Wittgenstein’s On Certainty: a sui generis form of foundationalism? Abstract This paper critically engages with the thesis according to which Wittgenstein in On Certainty propounded a highly original form of epistemological foundationalism. It starts by briefly presenting Avrum Stroll’s version of the thesis and goes on to discuss Michael Williams’ criticism of it. It argues that, while instructive, Williams’ objections are mainly terminological and do not provide a conclusive argument. Moreover, they hinge on an unfair reconstruction of Stroll’s position. In the second half, it presents a more substantial criticism of the foundationalist interpretation, arguing that (i) it fails to take into account the undeniable holistic strand in On Certainty, and (ii) it misunderstands Wittgenstein as being engaged in a project of justification of our knowledge. It concludes that Wittgenstein’s last notes do not commit him to any form of foundationalism. Keywords: Wittgenstein; foundationalism; certainty; scepticism 1. Introduction In this paper I discuss a particular reading of Wittgenstein’s final manuscripts in On Certainty (henceforth, OC) that takes Wittgenstein as suggesting a peculiar form of epistemological foundationalism. Although the idea was first advanced by Conway (1989), its most influential formulation was given by Stroll (1994). More recently, this interpretation has been taken up and defended by Moyal-Sharrock (2005). Although it is not a widely accepted reading – indeed, its proponents are certainly a minority – the issue is worth discussing for two reasons. For one thing, it makes an extremely bold claim. It ascribes to Wittgenstein an epistemological theory and that alone leads to an apparent clash with a fundamental tenet of Wittgenstein’s later thought: the antitheoretical methodological stance of his later (post-1929) period. Second, the foundationalist reading is supported by prima facie good textual evidence,1 and even though its defenders are a minority, they can count among them refined philosophers and Wittgenstein scholars such as Avrum Stroll and Danièle Moyal-Sharrock, who are also both proponents of (versions of) the so- As Stroll points out, “explicitly foundational language” occurs frequently in On Certainty. By foundationalist language he means words like “Boden”, “Grund”, “Fundament” and grammatical variations of them (1994, p. 142). 1 1 called “framework reading” of On Certainty – one of the most influential family of interpretations of Wittgenstein’s last work.2 In the following I start by giving a brief sketch of Stroll’s foundationalist account of On Certainty (Section 1). With that as a background, I present and discuss Michael Williams’ critique of Stroll’s view (Section 2). I argue that although Williams’ observations are commendable, the very nature of his approach allows the foundationalist to manoeuvre out of the criticism. Moreover, I suggest that Williams fails to give a fair reconstruction of Stroll’s position. However, I shall argue that there is space for another kind of critique which to my mind is even more damning for the foundationalist reading. In order to show that, I first prepare the field by dwelling more on Stroll’s original account and then conclude by proposing my own criticism of it (Sections 3-4). 2. Stroll: A sui generis form of foundationalism In the ninth chapter of his highly influential book on OC, Stroll offers a fairly loose characterization of foundationalism, which he understands as a structural notion. To do so he makes use of a conceptual model which is supposed to capture the basic insights that lie behind the various historical forms of foundationalism. Having done that, Stroll then moves on to identify the distinctive features of Wittgenstein’s alleged view. These supposedly allow for a response to the general preoccupation regarding the apparent inconsistency between any epistemological theory (foundationalism included) and Wittgenstein’s own meta-philosophical commitments – that is, his commitments to a descriptive philosophy which refrains from putting forth theses and replaces explanation with description (Wittgenstein 1953, §109, 126).3 Moreover, according to Stroll, the features that make Wittgenstein’s a sui generis form of foundationalism are precisely the same that allow it to overcome some of the historical shortcomings of more traditional forms of the theory. Stroll’s conceptual model for foundationalism takes the form of an inverted pyramid composed of two class of items, F and R, such that R depends on F but not vice versa, and F depends on nothing. Stroll’s idea is that any kind of theory that shares this form is correctly described as foundationalist. In logic, for instance, foundationalism is well represented by axiomatic systems, with axioms and principles of inference forming the basis (F) on which the set of theorems (R) depends.4 2 Other important members of the framework reading are Coliva and Pritchard. However, not all members of this group share the same views on the alleged foundationalist vein in OC. Coliva, for instance, strongly rejects it (see Coliva (2010, pp. 166-170). 3 I will not discuss this delicate issue here. Stroll himself does not argue extensively for it. He only points out Wittgenstein’s open endorsement of the method of description in OC §§189, 212, and his preference for descriptive over normative language in passages like OC §232. But although this goes some way towards showing Wittgenstein’s loyalty to his methodological tenets, it doesn’t help to resolve the issue of the compatibility between such principles and any foundationalist reading of OC. 4 In this case, the relation of “dependence” is the relation of derivability in the system. 2 Epistemological foundationalism is then obtained by interpreting R and F as items of knowledge and thus the relation of dependence as the relation of justification that holds between the foundations (F) of our knowledge and the reminder (R) of it. According to Stroll, one historical instantiation of epistemological foundationalism is the Cartesian picture of knowledge as resting on a restricted basis constituted by one privileged piece of knowledge (the cogito) which provides the foundations upon which the entire mansion of our knowledge rests. What distinguishes Wittgenstein from traditional foundationalists is his driving a categorical distinction between the foundations and the superstructure. In the case of Descartes, the foundation (the cogito) is itself an item of knowledge, albeit a privileged one: it shares the same “nature” of, or belongs to the same category as, the rest of the building it supports. Stroll labels this feature, shared by all traditional forms of foundationalism, the doctrine of “homogeneous foundations” (Stroll, 1994, p. 141).5 In On Certainty, on the other hand, we witness the rejection of such a doctrine. For the foundations are non-epistemic in nature. In addition, according to Stroll, Wittgenstein’s definitive account6 of the foundations takes them to be non-propositional since the notions of “being true or false, known or not known, justified or unjustified do not apply to them, and these are usually taken to be the defining features of propositions” (Stroll, 1994, p. 146). Our relation to them is best characterized as “certainty” rather than “knowledge”, the two being categorically distinct (OC §308). This idea of “driving a wedge between the notions of certainty and knowledge” is, in the eyes of Stroll, “one of the deepest ideas of On Certainty” (1994, p. 134). It is also what makes Wittgenstein’s purported foundationalism an effective response to the infinite regress problem, whereas classical versions of foundationalism fail.7 The reason for this is that the classical foundationalist’s claim that the justification of his basic beliefs does not depend on other beliefs (that is, they are not inferentially justified) may still be challenged by the sceptics. After all, what makes these basic beliefs more than arbitrary assumptions? If they are not inferentially justified, how can they be justified at all? The traditional appeal to some special property (such as selfevidence) that picks apart basic beliefs from non-basic ones succeeds in avoiding the issue of arbitrariness only in so far as we can make clear sense of that notion. And even assuming we can do that, it still remains to be shown that such a property picks out a class of beliefs wide enough to play 5 The same holds for axiomatic systems in logic: the axioms are themselves theorems. In the complex story told by Stroll, OC shows Wittgenstein oscillating between two different versions of the nature of the foundations. In one case he speaks of the foundations in propositional terms, in the other he identifies what is fundamental with various non-intellectual candidates, such as “acting, being trained in communal practices, instinct, and so on.” (1994, p. 146). According to Stroll, we can see in OC a non-linear development from the former view to the latter. The claim is far from unproblematic. 7 An idea also shared by Moyal-Sharrock (2016). 6 3 the foundational role assigned to it by the foundationalist programme. Fulfilling both these conditions has notoriously proved hard. On the (alleged) Wittgensteinian account, on the contrary, the sceptical challenge to the foundations is ruled out for grammatical reasons. Doubting them is deemed unintelligible (§§154, 231, 247, 261, 450) or outright meaningless (§§2, 56-58, 115, 310312, 372). Roughly, since the concept of knowing (and related concepts such as doubting, believing, learning, etc.) only meaningfully attaches to (empirical) propositions it is inapplicable to the non-propositional foundations of our knowledge. The categorical divide between the foundations and what rests on them cuts the grounds from under the sceptics’ feet. To borrow an expression from Pritchard, the categorical divide is what makes Wittgenstein’s an “undercutting” solution to the sceptical problem:8 Given his form of foundationalism the regress problem does not arise. It arises for traditionalists because they assume that the question, How do you know that that which stands fast for us is certain? is always applicable. And they assume that because they think that the foundation and what rests on it belong to the same category. But for Wittgenstein’s form of foundationalism the question is not applicable and, in fact, embodies a category mistake. One cannot sensibly ask of that which is certain whether it is known (or not known) or true (or false); for what is meant by certitude is not susceptible to such ascriptions. The sceptical question thus need not be answered. This shows again how radically Wittgenstein’s view differs from any traditional form of foundationalism. (Stroll 1994, p. 148)9 Here a first objection is in order. If the heterogeneity of foundations represents such a radical departure from the foundationalist programme as the quotation above suggests, it is unclear whether it really is legitimate to include Wittgenstein in the foundationalist family.10 At the very least, such inclusion would be at odds with the current use of the label “foundationalist” in epistemological jargon. A look at standard definitions of foundationalism shows that the idea of the justification of 8 For the distinction between undercutting and overriding solutions to the sceptical challenge, see Pritchard (2015, pp. 16-17). 9 Moyal-Sharrock here is on the same page as Stroll: “certainties have their place in a foundationalist structure as the grammatical underpinnings of our beliefs. This, it seems to me, is a modification of foundationalism, not the absence of it. That hinges lack some of the features of foundational beliefs as traditionally conceived should not prevent them from being foundational. In fact it is precisely their differing from the rest of our beliefs (in being nonpropositional and nonepistemic) that makes for the success of Wittgenstein’s foundationalism. It is the realization that what we have traditionally taken to be propositional beliefs (rationally posited or arrived at) are in fact ungrounded or logical ways of acting that allows Wittgenstein to put a stop to the regress of justification. Insisting that propositions lie at the foundation may satisfy epistemologists, but it will not solve the problem of regress.” (Moyal-Sharrock 2016, p. 110) 10 One way in which a Stroll could reply to this objection is by providing different epistemological theories that share Wittgenstein’s “heterogenous foundation” tenet and that we usually regard as foundationalist. Stroll hints at the existence of such theories when he says that Wittgenstein’s is “one of the few” versions of foundationalism to take this step (1994, p. 141). Unfortunately, though, he does not point to any other member of that restricted community. 4 the basic beliefs (the foundations) is usually built into the very notion of foundationalism.11 If the unjustifiability of the foundations is what defines Wittgenstein conception of the structure of our epistemic system, then rather than a foundationalism of a special character Wittgenstein’s “foundationalism” is no foundationalism at all. BonJour, for example, regards a view of the sort as an alternative to foundationalism, rather than a species of it – an alternative which, on his view, is in the end unsatisfactory (1985, pp. 22-23). Leaving this point aside, I will now focus on another objection against Stroll’s reading of OC, to be found in Williams (2005). There Williams argues that Wittgenstein can be regarded as a foundationalist only provided that we accept an extremely loose definition of foundationalism. Stroll’s reading, he claims, trades on the looseness of his formal model for foundationalism. I take it that Williams’ critique, while making a strong point against Stroll’s labelling Wittgenstein’s position “foundationalist”, does not go much further than the objection above in that it makes a merely terminological point. In addition, the objection itself can be challenged. However, I argue that Williams’ objection can be supplemented by a more substantial one, according to which even if we grant the correctness Stroll’s definition of foundationalism there is no room for Wittgenstein within it. 3. Williams’ terminological criticism Williams himself openly recognized that his criticism is mainly terminological, although he argued that this does not subtract from its importance: Is the question of Wittgenstein’s ‘foundationalism’ just terminological? In a way, it is. But words matter, not least because they have a history, in the course of which they pick up associations that cannot easily be cancelled. Thus, while the term ‘foundationalism’ can be applied with some latitude, there are limits. If the term is not to be virtually drained of meaning, we must recognize that foundationalism is more than the view that there are certainties of some kind or other, so that scepticism goes wrong somehow. It is a theoretical position in epistemology involving distinctive commitments, methodological and theoretical. (Williams 2005, pp. 49-50) The objection, I take it, is that Stroll can only ascribe foundationalism to Wittgenstein by constructing a deliberately loose notion of foundationalism, to the point that the term itself is “drained of meaning”. Williams then spells out the “distinctive commitments” of foundationalism as being commitments to: 11 Even though such justification cannot be itself inferential, on pain of infinite regress. 5 1. The universality of the foundations. A specific kind of knowledge must provide the foundations for knowledge tout-court. This constraint rules out the possibility of beliefs (or propositions) that are basic only with respect to some specific domain: “This commitment – says Williams – is what makes foundationalism a theory of knowledge” (i.e., of knowledge without further qualification). 2. The specifiability of the basic beliefs. Foundationalism requires a definite, qualitative, divide between basis and superstructure and a “principle of demarcation” for distinguishing items that belong to the former from those that belong to the latter. 3. The semantic and epistemic independence of the basis. According to Williams, the “insistence on the autonomy of basic beliefs is what separates foundationalism from its traditional rival, the coherence theory (whose adherents are typically semantic as well as epistemic holists)”. 4. The rational adequacy of the basis, which must “offer a way of rationally deciding what we should and should not accept at the non-basic level”. In order to do so, there must be a tight connection between the two levels, such that one doesn’t need to appeal to anything other than what is included in the foundations to ground all that which rests above it. (Williams 2005, pp. 50-51) If these commitments are indeed embedded in the very idea of foundationalism, then Wittgenstein’s failure to comply to them will undermine any foundationalist reading of OC. Let’s look at them one at a time. Looking at OC we are immediately aware of the variety of cases that Wittgensteins considers to be exempt from doubt and mistake. (According to Stroll this exemption is the criterion that makes something foundational.) Wittgenstein himself recognizes this seemingly irreducible variety in one of the closing passages of OC: There are, however, certain types of cases in which I rightly say I cannot be making a mistake, and Moore has given a few examples of such cases. I can enumerate various typical cases, but not give any common characteristic. (N. N. cannot be mistaken about his having flown from America to England a few days ago. Only if he is mad can he take anything else to be possible.) (OC §674) In fact, among the examples given by Wittgenstein in OC we can count propositions about our bodies (“Here is a hand” §1) and about familiar middle-size objects (“There is an armchair over there” §193), beliefs concerning basic scientific facts (“I have a brain” §4, “Water boils at 100° C” §293), beliefs regarding one’s own experience (“I have never been on the moon” §111) and beliefs 6 that constitute domains of inquiry such as history (“The earth has existed for many year past” §411). It seems clear that not all these beliefs can be said to be universal in the same sense. Indeed, one cannot underestimate the heterogeneity of the certainties. Some are arguably shared by every adult human being; others are culturally specific. Moreover, although the denial of a basic judgement never amounts to a mistake, what it does amount to varies greatly: sometimes we regard it as a sign of a mental disturbance, sometimes as the manifestation of a different worldview, etc.12 The specifiability condition is even more problematic. According to Williams’ reading, Wittgenstein openly denies the theoretical tractability of the distinction between basis and superstructure. There is no criterion to divide the one from the other, for they do not belong to two distinct theoretical kinds. This objection, if sustainable, threatens to crush Stroll’s account, which depends on there being a strong categorical distinction between the foundations and the mansion. Williams quotes passages from the beginning of OC, where Wittgenstein suggests the impossibility of formulating a criterion or a rule for distinguishing the circumstances where mistake (and therefore doubt) is excluded from the rest (§§25-29). Perhaps more to the point are the remarks presenting the river-bed metaphor, where Wittgenstein points out the lack of a clear-cut boundary between basic and non-basic proposition: It might be imagined that some propositions, of the form of empirical propositions, were hardened and functioned as channels for such empirical propositions as were not hardened but fluid; and that this relation altered with time, in that fluid propositions hardened, and hard ones became fluid. (OC §96) The mythology may change back into a state of flux, the river-bed of thoughts may shift. But I distinguish between the movement of the waters on the river-bed and the shift of the bed itself; though there is not a sharp division of the one from the other. (OC §97) It seems to me that this lack of a sharp boundary is a feature that remains constant throughout OC, in both earlier and later remarks: This situation is thus not the same for a proposition like ‘At this distance from the sun there is a planet’ and ‘Here is a hand’ (namely my own hand). The second can’t be called a hypothesis. But there isn’t a sharp boundary line between them. (OC §52) 12 Kusch (2016) makes the interesting proposal that our reaction to a denial of a certainty is a function of the cultural distance between us and the certainty-denier. 7 ‘The question [about the existence of the earth for many years past] doesn’t arise at all.’ Its answer would characterize a method. But there is no sharp boundary between methodological propositions and propositions within a method. (OC §318) According to Williams, the thesis of independence is at odds with Wittgenstein’s “limited semantic holism” which is implied by his conception of meaning (Williams 2005, p. 54). Whether or not this is true (I think it is), the independence thesis surely contradicts many sections of OC where Wittgenstein appears to endorse a form of epistemic holism. Those are the remarks where Wittgenstein claims that our knowledge forms a system and “only within this system has a particular bit the value we give it” (OC §410). Although a certain degree of asymmetry is granted – denying it would be to deny that our system is internally articulated (contra OC §213) – it is far from clear that a specific portion of that system carries all the burden of justification. Far from it, sometimes Wittgenstein seems to suggest that there is a relation of dependence in what Stroll would call the R–F direction, and that alone is incompatible with the idea of the autonomy of the basis. What Wittgenstein calls the “foundations” or “grounds” are dependent (in some sense of “dependence”) on what is grounded by them: I have arrived at the rock bottom of my convictions. And one might almost say that these foundation-walls are carried by the whole house. (OC §248, emphasis added) Besides, there are many passages throughout the text where Wittgenstein appears to commit to a holistic or quasi-holistic picture of our epistemic practices (OC §§102, 105, 141-144, 410-411, 603). This massive presence is troubling for any foundationalist reading. In many of those passages, the typically stratified picture of knowledge of the foundationalist gives way to the notion of “system”: The child learns to believe a host of things. I.e. it learns to act according to these beliefs. Bit by bit there forms a system of what is believed, and in that system some things stand unshakeably fast and some are more or less liable to shift. What stands fast does so, not because it is intrinsically obvious or convincing; it is rather held fast by what lies around it. (OC §144, emphasis added) Moreover, I believe it can be shown that in the later parts of OC Wittgenstein tries to deal with the consequences of such a picture, more specifically with the spectre of relativism.13 The fact that he goes to great pains to resolve the issues connected with this holistic picture suggests that he 13 See, for instance, OC §§594-595, 599. 8 endorsed the intuition behind it, an intuition that can hardly be reconciled with any form of foundationalism. Finally, the rational adequacy of the basic beliefs is a necessary condition for foundationalism if it is to answer the challenges of the sceptic. Only the universality of the grounds together with the tightness of the connection between the basic and the non-basic level can provide a solution to sceptical arguments from underdetermination.14 For the foundationalist needs a logical connection between the basis and the superstructure, if the basis is to “resolve any genuinely empirical question” (Williams 2005, p. 56). As we have seen above, not all Wittgenstein’s certainties can be regarded as universal. Moreover, Williams reads remarks like OC §92 as saying that our basic beliefs are not even rationally adequate because they do not suffice for deciding every dispute at the non-basic level: Men have believed that they could make rain; why should not a king be brought up in the belief that the world began with him? And if Moore and this king were to meet and discuss, could Moore really prove his belief to be the right one? I do not say that Moore could not convert the king to his view, but it would be a conversion of a special kind; the king would be brought to look at the world in a different way. (OC §92) However, I think that Williams here makes a mistake. In the quoted passage, Wittgenstein is not saying that our basic beliefs are not broad or powerful enough to decide some disputes at the nonbasic level. For if we were to locate the disagreement between Moore and the king, we would say that it is precisely at the basic level. It is an instance of what has come to be known as “deep disagreement”.15 Wittgenstein is here pointing out that there are (or it is conceivable that there might be) cases of disagreement which do not to allow for a rational resolution. In OC he goes to a great length to describe imaginary scenarios of disagreement in order to investigate how we would react to them. One running theme is that in those cases the appeal to rational arguments turns out to be ineffective and we are forced to either switch to different (non-argumentative) forms of “persuasion” or “conversion” (OC §§ 92, 262, 612), or simply to give up the dialogue altogether (OC §§120, 238). What I mean to say is that in writing those remarks Wittgenstein is not concerned with the issue of rational adequacy. What he is saying is that rational argumentation inevitably comes to an end (OC §110), and the reason for this is that all rational argumentation requires that certain presuppositions are fixed: inquiries can take place only within a system in which many things are not themselves 14 In short, those arguments that hinge on the fact that the whole of our basic beliefs (e.g., beliefs about our sense experiences) is compatible with different sets of beliefs about the world (e.g., that I am awake and conscious or dreaming, that I am a human being or a brain in a vat). 15 The debate on deep disagreement was initiated by Fogelin (1985) and has its roots in On Certainty. 9 doubted. This point is well expressed in the hinge metaphor of OC §341. That passage makes a structural point, it tells us about the characteristics of our inquiring, but by itself it does not settle many of the issues connected with the nature of those “hinges”. More specifically, it does not say whether hinges are in any way absolute – that is, fixed independently of the context. But Wittgenstein clearly thought that in certain conceivable contexts, such as the one depicted in §92, the disputants do not share enough presuppositions for a rational argument to take place. Lacking a system of presuppositions there can be no such thing as a proof, for the resources needed to decide what counts as a reason for what are lacking as well. It is only within a system that the game of evidence can be played. Moore and the king lack a shared system within which their discussion could take an argumentative form. This suffices to show why this specific passage Williams misses the mark. Wittgenstein’s problem is not that of the rational adequacy of the basis. That is a problem for foundationalists, for they must make sure that their basis is sufficiently broad to support the whole of our knowledge. Wittgenstein appears to be more concerned with the possibility of alternative systems of inquiry, and this is an that he shares with coherence theories and theories of a holistic spirit in general. Therefore, Williams’ last point could be amended thus. The problem with Stroll’s view is not that Wittgenstein’s “hinges” do not constitute a rationally adequate basis, but the fact that in OC we do not find Wittgenstein dealing with the issue of rational adequacy of the basis at all. (This ceases to be surprising if we consider that in that work a clear-cut distinction between basis and mansion is lacking as well.) What we do find in OC is Wittgenstein’s concerning himself with the issue of the possibility of altogether different epistemic systems—a possibility that arises from scenarios of disagreement in our most deeply entrenched convictions. Leaving aside these issues, Williams’ critique succeeds in showing that if the commitments outlined above are part and parcel of the very notion of foundationalism, then Wittgenstein’s reflections on our epistemic concepts and practices in OC cannot be regarded as leading to a foundationalist account of knowledge. Stroll’s account is dubious precisely because it appears to strip the concept of foundationalism of such commitments. As a result, his labelling of Wittgenstein a foundationalist trades on a very loose definition of foundationalism. As I mentioned before, though, this kind of objection risks to turn on a merely terminological point, for it allows for the conditional adequacy of the label. In other words, if we accept the loose definition, then Wittgenstein is a foundationalist. Of course, this does not mean that Williams’ critique is less concerning, for given the much more restrictive use of the term “foundationalism” in current literature it is unclear why we should go for a much looser definition. The burden lies on the shoulder of any defender of a foundationalist reading of OC to show why we should reject Williams’ restrictive conditions for foundationalism. 10 The objection that now I move on to make is quite independent from all of this. First, I argue that Stroll’s account of foundationalism is less loose than it seems at first sight. I then claim that for this reason Williams’ criticism is not entirely fair. Stroll’s account is more nuanced than Williams takes it to be. For one thing, a closer look reveals that some of the same commitments proposed by Williams are actually present, albeit implicitly, in Stroll’s conceptual model. Moreover, Stroll’s account doesn’t stop at such a skeletal model, but explicitly puts some meat on it. New commitments are thus embedded in the picture. However, I argue that his does not necessarily play in favour of the foundationalist reading. For those commitments, once properly displayed, can be shown to be at odds with what Wittgenstein says in OC. As a result, I will try to show that even if we accept Stroll’s enriched model for foundationalism, it turns out not to fit Wittgenstein’s position. Wittgenstein’s own account does not embody such a model. As one can easily see, if this kind of objection can be sustained it will rule out even the conditional adequacy of Stroll’s labelling. One could then say that Stroll’s account, as understood by Williams, is too loose to provide a meaningful definition of foundationalism, and understood in more detail turns out to be too restrictive to do justice to Wittgenstein’s reflections in OC. For this reason, we should just give up the foundationalist picture altogether. 3. A closer look at Stroll’s account First, I suggest we take a closer look at the seemingly innocuous conceptual model proposed by Stroll. Some features of the by-now-familiar twofold structure are presented thus: [This conceptual model] holds that there is some asymmetrical relationship of dependence between F and R, whatever these are taken to be, and that F is not dependent on anything else. [. . .]. The thrust of the model is that F somehow supports R and is itself not supported by anything. The idea that F is not supported by anything is generally taken to be another way of saying that F is foundational. [...] [W]e shall have to add another element to it. It is difficult to state this in a purely formal way and without giving specific examples, so some examples will follow. The formal point is that F will either be a single thing or a very limited number of things if there is more than one F, whereas R will be complex, possessing scope and amplitude. The formal point will thus flesh out the model, giving it topologically the shape of an inverted pyramid. The main body of the pyramid will rest on something equivalent to a simple base. It will be broader at its apex than at its base. We can call the base the foundation and what rests on it the mansion. (1994, pp. 143-144) From this passage we can extract four features of the conceptual model that up to now I had left in the shadow. The model thus presented turns on: 11 1. An asymmetric dependency relation between two classes of items, F and S, 2. The non-dependence of the foundational class F, 3. The difference in complexity between F and S (F being a simple and restricted class, S being complex and ample), 4. The uniqueness of the pyramid. All four of these commitments are embodied in Stroll’s model for foundationalism. What stands out is that the second and fourth points match the independence and universality commitments outlined by Williams. Hence, Stroll’s account is not that empty after all. However, this also means that the objections raised by Williams towards those commitments could be applied again here. Furthermore, the claim that the foundations form a simple and very restricted class of propositions or beliefs (point 3) clashes with the variety of Wittgensteinian examples of things that are “held fast” and with his claim that there are “countless general empirical propositions that count as certain for us” (OC §273). More importantly, later in the chapter Stroll gives more substance to his depiction of foundationalism by recourse to nine different features that – he argues – have been partially shared by all versions of foundationalism.16 Stroll’s depiction of foundationalism is more articulated than it seems at first sight. He argues that foundationalism, while it looks like a simple idea, is composed by “multifarious strands, each of which contributes to a rather complex intuition in a somewhat different way” (1987, p. 280). Each historical instantiation of it has shared some of the nine features with some of the other members of the foundationalist family. None of them, though, has embodied all the features – none, that is, until Wittgenstein’s (1994, p. 148). The striking proposal, then, is that Wittgenstein is the most foundationalist of the foundationalists. One might suspect that this way of defining foundationalism is not more restrictive (hence not more convincing) than the overtly loose one considered above. For while Stroll does in fact multiply the features that characterize foundationalism, he also makes it clear that adherence to all of them is not required. The fear is that even this formulation turns out to be too inclusive. However, Stroll assures us that the definition is restrictive enough, and points out that it rightly excludes epistemological theories that none of us would claim belong to the foundationalist camp. His examples are Quine and Popper. And this is interesting. For I believe that Wittgenstein’s epistemological views in OC have more in common with Quine that with the traditional foundationalist. As I have pointed out above, there is an undeniable holistic strand in OC. None of the Wittgenstein scholars that have The nine features of foundationalism are “(1) stratification, (2) aberrancy, (3) non-dependence, (4) particularism or methodism, (5) publicity, (6) negational absurdity, (7) absorption, (8) certitude, and (9) standing fast.” (1994, p. 148) The list was already formulated in Stroll (1987), with the only difference of “vagueness” now replaced by “aberrancy”. 16 12 worked on OC will deny this. Some find it natural to compare those passages with the form of holism expressed by Quine’s “Two Dogmas of Empiricism”, although all of them also make sure to stress the differences between the two.17 Generally speaking, Quine’s point that every item in the “web of beliefs” is in principle revisable, does not necessarily clash with Wittgenstein’s claim that our method of inquiry presupposes that at any given point something is exempted from doubt, while it fits well Wittgenstein’s river-bed metaphor and its suggestion that nothing is absolutely immune to revision. Stroll himself recognizes an ambiguity in Wittgenstein’s account of certainties. Along with foundationalist-sounding remarks, there are passages in OC that have a strong relativistic flavour, in that they seem to allow for the possibility of there being different sets of immune commitments. Thus, after quoting the river-bed sections, Stroll writes: Wittgenstein’s idea here - frequently asserted throughout On Certainty - is that in any activity or principle something must remain immune to doubt, evidence, consideration, and justification. But that in principle, depending on the activity, what is exempt from consideration at a given time may come under consideration at another. But nothing per se is absolutely immune to all revision and consideration. There are thus no propositions which absolutely and independently of any contextual circumstances stand fast for us. Though this is unquestionably a form of foundationalism, there is the danger of interpreting it as if it were a variant of Popper’s boot-strap theory and thus as ultimately non-foundational at all. (1987, p. 297) For this reason (to avoid relativistic conclusion and undue assimilations to Popper), he argues that we would “do better in emphasizing the absolutist interpretation”, which can also find some textual support in other sections of OC. Here a couple of remarks are needed. First, it is unclear (to me) what it is that makes OC §§96-97 “unquestionably” foundationalist. If anything, these remarks row towards the opposite bank, in the direction of anti-foundationalism. This seems to be also the most natural conclusion to draw from what Stroll says in that same passage: “nothing per se is absolutely immune to all revision and consideration. There are thus no propositions which absolutely and independently of any contextual circumstances stand fast for us”. What Stroll does get right, however, is the fact that such relativistic-sounding remarks occur throughout OC. I believe that for this reason we should be wary of any attempt to underestimate their importance. Unfortunately, later on, Stroll himself falls prey to this mistake when he claims that Wittgenstein progressively abandoned this relativistic picture and moved towards an absolutistic one, according to which our basic commitments are embedded in our 17 See, for instance, Gibson (1996). For a classic discussion on the affinities and differences between the two philosophers (with a stress on the differences), see chapter 7 of Hacker (1997). 13 inherited background which is not amenable to revision or rejection at all (1994, p. 157). In his mind, this absolute immunity from revision characterizes what became the dominant view in OC, and it is what distinguishes Wittgenstein’s certainties from Peirce’s rather relativistic “regulative assumptions”. The problem here is that this diachronic reading is not sufficiently supported by the textual evidence. In Stroll’s view the movement towards the absolutistic conception of our basic beliefs coincides with Wittgenstein’s abandonment of a propositional reading of them in favour of a non-propositional account. In other words, Stroll’s sees in OC a development from a relative foundationalism that takes an ever-changing set of privileged propositions as forming the groundless ground of our beliefs to a foundationalism grounded on our non-reflective ways of acting, or an animal-like instinct, or a picture of the world absorbed from birth and early training. But this chronological reading is unwarranted, for Wittgenstein talks about propositions exempted from doubt throughout the text.18 It remains to be shown, then, why one should privilege the absolutist reading over the relativistic one. 5. Foundationalism without certainty? To conclude, I now take issue with one specific point in Stroll’s account of foundationalism. If my critique is sound, then Wittgenstein is not a foundationalism even according to Stroll’s standards. Among the nine features of foundationalism, Stroll lists “certainty”. From what he writes it is clear that Stroll regards it as a necessary but non-sufficient feature of foundationalism. He writes that it is an interesting question “whether someone could hold a form of foundationalism while denying there is certitude. To my knowledge no notable philosopher has done so.” (1994, p. 155).19 However, he does not make clear what he means by “certitude” here. In our common understanding of the word, “certain” often indicates some psychological state, expresses complete conviction (OC §194). But subjective certainty is compatible with my being mistaken, and so with my not knowing. In the relevant epistemological sense, though, for something to be certain is for it to be a true and indefeasible belief, in the sense that come what may any doubt about it is impossible. This is the sense of “certain” that Descartes ascribes to the cogito. Its being certain is tied to its being impossible to doubt, where impossible means logically impossible, and by “logically” we mean something like “in virtue of the laws of thought”. Doubting the cogito is impossible because it is 18 Having recognized this, Moyal-Sharrock argues that we shouldn’t “think of Wittgenstein’s ‘propositional’ and nonpropositional accounts as consecutive, but as indicative of an ongoing, nonlinear and nonprogressive struggle, throughout On Certainty, to understand the nature of our foundational beliefs” (2005, p. 89). 19 This is also clear in Stroll (1987). There he takes certitude as being one of the four central features of foundationalism, together with stratification, non-dependence and methodism or particularism, the other five being more peripheric. Interestingly enough, when listing the nine strands in 1994 Stroll puts certitude in the eighth place. Nowhere does he say that the numeration reflects a hierarchical structure, but I think that the change remains suspicious. I suspect that he (rightly) came to see certitude as being the weak link in the chain. 14 inconsistent. Among other historical examples of defenders of certainty Stroll counts Plato. In Plato, all knowledge is certain in virtue of the ontological status of its object, the ideas. In Wittgenstein there is nothing of this sort. He does say that in certain cases doubt is ruled out as nonsensical, and in other cases as unreasonable, with no clear distinctions between the two (OC §454). He also says that this fact belongs to the logic of our inquiries (OC §342). But what he means by logic has nothing to do with metaphysically robust laws of thought, and everything to do with his notion of language-game. To say that doubt is logically excluded is to say that doubt is not permitted by the rules of the relevant language-game. There is very little in common between this notion of certainty and the platonic or cartesian one. Indeed, I believe that Wittgenstein was very suspicious about the use of “certainty” to characterize our relation to the special propositions that he discusses in OC. A large part of the remarks in OC show Wittgenstein’s attempts at finding the correct way of expressing that relation, and “certainty” is never really considered. A more promising candidate expression is “standing fast” (“feststehen”) (OC §§116, 125, 144, 151-152, 173, 225, 234-235), the immediate upside of which is that it comes with an implicit reference to a subject that holds something fast (accentuated by Wittgenstein’s own expressions “It stand fast for me” (OC §116) or “for him” (OC §151), “I hold fast” (OC §225)), whereas “certain” suggests the reference to a property or quality possessed by the belief itself. Moreover, Stroll curiously indicates “standing fast” as the ninth strand of foundationalism, separate from certainty. However, when he comes to discuss it, he doesn’t actually distinguish the two. Instead, he reduces the former to a metaphorical expression for certainty (1994, p. 155). Another way to put it is to say that “standing fast” is the form of that “certainty” takes in OC. Or better, that standing fast is what Wittgenstein replaces certainty with. One must explain the reason behind this apparent redundance in Stroll’s list of foundationalist features. The reason why Stroll is unwilling to let certainty go is, I believe, the fact that “standing fast” by itself allows for a relativistic reading in a way that “certainty” does not. It is this reference to certainty that allows Stroll to anchor Wittgenstein’s view to the foundationalist tradition in epistemology. However, there is no such thing as certainty (in the traditional sense) in OC. What we do find, instead, is communities and individuals taking things for granted and holding to them. We find them by way of observation, including observation in the peculiarly “speculative” sense adopted by Wittgenstein. And what we see does bring to the surface the possibility that what we held fast might change through space and time, and with it, the issue of relativism. 15 That is not to say that we should ascribe to Wittgenstein a form of epistemic relativism.20 His point, I suggest, is that our knowledge is in no need of foundations. The search for foundations is prompted by our having a mistaken picture of knowledge. Looking for foundations is looking for ways to justify our body of knowledge and to defend it against sceptical doubts, and Wittgenstein cannot do this successfully by simply claiming that “this is how we act”. By taking this step, Wittgenstein doesn’t mean to provide foundations for knowledge. Towards the end he is clear that knowledge is neither grounded nor ungrounded: You must bear in mind that the language-game is so to say something unpredictable. I mean: it is not based on grounds. It is not reasonable (or unreasonable). It is there—like our life. (OC §559) And the concept of knowing is coupled with that of the language-game. (OC §560) Although I disagree with the foundationalist reading that Stroll proposes, I believe that he took off from a correct insight. Even though he does not put it in these terms, Stroll does see that there is a pragmatic vein in OC.21 By that I mean that he rightly recognizes in OC the idea that knowledge has an intrinsic relation with the epistemic practices of a community. But he gives it an unwarranted foundational turn when he understands that relation in terms of justificatory dependence, i.e., as the relation between the ground and what is grounded by it. Wittgenstein is saying that all rational evaluation and all normative discourse takes place within a framework of inquiry that goes unmentioned, within a language-game which comprises discursive and non-discursive practices shared by a community of human beings. He can also be understood as saying that these practices play a foundational role, in the loose sense that it is only by taking them for granted that our game of giving and asking for reasons takes place. But this thesis, which is the upshot of observation and description, does not make Wittgenstein a foundationalist because he doesn’t take it as justifying our epistemic system. It might be that these observations on our actual epistemic practice succeed in disallowing scepticism, by showing that seriously entertaining a sceptical doubt would bring about the collapse of our very “method of doubt and inquiry” (OC §151). Sceptical doubt is then ruled out because it undermines the very framework within which what we call “doubting” takes place. But as long as they do not purport to justify our knowledge on the face of sceptical challenges, they do not commit Wittgenstein to any form of epistemological foundationalism. 20 This issue is quite complex and has commentators distributed on both sides. For a recent defence of a relativist reading of OC see Kusch (2013, 2016). For the opposite view, see Coliva (2010, pp. 188-203) and Moyal-Sharrock (2017). 21 This affinity with pragmatism was recognized, albeit not explicitly embraced by Wittgenstein himself (OC §422). 16 References BonJour L (1985) The structure of empirical knowledge. Harvard University Press, Cambridge (Mass.) Conway G D (1989) Wittgenstein on foundations. Humanities Press International, New Jersey Fogelin R (1985) The logic of deep disagreements. Informal Logic 7:1–8 Hacker P M S (1997) Wittgenstein’s place in twentienth-century analytic philosophy. Blackwell, Oxford Gibson R F (1996) Quine, Wittgenstein and holism. In: Addington R L, Glock H- J (eds) Wittgenstein and Quine. Routledge, London and New York Moyal-Sharrock D (2005) Understanding Wittgenstein’s On Certainty. Palgrave Macmillan, Basingstoke. Moyal-Sharrock D (2016) The animal in epistemology: Wittgenstein’s enactivist solution to the problem of regress. International Journal for the Study of Skepticism 6:97-119. Moyal-Sharrock D (2017) Fighting relativism: Wittgenstein and Kuhn. In: Kanzian C, Kletzl S, Mitterer J, Neges K (eds) Realism - Relativism – Constructivism. De Gruyter, Berlin/Boston Moyal-Sharrock D, Brenner W H (2005) (eds) Readings of Wittgenstein’s On Certainty. Palgrave Macmillan, Basingstoke Kusch M (2013) Annalisa Coliva on Wittgenstein and epistemic relativism. Philosophia 41:37-49 Kusch M (2016) Wittgenstein’s On Certainty and relativism In: Rinofner-Kreidl S, Wiltsche HA (eds) Analytic and continental philosophy: methods and perspectives. De Gruyter, Berlin/Boston Pritchard D (2016) Epistemic angst: Radical skepticism and the groundlessness of our believing. Princeton UP, Princeton Stroll A (1987) Foundationalism and common sense. Philosophical investigations, 10(4):279-298 Stroll A (1994) Moore and Wittgenstein on certainty. Oxford UP, New York and Oxford Williams M (2005) Why Wittgenstein isn’t a foundationalist. In: Moyal-Sharrock D, Brenner W H (eds) Readings of Wittgenstein’s On Certainty. Palgrave Macmillan, Basingstoke Wittgenstein L (1953) Philosophical investigations. Blackwell, Oxford. Wittgenstein L (1969) On certainty. Blackwell, Oxford 17