute specific logical structures to the system, implied within the description and use of the five modes, but they rarely agree on how exactly that structure should look.

Moreover, one may wonder if there really is a specific, organized way in which the modes work together in any conversation. Perhaps the sceptic will use them at her discretion, deciding which one to use depending on the circumstances; or maybe she can use them randomly, without a specific order and not having to follow a specific set of rules. If one of the latter alternatives is compatible with the textual evidence, does it mean that the traditional depiction of the system speculates way beyond what is needed to make sense of the texts?

Even beyond a purely historical concern, there is another more troubling problem. Williams (1988, 556), for example, once thought that ‘Pyrrhonian scepticism cannot be systematised, formalized, or reduced to a rule;’ not, at least, without seriously misinterpreting the texts and compromising its coherence. Or take Barnes (1990b, 114) again, who claims that the system of five modes, if there is such a thing, ‘is a curious thing: it is positively rococo in its complexity, yet it possesses neither aesthetic elegance nor philosophical cohesion.’ So, even if the text clearly showed the modes working together in a specific way, it is unclear if it is possible to construe that into a formal schema and how exactly one ought to understand its logical form and whether its schematization would fit together with the sceptical project more generally.

These difficulties, of course, have not stopped some people from trying such reconstructions. There are two main families of proposals in the literature: those who divide the modes into a triggering strategy and a sceptical argument, and those who argue that the five modes are a single dialectical strategy or technique. However, as I shall show, it is unclear whether the models offered by both of these families of proposals really escape from speculating, misreading the text, or compromising the coherence of the sceptical project.

1.1 The five modes as a triggering strategy plus sceptical arguments

Take for instance the first family of proposals, popular with many contemporary epistemologists. The idea is to reduce the five modes into a triggering strategy that is sometimes obviated, and an argument, normally a trilemma (the famous ‘Agrippa’s trilemma’).3 The appeal in these attempts is clear. They try to offer a neat, manageable and elegant interpretation of the modes. There are many variations of the trilemma, but it can be illustrated as follows (this is my own version). Imagine you have been challenged to justify one of your claims, and after you offer an answer your interlocutor presents you with the following argument:

Agrippa’s trilemma (AT)

  1. The justification for your claim is either an unjustified hypothesis, or requires in turn a chain of justification that falls into an infinite regress, or is circular.

  2. If it is an unjustified hypothesis, you should suspend judgement.

  3. If your justification falls into an infinite regress, you should suspend judgement.

  4. If it is circular, you should suspend judgement.

  5. Therefore, you should suspend judgement (from 1-4).

This is as good as one could hope for a philosophical argument. It is valid, its premises are plausible, and its conclusion shocking. However, this type of formalization of the modes has received heavy criticism, some of which is undeserved. A common complaint is that it makes the sceptic endorse the premises of her argument. But that does not necessarily follow. Anyone can offer an argument without committing to its validity or the truth of its premises, in a dialectical interchange what really matters is if one’s interlocutor accepts the argument’s premises. A similar but distinct complaint is that the sceptic would not be able to apply this argument in herself, because in order for it to work one needs to accept dogmatically its premises and rule of inference. However, this might not be a problem either. An urbane reading of Pyrrhonian scepticism would not be affected (since it means that the sceptic does not suspend judgement about everything), and a rustic interpretation could understand the modes as just one step into a longer path to a global scepticism.4

However, one may worry that this formalization leads to another problem. The argument seems to allow easy responses, since people could just reject one of its premises. In fact, this is the reason some epistemologists like this type of argument so much. It provides the background for introducing foundationalism, infinitism, and coherentism.5 However, things are not as easy as they seem. Whatever one answers to the above argument, the sceptic can reapply it directed to one’s new claim since the five modes are topic-neutral. Thus, even in this reconstruction, the five modes do not necessarily compromise the coherence of scepticism and pose an interesting meta-epistemological challenge.6

However, this reconstruction does a poor job of expressing the dialectical context in which the argument is supposed to be offered. Leaving the dialectical aspect implied is the main reason people accuse the sceptics of dogmatism. To avoid this problem, Lammenranta (2008, 8) supplies Agrippa’s trilemma with a formalization of the disagreement mode, which he thinks is the central feature of the system and the mode that clarifies its dialectical nature. For his reconstruction, he asks us to suppose a question to which there are only two possible answers (p and not-p), he then offers an argument with the following form:7

Disagreement argument (DA)

  1. S 1 claims that p.

  2. S 2 claims that not-p.

  3. At most, one of them is right.

  4. The disagreement between S 1 and S 2 is irresolvable.

  5. We should suspend judgement about p.

If one denies premise (4), Lammenranta argues, the sceptic would ask how the disagreement is to be resolved. If one gives a reason r for p, the sceptic points out that there is also an irresolvable disagreement about that too. If we keep rejecting premise (4), the sceptic could repeat the disagreement mode over and over or introduce Agrippa’s trilemma. If we put both of these arguments together, we get a challenge with four out of the five modes.

Now, Lammenranta explains that the dialectical strategy of his disagreement argument is based on the normative principle that under irresolvable disagreement we should suspend judgement. But notice that this also applies to Agrippa’s trilemma. Thus, a full reconstruction in these terms would include one normative principle (NP) and two arguments (DA and AT). The principle can be expressed as follows:

NP: If a judgement about whether p or not-p is under irresolvable disagreement, or is grounded on an unjustified hypothesis, infinite regress, or circular reasoning, we should suspend judgement.

However, in this reconstruction, the sceptic is vulnerable to a simple objection. As Lammenranta notices, it is possible to avoid scepticism by rejecting the normative principle.8 This move comes with the price of breaking the link between claiming something and the requirement to offer some justification for it. However, some people would bite the bullet and acknowledge that the scope of the modes is restricted to people who are in the business of justifying their claims. Yet, the reconstruction might not really require a normative principle. I shall come back to this point later, but the attempt to reduce the five modes to a couple of arguments faces another problem. Reconstructions like the above are inventions based rather loosely on the original sources. Thus, one may be unsatisfied with them as historical reconstructions. Therefore, if this proposal fails on both fronts, as a historically accurate interpretation and as a strong and challenging formalization of the modes, it loses some of its original appeal.

1.2 The five modes as a single dialectical strategy

The second proposal has given rise to many variants but they all understand the modes as challenges in a dialectical exchange coordinated in a single strategy.9 This makes clearer that the sceptics do not offer proof, nor are they committed to any premise or rule of inference. Even if this proposal may be closer to the original texts, people often speculate over the exact way the dialectical strategy works. As with the previous proposal, they also rely on a normative principle and some models incorporate only four of the five modes.

Barnes’ (1990b, 114-15) system, for example, offers a reconstruction with four of the five modes, leaving aside relativity. He starts with disagreement about a problem, and then shows that to resolve it we can just affirm one of the solutions, or offer some reasons to support it. If we chose the first one, the sceptic would reply with the hypothetical mode. Thus, it seems that we must choose the second option and offer some reasons to support our solution to the problem. But the sceptic will adduce disagreement about these reasons and we cannot just affirm them because of the hypothetical mode. Therefore, we produce some reasons in favour of our initial reasons. However this new set of reasons is either identical to the previous one, in which case the sceptic will use the reciprocity mode, or is new, in which case it will be subject to disagreement again. In this way, by repeated application of disagreement, hypothesis and reciprocity, we are led into an infinite regression, which is forbidden by the mode of regress.10

Hankinson (1995, 171) thinks that the texts ‘clearly indicate that the [five] Modes form a coherent collective sceptical strategy’, and offers a similar version of the sceptical system. According to him, if we suppose a claim C1, there will be disagreement as to whether it should be accepted on its own terms or whether it needs further confirmation. Such a disagreement can either be decided or not. If not, we should suspend judgement. If the disagreement is decidable, it is decidable on the basis of another claim C2, or it is just assumed to be true. If the latter, then the mode of hypothesis applies. If the former is the case, C2 is either supported by another claim C3 or by C1. If it is by C1, then the reciprocal mode applies. If it is supported by C3, the previous steps will repeat, leading either to suspension of judgement or to infinite regress.11

Fogelin (1994b, 116), put forward another influential variation. He divides the five modes into two subsets, what he calls the challenging modes (disagreement and relativity), in charge of triggering a demand for justification, and the dialectical modes (hypothesis, regress, and circularity), which shows that justification is impossible. In his reconstruction, the target of the five modes is someone committed to the possibility of knowledge and a strong normative principle according to which it is epistemologically irresponsible to choose without an argument between competing claims.12

I suggested a different model in Vázquez (2009, 46-49, 52). There, I argue that all the modes are questions or challenges that can start the dialectic, and each mode can lead to suspension or, if rejected, to any other mode. In this proposal, the suspension of judgement is not only the final outcome of the five modes but also the ‘the dynamic process of going from one mode to another in a permanent movement of thoughts’ (52). The suspension, then, does not have to be a passive final state, but an active and continuous dialectic.13

How exactly shall we represent the logic behind these reconstructions? They have sometimes been visualised in tree-like schemas but their format varies and their structure is not always transparent. However, these models can be understood as a sequence of steps used by the sceptic. This can be represented in flowcharts, which show the differences between the models offered above more clearly. Take for instance Barnes’ and Hankinson’s models (Figure 1 and 2 below).

Figure 1: Barnes’ model 

Figure 2: Hankinson’s model 

The flowcharts show the different structures of these models and their differences in terms of specificity and detail. Note, for example, that Barnes’ model is more explicitly dialectical. Hankinson’s model, however, can be applied not only in a conversation with the dogmatist but also in the sceptic’s internal dialogue. In both models, there is a point where the steps have to be repeated. Their proposed structure, however, expresses this important aspect rather poorly. Hankinson’s model would grow indefinitely and trivially, whereas Barnes’ heavily relies on the reapplication of disagreement (and if that is all one needs, the other modes seem pointless).

None of these models, however, offer all of the advantages while avoiding all of the problems. The details of how each mode connects with the others and which one should be reapplied are highly speculative (their authors offer no reason to prefer one model over the other nor do they offer any specific textual evidence for the specifics of their reconstructions). Moreover, most of these models are incomplete or curtail the scope of the system’s application by assuming that the sceptic is committed to a normative principle. But again, there is no textual basis for that nor does it seem a logical requirement. Sextus never presents suspension of judgement as something the sceptic should do, but as something that happens to her, that is brought about by the modes where the sceptic has no say (see PH 1, 35, 177).14 In the next section, I turn to the original textual evidence to reassess the situation and to show that there is still room for improvement in our characterisation of the system of the five modes.


There are only two surviving sources for the five modes for the suspension of judgement. One is Sextus Empiricus’ PH 1, 164-169 and the other is a parallel passage in Diogenes Laertius’ Lives 9, 88-8915 In addition, Sextus deploys the modes in numerous places, most notoriously in PH 1, 170-177, 178-179, 185-186 and 2, 20. Although Diogenes’ passage is briefer, it offers variations and information missing in Sextus, like attributing the modes to Agrippa.16 A plausible explanation for the similarities in Sextus and Diogenes is that both were copying from a common source now lost. However, the authorship of this common source is very difficult to establish.17 The parallel texts read as follows:

Sextus Empiricus Diogenes Laertius
[164] The more recent Sceptics offer the following five modes of suspension of judgement: first, the mode deriving from dispute; second, the mode throwing one back ad infinitum; third, the mode deriving from relativity; fourth, the hypothetical mode; fifth, the reciprocal mode. [88] […] But those in Agrippa’s circle posit another Five Modes: one that argues from Disagreement, one from Infinite Regress, one from Relativity, one from Hypothesis, and one from Reciprocity.
[165] According to the mode deriving from disagreement, we discover that undecided dissension about the matter proposed has come about both in ordinary life and among philosophers. Because of this we are not able either to choose or to reject anything, and we end up with suspension of judgement. The mode from Disagreement exposes any question put forward among philosophers or in ordinary life as full of utmost conflict and complete confusion.
[166] In the mode deriving from infinite regress, we say that what is brought forward as a source of conviction for the matter proposed itself needs another such source, which itself needs another, and so ad infinitum, so that we have no point from which to begin to establish anything, and suspension of judgement follows. The mode from Infinite Regress does not permit that which is under investigation to be established, because one thing receives its credibility from another and so on ad infinitum.
[167] In the mode deriving from relativity, as we said above, the existing object appears to be such-and-such relative to the subject judging and to the things observed together with it, but we suspend judgement on what it is like in its nature. [89] The mode from Relativity says that nothing is apprehended by itself, but always with something else. Therefore, it declares these matters unknown.
[168] We have the mode from hypothesis when the Dogmatists, being thrown back ad infinitum, begin from something which they do not establish but claim to assume simply and without proof in virtue of a concession. The mode from Hypothesis is introduced in response to those who believe that one should take as trustworthy the starting-points of things straightaway, rather than put them into question. But this is pointless: someone else will set up the opposite hypothesis.
[169] The reciprocal mode occurs when what ought to be confirmatory of the object under investigation needs to be made convincing by the object under investigation; then, being unable to take either in order to establish the other, we suspend judgement about both (PH 1, 164-169).18 The mode from Reciprocity comes up when that which should confirm the matter under investigation is itself in need of being confirmed by the very thing that is investigated, as when someone, seeking to confirm that there are pores by appealing to the occurrence of emanations, might adduce pores as confirmatory of there being emanations (DL 9, 88-89).19

There are some important differences between Sextus’ and Diogenes’ version of the five modes.20 Diogenes makes no reference to the suspension of judgement, gives no indication whatsoever that the modes are used together in a systematic way, and seems to conceive the modes as arguments that prove, establish conclusions, and reject other people’s arguments. Strictly speaking, then, Diogenes’ version of the five modes does not argue for scepticism but for negative metadogmatism.21 If there is a systematic and more dialectical use of the modes, whatever that is, it is only present in Sextus’ report, and it could have been a later development not originally present in Agrippa’s version. In Sextus’ report, the suspension of judgement plays a central role and is mentioned five times. The text also hints at some connections between the modes and is careful not to say that they prove or imply accepting any claim or rule of inference. There is no normative language. The modes are treated as strategies that lead to the suspension of judgement and fit with Sextus’ general sceptical agenda.

Finally, it seems that Diogenes used Peripatetic terminology whereas Sextus used a Stoic one.22 These terminological differences suggest that even if Diogenes and Sextus used a common source, their reports belong to two very different interpretative traditions. This makes very risky to mix and match modes from both lists.

The report of the mode of disagreement is also strikingly different. Diogenes’ version of disagreement only gives the conclusion of what the mode is supposed to prove: that any inquiry sets out an extremely negative outcome at a psychologically and interpersonal level. It is difficult to infer the rest of the argument, but the result of this mode, if successful, is a claim about the nature of inquiry: that regardless of its content, it leads to conflict and distress. The mode promises an astonishing result since it turns the inquiry into a forbidden zone. The text, however, fails to tell us how that outcome is achieved. The mere fact of disagreement, even if pervasive, is not enough to produce the level of contention and disturbance promised by the version of the mode reported by Diogenes.

In Sextus, however, the mode from disagreement has a different form. It is, strictly speaking, the mode resulting from undecided disagreement.23 A plain disagreement will not necessarily produce distress, a heated dispute, or suspension of judgement. In fact, Sextus’ version does not even assume that undecided disagreement produces the negative effects advertised in Diogenes’ version. In Sextus, this mode only leads to suspension if the person discovers (εὑρίσκομεν) that, under the current circumstances, there is an undecided dispute.

Furthermore, disagreement in Sextus seems to presuppose a previous interchange or awareness of arguments or evidence for and against the topic of discussion. The recognition of a specific undecided disagreement seems to depend on whether there are (or seem to be) opposing irreconcilable views about a given topic already mentioned. One may object, following Barnes (1990b, 21), that undecided disagreement does not produce suspension of belief. Barnes suggests that we are supposed to read Sextus’ ‘we are not able to choose or to rule out anything’ as ‘we should not choose or rule out anything.’ If this were correct, the sceptic would be clearly committed to a normative principle. But that is not in the text.

The result can be read, instead, as a report of an appearance. Given the sceptic’s appearance that the dispute is undecided at a specific moment and place, she reports that she is unable to choose. When the sceptic says that she is not able to choose, she is being literal about it. She is reporting his inability to take a decision, and for that she needs no normative principle. If she were able to choose, then the dispute would obviously be decidable. Suspension of judgement may not necessarily follow in every case and for everybody, though. The sceptic is only reporting that it does sometimes happen, to some people. The reasons are not said. The sceptic makes no claims about them.

The passages also report two different versions of the mode from relativity. In Diogenes, relativity looks like an abbreviated version of the sixth mode of Aenesidemus (PH 1, 24; DL 9, 84-85)24 or a variation of one of the two modes, a list reported by Sextus but absent from Diogenes (PH 1, 178). In contrast, Sextus’ version of the relativity mode refers back to Sextus’ version of the eighth mode of Aenesidemus (PH 1, 135-136), which is different from the eighth mode reported by Diogenes (DL 9, 86) and seems to be a later invention.25 The modifications found in Sextus aim to improve the cohesion of the sceptic’s modes.

However, in both of its versions, this mode might be deemed problematic. The main worry is that it is dogmatic, superfluous, or fails to integrate with the other four.26 Brennan & Lee (2014) use Diogenes Laertius’ version of the relativity mode in an attempt to incorporate it into the system. However, if Brennan and Lee (2014, 251) accuse Sextus’ version of relativity of not adding anything to the system of the five modes not already secured by disagreement, Diogenes’ version can be accused of not adding anything not already secured by the hypothetical mode, even if it targets a distinct relation, as Brennan and Lee suggest.

Moreover, as I have shown, it is unwise to mix the information from our two surviving sources because they seem to belong to different stages of the development of the five modes and to two different interpretative traditions. Otherwise, in the hope of finding Agrippa’s original system of the five modes, we run the risk of unintentionally ascribing to him a mere invention that has insufficient historical grounds. Therefore, it is best to look at Sextus’ report alone, which is the only source that offers some hints of a systematic use of the modes.

Looking at the evidence it is also necessary to specify what it means to say that the five modes conform to a system. At PH 2, 48, Sextus refers back to his use of the five modes at PH 2, 19-21 as ἐμμέθοδος, which Annas & Barnes translate as a ‘methodical procedure’ and means ‘according to a rule or system.’27 However, Sextus does not explain what exactly he means by that, it could be just a loose reference to the use of standardized modes. Thus, even if there is no doubt that Sextus uses the five modes together, and considers this a methodical procedure, the question is whether there is a more specific strategy or rule behind their use, and if so, what is the best way to understand its formal structure.

There are, at least, three different scenarios. In the most deflationary and economical, we could think that Sextus uses the modes one after the other, but without any specific order or commitment to use them all or following any rule more specific than that. After all, there are multiple passages that might suggest this is what is going on. This minimalist version could also sound attractive because if there is no strategy, the sceptic might think that she can escape from the accusation of dogmatism more easily (although, one could complain that using the modes randomly is also following a rule).

A second alternative is to think that Sextus has various general strategies to apply the modes: sometimes in relation to the two modes, others to the ten modes, and sometimes using only the five or a subset of the five modes, but without following any rule more precise than that. A sceptic may, depending on what appears to her at any given moment, combine the different lists of modes tailoring the best response in a given dialectical context. Besides, Sextus does say that the five modes add variety and possibilities to the ten modes and that the sceptic can use them together:

Such are the Five Modes which have been handed down by the more recent Sceptics. They put them forward not as rejecting the Ten Modes but in order to cross-examine the rashness of the Dogmatists in a more varied way by using both sets together (PH 1, 177).

A third option would be to find a specific strategy that explains the way the modes are deployed in a clear rule or set of rules. Perhaps there is a single rule that can explain how to use the five modes, how to combine them with other modes, and why sometimes only some of them are required. This view of the system of the five modes seems to be assumed by some of the available models described before. And even if not all of them make a historical claim, they often assume that by systematising the modes in specific ways they are offering the most plausible or elegant arrangement, doing the sceptic a favour.

Now, how far does the textual evidence can really take us? There are only a few hints in the text that one needs to be careful not to overemphasize. For example, it has been noted that the order in which the modes are listed may signal the standard order of their use (namely, disagreement, regress, relativity, hypothesis, and reciprocity; see PH 1, 164, 170-177 and DL 88.315-318).28 However, in many passages, Sextus only uses some of the modes, sometimes in a different order or together with other lists. Moreover, he offers no reason to prefer a certain arrangement to another. Thus, one may conclude that the fact that Diogenes’ list has the same order as Sextus’, only amounts to the claim that there is a standard order for listing the modes, but not necessarily a standard way to use them. Another hint that the modes are meant to work together in a specific way is Sextus’ account of the mode of hypothesis. There, he explicitly says that hypothesis is introduced after the regress mode. However, if we look at other places, Sextus does not always respect this order (see, for instance, PH 1, 185-186; 2, 18-20).

One thing Sextus does say is that every object of investigation can be referred back to the five modes (PH 1, 169). Perhaps a sceptic does not always need all of them, or she can mix and match them with other lists, but the idea is that the five are sufficient, even if not all of them are necessary all of the time. But yet again, this claim can be read in a deflationary way, without implying a specific rule of application. It does not even necessarily mean that the modes have to be used together, but that all of them are general enough as to apply to any possible subject matter.

The final piece of information is the passages where Sextus demonstrates how he uses the five modes together, especially where he seems to indicate that there is some logic behind determining which mode would be introduced at a given moment of the dialectic. But the same passages also suggest that the sceptic has a wide range of actions for improvisation, adaptation and the use of distinctions or concepts that fit the dialectical context or are used by the interlocutor (some examples are PH 1, 170-177, 178-179, and 185-186).

Then, one may wonder again, if there really is a way to make sense of the textual evidence beyond the deflationary reading. In other words, the exegetical challenge is to determine whether there is, in fact, a way to unify in a specific rule or system all the information we have without compromising the coherence of the sceptical project. I think in a way, there is. But is not to be taken as a general specification of how sceptics ought to use the five modes. Instead, I think it is about how Sextus applies the five modes as a sceptical practitioner. Of course, from his use the modes in certain patterns, it does not follow any normative suggestion. But it is still interesting to see that he tends to follow certain more specific patterns without compromising his coherence. If we look at the order in which the modes are used in the different passages, we can advance in this direction. Sextus uses the five modes in the following order:

  • 1. Disagreement, regress, relativity, hypothesis and reciprocity (PH 1, 164, 170-177; see also DL 88.315-318).

  • 2. Disagreement, regress, reciprocity, relativity and hypothesis (PH 1, 185-186).

When Sextus uses four of the five modes he does it in the following order:

  • 3. Disagreement, reciprocity, hypothesis, regress, and then he goes back to reciprocity (PH 2, 18-20).29

Finally, we also know that after regress, the sceptic can use the hypothetical mode:

  • 4. […] regress, hypothesis […] (PH 1, 168).

From this, we can observe certain features of Sextus’ use of the modes. First, that disagreement is always first. Second, that relativity is only followed by hypothesis and is never at the beginning or after disagreement, and thus, does not trigger the discussion. Third, notice that the other modes connect with two others, except for regress that connects with three. Disagreement is followed by either regress or reciprocity. Regress is followed by relativity, reciprocity, and hypothesis. In turn, hypothesis is followed by reciprocity or regress. Finally, reciprocity is followed by relativity or hypothesis. Moreover, Sextus tells us that all the modes could also lead to the suspension of judgement (PH 1, 164-169 and 186 for the hypothetical mode). Now, it is possible to find some specific rules of application derived from Sextus use of the modes (which, I want to reiterate, do not have normative value but illustrate Sextus’ personal practice).


Despite the variations in Sextus’ use of the modes, the information gathered in the previous section can be put in a single list of steps. A first version would look like this:

  1. Is there a claim? (If there is, go to step 2; otherwise go to step 8).

  2. Is there an undecided disagreement over the claim? (If so, go to step 7; otherwise go to step 3 or 6).30

  3. Is there an infinite regress? (If so, go to step 7; otherwise go to step 4, 5, or 6).

  4. Is it relative? (If so, go to step 7; otherwise go to step 5).

  5. Has an unjustified hypothesis been proposed? (If so, go to step 7; otherwise go to step 3 or 6).

  6. Has a reciprocal justification been offered? (If so, go to step 7; otherwise go to step 4 or 5).

  7. Has suspension of judgement been discovered? (If so, go to step 8; otherwise go back to step 1).

  8. End.

This sceptical recipe implies that the dialectic can take various routes. But it either leads to suspension of judgement or loops the conversation going from one mode to the next one following the above rules, conveying the recursive aspect of the system too. The advantage of thinking about the application of the modes in this way is that one does not need to hold beliefs to follow these rules. Even a machine like an elevator or an automatic door can be programmed to do certain actions when they receive the specified input.31 According to Sextus, the sceptic can act following his involuntary appearances with the need for no beliefs (PH 1, 19-20, 22-23). So the sceptic can follow recipes like this by following what appears to her at the time she asks each question. Now, the sceptic, unlike an elevator, may require a great amount of preparation, memorization, experience, and dialectical talent to follow the above rules. But even if she is more like a grandmaster chess player, notice that a machine can also achieve this level of expertise by following programs and without the need for beliefs.

In addition, notice that a list of rules does not imply a normative principle of rationality. The list does not end up recommending or demanding one’s suspension of judgement as the expected outcome of a rational interchange. In my reconstruction, step 7 only asks for confirmation of a psychological fact, the discovery of suspension of judgement, of a ‘standstill of the intellect’ (PH 1, 10). But notice that there are neither demands nor any assumption that one has to suspend judgement (suspension just ‘comes about’ PH 1, 31). Step 7 only asks us to report what appears at a given moment and time (PH 1, 19-20). There are no guarantees, nor does suspension follow logically, but if it does not happen, Sextus will just go back to step 1 and continue the application of the modes. The dialectical strength of the system is, instead, that once the sceptic initiates the application of the five modes there is no easy way out from the loop in which one is immersed. If one tries, for example, to ban the sceptic from using the system as a valid dialectical strategy at a certain level of conversation, the sceptic can use the five modes to challenge those rules of interaction, and so on.

Moreover, the language of discovery is to be understood as the finding of appearances, passive and unwilled, that the sceptic or her interlocutor will report (PH 1, 22-23). We should see these sceptical instructions as a therapeutic method against the mental disease of dogmatism (PH 3, 280). This therapy can be tailored to each individual case and is a medicine that might cure the patient but where there are no guarantees that she will be completely cured. However, there are testimonies that it sometimes works (PH 1, 26), and the sceptic could say that simply by engaging with the method one may neutralise some of the symptoms of dogmatism, like rashness. As long as one discusses with the sceptic, the discussion over a claim stays open.

Finally, notice that the list can be followed regardless of whether one is a perfect sceptic (with no claims to assess), an apprentice, or a full-on dogmatist. It can be applied as a self-imposed therapy or as a dialectical interchange with someone else.

Now, the list of steps I offered to explain Sextus sceptical practice still leaves some questions open: Is there a way to specify the alternatives the sceptic should take at each turn? Why would some modes only lead to some others and not to all of them?

Consider the mode which comes from relativity. The fact that rejecting it always sends us into the hypothetical mode might mean that it deals with attempts at grounding claims on appearances. One may think that the disagreement over a subject is decided by appealing to one’s own incontrovertible appearances. An example might help to explain what I mean. Someone may agree that there is disagreement over God’s existence but reply that it is apparent to her that God exists, so the disagreement is decidable in favour of God’s existence. A first-person testimony of an appearance and the assumption that this trumps other people’s opinions and arguments are used as grounds for the claim that God exists. A person who grounds her judgements on what appears to her is immune to disagreement precisely because she gives more credit to her personal appearances than to the discussion about the topic. Now, Sextus will not fight over appearances but over assuming the primacy of the first-person’s report by pointing out that one’s appearances are relative. The mode of relativity, then, challenges the unclear jump from appearance to judgement. Notice that this interpretation of relativity is not superfluous nor can it be subsumed by the mode of disagreement. On the contrary, it can tackle responses that the sceptic cannot redirect to disagreement.

Notice, however, that appealing to one’s appearances is not a straightforward unjustified hypothesis. Appearances as such are clear, so they are outside the scope of Pyrrhonian scepticism (see PH 1, 19-20; 2, 8). When someone claims something based on appearances and disregarding disagreement she might be just a step away from suspending judgement because that person could just not realise that others might have different appearances (and that this fact does not imply they have to change their actions because they can just follow their appearances).32

However, can Sextus use another mode after relativity, apart from the hypothetical mode? It does not seem so. The justification does not go to an infinite regress because the person’s appearance is supposed to be the end of the justification chain. If someone asks our theist why it appears to her that God exists she might not know, but that would be beside the point since what makes the appearance a good candidate for grounding one’s claims is its clarity and immediacy, not that it is grounded on something else. Since the report of an appearance is not another unclear theoretical proposition, but the report of a person’s phenomenological and psychological state, it is not a case of reciprocity either (remember the theists who say ‘God exists because it appears so to me, I sense its presence’). All this explains why the mode of relativity plays a crucial role in the system and cannot simply be regarded as trivial.

The decisions required after introducing the mode of disagreement seem also difficult to explain. If one acknowledges that disagreement over a topic is decidable, there are two options. Sextus could either introduce the mode of reciprocity or the regress mode. But we also know that Sextus can also introduce disagreement again. Why he does not? The reason might be that Sextus’ interlocutor might also disregard disagreement in virtue of a justification chain. Surely there is disagreement over climate change or whether we should vaccinate children but there are some good scientific justifications to believe in both of these claims. Here is when the regress and reciprocity mode would be introduced instead of applying disagreement again. When the sceptic reintroduces disagreement after disagreement, she is not using the system, but only one of its modes independently. Now, the systematic use of the modes is only one way of using them and depending on the details of the dialectical interchange (or personal reflections), Sextus might decide to interrupt the systematic use to move to a different list or to use one of the modes separately. None of this would be a problem since Sextus can come back to the systematic use at any given time.

Now, according to our information, rejecting that the justification goes on ad infinitum is followed by the modes of reciprocity, relativity or hypothesis. The decision as to which to select should be clearer by now. Relativity is introduced if the end of the justification chain is an appearance. If what stops the justification is another claim, there are two options. Either what stops the justification is an old claim, as Barnes calls them, or a new claim. In the first case, reciprocity is introduced, in the latter, hypothesis.

To dispute the accusation that one’s justification is reciprocal, one may reply that we are relying on an appearance, to which Sextus will respond by introducing the relativity mode. But if one denies that the justification is reciprocal, but it is a claim, then Sextus will introduce hypothesis. Finally, if one denies that the last piece of the justification is an unjustified hypothesis, Sextus will lead us back to regress or reciprocity.

I can now offer a systematic list of instructions for using the five modes based on Sextus practices. Unlike the previous list, this version includes information about how to choose the next mode at every stage (see Figure 3 for a graphical representation). The list runs as follows:

  1. Is there a claim? If there is go to step 2, if not go to step 12.

  2. Is there an undecided disagreement over the claim? If it has, go to step 11, or else go to step 3.

  3. Does the claim require a non-reciprocal chain of claims? If it does, go to step 4, if not go to step 9.

  4. Is there an infinite regress? If there is, go to step 11, or else go to step 5.

  5. Is the claim grounded by an appearance? If it is, go to step 6, or else go to step 7.

  6. Is it relative? If it is, go to step 11, if not go to step 8.

  7. Is the claim grounded by a new claim? If it is, go to step 8, if not go to step 9.

  8. Has an unjustified hypothesis been proposed? If it has, go to step 11, or else go to step 3.

  9. Has a reciprocal justification been offered? If it has, go to step 11, or else go to step 10.

  10. Has the claim been accepted by virtue of a concession? If it has, go to step 8, if not go to step 6.

  11. Has suspension of judgement been discovered? If it has, go to step 12, if not go to step 1.

  12. End.

Figure 3: new model for the use of the five modes 

Like the previous reconstruction, this list offers no guarantee that Sextus’ interlocutor will arrive at the suspension of judgement. But it shows how the conversation will remain in a loop unless someone quits the discussion or discovers suspension. This is not a weakness of the system. Sextus never promises a bulletproof therapy that will necessarily lead to suspension. One may wonder, however, whether staying in the loop count as suspending judgement? Perhaps in a sense, it does. Whoever is trapped in the loop cannot justify her claims. But sceptical suspension, the one that leads to happiness, includes the discovery, the realization of this very fact. It is a discovery of one’s own cognitive state at the time it is unable to justify a given claim and someone could remain in the sceptic’s maze without realising this fact.

To conclude, let me assess my new model with respect to my original objectives. My model incorporates the five modes, including a non-trivial version of relativity that deals with attempts at grounding claims through appearances. I do this without falling into the problematic practice of mixing Diogenes and Sextus’ report of the modes. More importantly, I offer a reconstruction of the modes that clarifies the underlying logic of the use of the modes without committing Sextus to a normative principle, a rule of inference or a claim.33 Finally, my reconstruction of the inner workings of the system is more firmly grounded on the textual evidence of Sextus’ description and application of the five modes.


ANNAS, J. and JONATHAN BARNES. The Modes of Scepticism. Ancient Texts and Modern Interpretations. Cambridge: Cambridge University Press, 1985. [ Links ]

BAILEY, A. “Pyrrhonean Scepticism and the Self-Refutation Argument.” The Philosophical Quarterly 40 (158): 27, 1990. [ Links ]

BARNES, J. Aristotle Posterior Analytics. 2nd 1993. New York: Oxford University Press, 1975. [ Links ]

_____. “The Beliefs of a Pyrrhonist.” Proceedings of the Cambridge Philological Society, no. 28: 1-29, 1982. [ Links ]

_____. “An Aristotelian Way with Scepticism.” In Aristotle Today, edited by M. Matthen. Edmonton: Academic Printing, 1987. [ Links ]

_____. “Some Ways of Scepticism.” In Epistemology, edited by Stephen Everson, 1990a. [ Links ]

_____. The Toils of Scepticism. New York: Cambridge University Press, 1990b. [ Links ]

_____. “Diogenes Laertius IX 61-116: The Philosophy of Pyrrhonism.” In Aufstieg Und Niedergang Der Römischen Welt, edited by W. Hasse and H. Temporini, II:4241-4301. Berlin: de Gruyter, 1992. [ Links ]

BETT, R. “Pyrrhonian Skepticism.” In The Routledge Companion to Epistemology, edited by Sven Bernecker and Duncan Pritchard, 403-13. New York: Routledge, 2010. [ Links ]

BRENNAN, T., and JAMES LEE, J. “A Relative Improvement.” Phroensis 59: 246-71, 2014. [ Links ]

BULLOCK, J. B. “The Challenges of the Modes of Agrippa.” Apeiron 49 (4): 409-35, 2016. [ Links ]

CORTI, L. Scepticisme et Langage. Paris: Vrin, 2009. [ Links ]

EMPIRICUS, SEXTUS. Against the Logicians. Edited by Richard Bett. Cambridge: Cambridge University Press , 2005. [ Links ]

FOGELIN, R. Pyrrhonian Reflections on Knowledge and Justification. New York: Oxford University Press , 1994a. [ Links ]

_____. Pyrronian Reflections on Knowledge and Justification. Oxford: Oxford University Press, 1994b. [ Links ]

GERKEN, M. “Discursive Justification and Skepticism.” Synthese 189 (2): 337-94, 2012. [ Links ]

GRECO, J. “Virtue, Luck and the Pyrrhonian Problematic.” Philosophical Studies 130 (1): 9-34, 2006. [ Links ]

HANKINSON, R.J. The Sceptics. New York: Routledge , 1995. [ Links ]

JANÁČEK, K. “Skeptische Zweitropenlehre Und Sextus Empiricus. ” Eirene 8: 47 - 55, 1970. [ Links ]

_____. Sextus’ Empiricus Methods. Prage: Universita Karlova Praha, 1972. [ Links ]

KLEIN, P. “How a Pyrrhonian Skeptic Might Respond to Academic Skepticism.” In The Skeptics: Contemporary Essays, edited by Steven Luper, 75-94. Aldershot: Ashgate, 2003. [ Links ]

LAMMENRANTA, M. “The Pyrrhonian Problematic.” In The Oxford Handbook of Skepticism. Oxford: Oxford University Press , 2008. [ Links ]

LONG, A. A. “Aristotle and the History of Greek Scepticism.” In Studies in Aristotle, edited by D. J. O’Meara, 79-106. Washington, DC: Catholic University of America Press, 1981. [ Links ]

MACHUCA, D. E. “The Pyrrhonian Argument from Possible Disagreement.” Archiv Für Geschichte Der Philosophie 93 (2), 148-61, 2011. [ Links ]

MANSFELD, J. “Diaphonia: The Argument of Alexander De Fato Chs. 1-2.” Phronesis 33 (2): 181-207, 1988. [ Links ]

PERIN, C. The Demands of Reason. Philosophy. Oxford: Oxford University Press , 2010. [ Links ]

PRITCHARD, D. “Doubt Undogmatized: Pyrrhonian Scepticism, Epistemological Externalism and the ‘Metaepistemological’ Challenge.” Principia 4 (2): 187-214, 2000. [ Links ]

_____. What Is This Thing Called Knowledge? Oxon: Routledge, 2006. [ Links ]

SEXTUS EMPIRICUS. Outlines of Scepticism. Edited by Julia Annas and Jonathan Barnes. Philosophy. Cambridge: Cambridge University Press , 2000. [ Links ]

STRIKER, G. “Scepticism as a Kind of Philosophy.” Classical Quarterly, 113-29, 2001. [ Links ]

THORSRUD, H. Ancient Scepticism. Stocksfield: Aucmen, 2009. [ Links ]

TURRI, J. “Pyrrhonian Skepticism Meets Speech-Act Theory.” International Journal for the Study of Skepticism 2 (2): 83-98, 2012. [ Links ]

VÁZQUEZ, D. “Reason in Check: The Skepticism of Sextus Empiricus.” Hermathena, no. 186: 43-57, 2009. [ Links ]

VOGT, K. M. “Appearances and Assent: Sceptical Belief Reconsidered.” The Classical Quarterly 62 (2): 648-63, 2012. [ Links ]

_____. and Sharffenberger, E. “Text, Translation and Commentary.” In Pyrrhonian Skepticism in Diogenes Laertius, edited by Katja Maria Vogt, 16-74. Tübingen: Moher Siebeck, 2015. [ Links ]

WILLIAMS, M. “Scepticism without Theory.” Philosophy 41 (3): 547-88, 1988. [ Links ]

_____. “The Agrippan Argument and Two Forms of Skepticism.” In Pyrrhonian Skepticism, edited by Walter Sinnott-Armstrong, 121-45. New Mexico: Oxford University Press, 2004. [ Links ]

_____. “The Agrippan Problem, Then and Now.” International Journal for the Study of Skepticism 5 (2): 80-106, 2015. [ Links ]

WOODRUFF, P. “The Pyrrhonian Modes.” In The Cambridge Companion to Ancient Scepticism, edited by Richard Bett, 208-31. Cambridge: Cambridge University Press , 2010. [ Links ]

WRIGHT, S. “Does Klein’s Infinitism Offer a Response to Agrippa’s Trilemma?” Synthese 190 (6): 1113-30, 2013. [ Links ]

1For the description of each of these modes see section 2, below. For the notion of ‘sceptical mode’ in general, see Sextus Empiricus’ Outlines of Scepticism [abbreviated hereafter as PH] 1, 31-35.

2See, for example, Bailey (1990), Barnes (1990b, 1990a), Fogelin (1994a), Pritchard (2000, 190), Striker (2001), Hankinson (1995, 163), Bett (2005), and Machuca (2011).

3See, for example, Greco (2006, 9) and Turri (2012). Brennan and Lee (2014, 257-58) have recently offered a version of the argument as a tertalemma, including the relativity mode, traditionally ignored or understood as part of the triggering strategy. I shall come back to the inclusion of relativity later.

4For the distinction between urban and rustic scepticism, see Galen, diff. puls. 7.711K, praenot 14.628K. The labels were popularized in the contemporary literature by Barnes (1982).

5See, for example, Klein (2003) and Pritchard (2006, chap. 4).

6For recent discussion on Agrippa’s trilemma see, for example, Gerken (2012), Wright (2013) and Williams (2015)

7In Lammenranta’s original version he talks about beliefs, but I think it is more accurate to talk about claims in a dialectical interchange.

8See also Perin (2010) and Fogelin (1994, 114-116).

9Compare, for example, my the proposal in Vázquez (2009), Thorsrud (2009, 147-60), Machuca (2011), Brennan & Lee and Bullock.

10Later, Barnes (1990b, 116-20) offers two more models; one using the two modes (PH 1, 178-179) to deploy disagreement, regress and circularity, and another with three modes (hypothesis, reciprocity and regress).

11See also Woodruff (2010, 226).

12See also Williams (2004, 121-22) and Thorsrud (2009, 151). Similarly, Hankinson (1995, 163) divides them into ‘material’ and ‘formal’ modes, and seems to agree that this is a helpful distinction, but not necessarily that it is the way they are used by the sceptic.

13For a similar proposal see Bullock (2016).

14This aspect of Pyrrhonian scepticism has been noted by various scholars but few reconstructions of the system of the five modes take notice.

15There are some antecedents to the five modes, most clearly in Aristotle’s Posterior Analytics A 3. See Long (1981) and Barnes (1975, 103-4, 1987, 1990b, 120-21). Some modes may be traced back to Plato’s Teaeteetus (152d-e), Lysis (219c8-10) and Euthydemus. However, I leave the discussion about the origins of each of the five modes for another time, and focus on their form as preserved in Diogenes Laertius and Sextus.

16This is a sceptic of whom we know nothing else. Strictly speaking, though, the phrase Οἱ δὲ περὶ Ἀγρίππαν in DL 9, 88 is ambiguous between Agrippa, Agrippa and his associates, or Agrippa’s followers. See Barnes (1990a, 4266). See also Barnes (1990b, viii). Sextus, in contrast, only reports that the five modes were introduced after Aenesidemus. See PH 1, 36 and 164.

17For the hypothesis of the common source and its possible authorship, see Barnes (1992), who ends up unpersuaded by any of the possible options, but finds the idea that it could have been Agrippa himself attractive. Janáček (1970), however, thinks that Diogenes is a more faithful copyist of the original source. For the discussion see Barnes (1992, 4268-73). I think that another plausible option is that Diogenes is copying from an earlier source, probably Agrippa, and that Sextus has access to a modified and more dialectical version of the five modes, unless the ‘more recent sceptics’ is a veiled reference to himself and his own philosophical school.

18All Sextus’ transl. by Annas & Barnes (2000), with minor modifications.

19Transl. by Scharffenberger & Vogt (2015).

20For a detailed comparison and discussion, see Barnes (1992, 4263-73, 1987, 1990a) and Janáček (1970, 1972).

21 Barnes (1992, 4254) defines a metadogmatist as ‘someone who holds views about the cognitive status of certain propositions - that we do or do not, can or cannot, know or believe that certain things are thus and so.’

23Note that in PH 1, 165, δι' ἣν refers to ἀνεπίκριτον στάσιν. What causes the suspension of judgement is the fact that the disagreement is undecided. For the decision to translate ἀνεπίκριτος as ‘undecided’ instead of ‘undecidable,’ see Barnes (1990b, 16-20). See also Machuca (2011).

24This has been already suggested by Vogt (2015, 63).

25See Annas & Barnes (1985, 142-43) and Barnes (1992, 4274-75).

26See Annas & Barnes (1985, 97-98, 144-45), Barnes (1990b, 113), Hankinson (1995, 163, 166), Woodruff (2010, 224), and Bett (2010).

27See also PH 3, 37; M 8, 337a.

28See, for example, Williams (2015, 84).

29Finally, we might be tempted to include Sextus’ exposition of the two modes at PH 1, 178-179. However, in this passage it is not so clear how many modes are being used. A first reading suggests only three modes are at play, namely disagreement, then reciprocity or regress, and finally, instead of using the hypothetical mode, Sextus sends us back to disagreement. Alternatively, one might think that he implicitly starts with relativity, which leads to disagreement, and that at the end we do have a proper deployment of the hypothetical mode and only then we go back to disagreement. Whatever the case, the only useful piece of information is that this passage makes explicit that after using the other modes the sceptic can reintroduce disagreement.

30To decide the disjunctions in this list the sceptic could simply choose one of the options based on his appearances at the moment.

31See, for example, Corti (2009).

32For the difference between assent and appearances, see Vogt (2012).

33One may worry that by following these instructions, Sextus is accepting or denying claims and if that is so, it seems that even if he ends up suspending judgement over a topic, he also ends up committed to a larger set of claims in the process. If that were true, the use of the five modes backfires against the sceptic. At every round, the dogmatic or the sceptic using the modes as self-therapy will be committed to more and more claims instead of less. But since these are steps in a list of instructions, meaning actions to be taken, the sceptic can follow her criterion for action, which are her appearances here and now (so the ‘yes’ and ‘no’ answers in the representation are a shortcut for ‘it appears so’ and ‘it does not appear so’, respectively). So, the sceptic will not gather more claims, but will only report his appearances at the time.

Article info CDD: 501

Note This research was partially funded by a FAPESP fellowship number 2016/11848-9. I also wish to thank Gabriela Martínez Sainz for her support, and audiences at Oxford and Porto for their feedback.

Received: July 23, 2019; Revised: September 11, 2019; Accepted: September 11, 2019

Creative Commons License This is an open-access article distributed under the terms of the Creative Commons Attribution License