The Dynamic Strategy of Common Sense Against Radical Revisionism

Abstract

Common-sense philosophers typically maintain that common-sense propositions have a certain kind of epistemic privilege that allows them to evade the threats of skepticism or radical revisionism. But why do they have this special privilege? In response to this question, the “Common-Sense Tradition” contains many different strands of arguments. In this paper, I will develop a strategy that combines two of these strands of arguments. First, the “Dynamic Argument” (or the “starting-point argument”), inspired by Thomas Reid and Charles S. Peirce (but which will be strengthened with the help of Gilbert Harman’s epistemology of belief revision). Second, G.E. Moore’s “greater certainty argument” (interpreted along the lines of Soames’ and Pollock’s construal). This combined strategy, I will argue, is the strong core of Common-Sense Philosophy, and relies on extremely modest and widely held assumptions.

What is the significance, in contemporary analytic philosophy, of the “Philosophy of Common Sense”—the tradition of philosophical methodology going from Thomas Reid to G.E. Moore and Roderick Chisholm?¹ In 2003, looking back on the history of analytic philosophy between 1900 and 1975, Scott Soames considered that the reliance on common sense was one of “the two most important achievements that [had] emerged from the analytic tradition in this period”. He wrote:

One of the recurring themes in the best analytic work during the period has been the realization that no matter how attractive a philosophical theory might be in the abstract, it can never be more securely supported than the great mass of ordinary, pre-philosophical convictions arising from common sense, science, and other areas of inquiry about which the theory has consequences. All philosophical theories are, to some extent, tested and constrained by such convictions, and no viable theory can overturn them wholesale. (Soames 2003, p. xi)

And according to Kit Fine, this important lesson of the XXth century was still remembered and approved at the turn of the XXIst:

In this age of post-Moorean modesty, many of us are inclined to doubt that philosophy is in possession of arguments that might genuinely serve to undermine what we ordinarily believe. (Fine 2001, p. 2)

Thomas Kelly (2005, pp. 179–180) notes that this post-Moorean consensus includes philosophers as prominent as David Lewis (1996) or David Armstrong (1999).²

And yet, these optimistic claims in favour of a consensus on common-sense methodology are not themselves consensual. According to Alex Byrne and Ned Hall, “there is no consensus on whether, or to what degree, philosophy should be subordinate to common sense” (Byrne and Hall 2004). Chris Daly, in his general Introduction to Philosophical Methods, concurs (Daly 2010, p. 36). What does it take to count as a card-carrying common-sense philosopher? Does it mean that one considers common-sense propositions as absolutely unrevisable? This would be a form of “indefeasible conservatism” which very few philosophers after Quine would consider as a proof of methodological “modesty” (Daly 2010, pp. 20–21)! Does “common-sense philosophy” mean that one agrees with the main elements of G.E. Moore’s response to skepticism, in particular his most famous “proof of an external world” (Moore 1939)? If that is what it means, then the contemporary consensus (if any) is rather against the neo-Mooreans, as Crispin Wright emphasizes:

Familiarly, there has been general agreement both that the [Moorean] Proof is unsuccessful […] and that one who offers it as a response to material world scepticism is somehow naïvely missing the point, or underestimating the severity of the challenge that the sceptical arguments present. (Wright 2007, p. 26)³

The difficulty to evaluate the legacy of common-sense philosophy is due in large part to the existence within the “Tradition” of many different conceptions of “common sense” and many different lines of argument in defence of it. This is true of course when one moves from one author to another: neo-Reidians tend to emphasize the principle of reliability of our natural faculties,⁴ where neo-Mooreans are usually more interested in the comparative certainty of different intuitions.⁵ But it is true also within the works or followers of one and the same author: for instance, it seems that an author’s evaluation of Moore’s legacy (and their willingness to self-identify as neo- or post-Moorean) depends a lot on whether the focus is placed on Moore’s “offensive” strategy in “Proof of an External World” (Moore 1939), or rather on his “defensive” strategy in “Hume’s Theory Examined” (Moore 1953) or in “Four Forms of Scepticism” (Moore 1959).⁶

Though I am willing to identify myself as a “common-sense philosopher”, I think it should be acknowledged that this label by itself is not clear enough to refer to any identifiable claim: the “Common-Sense Tradition” just isn’t sufficiently unified around one shared thesis. Because of this, I believe that the label “common-sense philosopher” would be better used with a qualification that would indicate which kind of common-sense strategy one endorses: we might perhaps distinguish “particularist common-sense philosophers” (Lemos 2004; Bergmann 2021) from “greater certainty common-sense philosophers” (Lycan 2001, 2019), “dogmatist common-sense philosophers” (Pryor 2000), or “reliable faculties common-sense philosophers” (Peels 2020). On account of the argument I’ll be defending in this paper, I would self-identify as a “dynamic common-sense philosopher” or “starting-point common-sense philosopher”.

The purpose of the present paper is to (i) distinguish the different lines of arguments present in the common-sense tradition, in order to (ii) isolate and defend one of them—the Dynamic Strategy—which I take to be the strong core of the tradition.⁷ In Sect. 1, I will present the various lines of arguments present in the common-sense tradition as different versions of one core intuition, i.e. the reaction to “radical revisionism”. In Sect. 2, I will focus on one of these lines of arguments, namely the Dynamic Strategy, and I will develop this strategy with the help of Gilbert Harman’s epistemology of belief revisions. The advantage of this section will be to isolate the proper argumentative force of this strategy alone, but it will also make clear that this strategy alone offers a relatively weak safeguard against radical revisionism. In order to offer a more secure protection against radical revisionism, it seems necessary to complement the Dynamic Strategy with Moore’s “greater certainty argument”. This is what I will do in Sect. 3. The result will be a common-sense strategy that combines two traditional lines of arguments: the Dynamic Argument and the greater certainty argument.

1 The Varieties of Common-Sense Strategies

Despite the variety of lines of arguments defended by various common-sense philosophers (and sometimes by one and the same author in different places), it seems possible to gather these different lines of arguments as different versions of one and the same scheme or core intuition. The common feature of different common-sense philosophers, I would suggest, is their initial reaction when reading (or hearing) some philosophical arguments that seem to establish extremely striking claims, such as the conclusion that nothing is in movement (Zeno), that there is only one substance (Spinoza), that time is an illusion (McTaggart) or that we have no knowledge of material objects (Hume).⁸ All these conclusions have in common the fact that they try to overturn some belief B that we initially had. Confronted with such striking philosophical conclusions, a “common-sense philosopher” is one who reacts with the following core intuition:

Core Intuition (CI): This argument, by philosopher X, is irrational. There is no way a belief like B could be so easily overturned. The argument is irrational not because of some specific defect, but for the general reason that the belief B belongs to a class of beliefs C (which we will call “common-sense beliefs”) that has a certain epistemic privilege P that allows them to evade the threats of such revisions.

This, I suggest, is the common core of the common-sense Tradition, but it should be noted that it contains only a proposition-scheme and that the plausibility of the Core Intuition will depend a lot on the value we give to the variables C and P. In other words, in order to make a real philosophical proposal, a common-sense philosopher has to respond to the following three questions:

The Classification Question (CQ): Which class of beliefs C should be counted as “common-sense beliefs”?

The Privilege Question (PQ): What is the epistemic privilege P of common-sense beliefs?

The Justification Question (JQ): Why is it that beliefs of class C have privilege P?

And it seems fair to acknowledge that different authors of the common-sense tradition (sometimes the same author in different places) have offered very different responses to these questions.

Before I present the various possible responses to CQ, PQ and JQ, I would like to make one important comment about the target of common-sense strategies in general. It is often supposed that the target of common-sense philosophy is skepticism. I think this is ambiguous and misleading, and that we should rather understand common-sense philosophy as an attempt to respond to (or fight against) radical revisionism. Here is what I mean.

The word “skepticism” is used in two importantly different senses. In the most classical, philosophical sense, “skepticism” refers to the philosophical view (or the family of views) according to which our epistemic situation is pretty poor. Views of this family include, for example, the view that we can’t have knowledge; or the view that we could have knowledge, but as a matter of fact we don’t; or the view that we lack justification (if not knowledge), or at least the standing to claim that any of our beliefs is justified (or known). What is common to all these views is the idea that our beliefs cannot reach a certain important epistemological standard (be it knowledge, or certainty, or justification, or entitlement, etc.). I will call this family of views epistemological pessimism.⁹ This family of views belongs to epistemology: it is a family of views about knowledge, justification or other epistemological notions. If this were the only use of the word “skepticism”, it would be very surprising to learn that Spinoza is sometimes described as a “Free Will skeptic”. For one thing, few philosophers are as optimistic as Spinoza about our capacity to reach (more geometrico) knowledge and certainty. Furthermore, his views about Free Will are views about the proper metaphysics of action and causality, not views about epistemology. So what is the meaning of “skepticism” in such usage? The idea is that Spinoza is a “skeptic” about Free Will, not because he denies knowledge (or justification) of anything (indeed he claims to know with certainty that Free Will is an illusion) but because his view goes strongly against what human beings ordinarily believe. David Hume uses the work “skepticism” in this same sense when he says that Berkeley, in spite of his claims to be trying to refute skepticism, is in fact a skeptic malgré lui.

Most of the writings of [Dr. Berkeley] form the best lessons of scepticism, which are to be found either among the ancient or modern philosophers […]. He professes, however, in his title-page (and undoubtedly with great truth) to have composed his book against the sceptics […]. But that all his arguments, though otherwise intended, are, in reality, merely sceptical, appears from this, that they admit of no answer and produce no conviction. Their only effect is to cause that momentary amazement and irresolution and confusion, which is the result of scepticism. (Hume 1894, sec. xii, part 2, p.155 fn, emphasis mine)

Berkeley, like Spinoza, is no skeptic in the sense of being an epistemological pessimist. But his view about material objects flies in the face of common sense, just like Spinoza’s claim that Free Will is an illusion is strikingly unbelievable. Both philosophical claims ask us to revise some views we all ordinarily have, and the revision that they ask us to do is strikingly radical. This other sense of “skepticism” I will call radical revisionism.

It should be observed that epistemological pessimism (about knowledge) is an instance of radical revisionism: since we ordinarily think that we know many things (our own name, what we have eaten at breakfast, etc.), saying that we don’t know anything is strikingly revisionary.¹⁰ But the category of radical revisionism is much larger than just epistemological pessimism—and independent from it, since (as is the case with Spinoza or Berkeley) one can be a radical revisionist about some domains of philosophy without being an epistemological pessimist.

The Core Intuition of common-sense philosophers is a reaction to radical revisionism generally, and not specifically to epistemological pessimism. But the ambiguity of the word “skepticism”, together with historical accidents (in particular the fact that the revisionist to whom the common-sense tradition first replied, namely David Hume, was also specifically an epistemological pessimist) have led to the result that most discussions about common sense are centered on epistemological pessimism and arguments for that specific view (especially Hume’s argument against knowledge of external objects or Russell’s arguments against certainty). I think this focus generates obscurities and confusions (“level confusions”) and that common-sense philosophers would do better to focus on forms of radical revisionism that are not epistemologically pessimist (such as Zeno’s argument against the reality of movement, or McTaggart’s argument against the reality of time). Here is why: even when we focus on the discussion of revisionisms whose content is not epistemological (views about movement or time, not about knowledge or justification), there is great chance that part of the discussion between the common-sense philosopher and the radical revisionist will hang on the epistemological status of our (pre-theoretical) beliefs or intuitions (our beliefs or intuitions about movement or time, for instance, not about knowledge or justification). Now, if instead of beliefs or intuitions about movement or time we focus on beliefs or intuitions about knowledge itself or justification, then epistemological notions will appear at two different levels of the inquiry and this is very likely to generate level confusions or level ambiguities, that can be easily avoided by changing the examples.¹¹ This is why, in the remainder of this paper, I will focus mainly on examples of revisionist arguments that are not arguments for epistemological pessimism (or “skepticism” in the narrow sense).

Let me now turn to the various possible versions of the Core Intuition, i.e. the various possible responses to CQ, PQ and JQ. Possible responses to CQ include¹²:

Common-sense beliefs are

C1. beliefs that have a maximal or very high degree of subjective certainty or confidence.

C2. basic beliefs that constitute the very foundations of our architecture of knowledge.

C3. beliefs that we have before we start philosophical inquiry.

C4. beliefs that (almost) all human beings in (almost) all societies at (almost) all times have believed.

C5. beliefs that are produced of necessity in human beings in virtue of the constitution of their nature.

C6. beliefs B that are accompanied with a second-order belief or a second-order seeming that “B constitutes knowledge”.

C7. beliefs whose denials are pragmatically incoherent.

And of course, various conjunctions or disjunctions of classes C1-5 can also count as possible responses to CQ.

Possible responses to PQ include:

Common-sense beliefs (i.e. beliefs in class C) are

P1. absolutely indefeasible.

P2. not revisable by philosophical argument (but possibly revisable by scientific evidence).

P3. prima facie justified (or justified by default) but revisable, even by philosophical argument.

P4. always entitled even if not justified.

P5. more justified than any other beliefs in our belief system.

Different combinations Cn-Pn make as many possible claims a common-sense philosopher might want to make. Obviously, some of these combinations are so strong as to be absolutely indefensible: how could anyone plausibly defend that the class of all beliefs we have pre-theoretically (folk theories like folk physics, folk biology, folk theory of mind, etc.) are absolutely indefeasible, even by science [C3-P1]? Perhaps some critics of the common-sense tradition have motivated their rejection of common-sense philosophy in general by focusing on such indefensible combinations. And it is not uncommon for common-sense philosophers to complain that their critics have thereby criticized a straw man, because “of course” no one in the Tradition has ever claimed that folk prejudices were indefeasible. But even though the critic might be making a particularly uncharitable reading of C and P, I think it should be acknowledged that the “tradition” does not give a clear and univocal response as to how C and P should be read. Sometimes, common-sense philosophers try to defuse the straw man by strengthening C—saying for example something like this: “you have misunderstood the point; the beliefs we claim are indefeasible are those that play a foundational role at the very basis of our knowledge architecture” [C2-P1]. Other times, they’ll defuse the straw man by weakening the epistemic privilege—saying for example something like this: “you have misunderstood the point: we are not claiming that our pre-theoretical beliefs are indefeasible, just that they are innocent until proven guilty” [C3-P3]. And of course, the common-sense philosopher can use both strategies at the same time (both strengthening C and weakening P). But what is clear is that she is committed to providing a more explicit answer to CQ and PQ, otherwise she is just leaving herself wide open to an uncharitable reading.

Once the common-sense philosopher has made clearer which claim she wants to support (by responding more explicitly to CQ and PQ), her task is not over: she still has to provide an argument or a justification, in order to explain why beliefs of class C have the epistemic privilege P.¹³ In response to this Justification Question (JQ), the tradition contains at least the following lines of arguments (which are clearly distinct, though sometimes combined by one and the same author)¹⁴:

J1. The “foundations” argument (or synchronic argument): basic foundational beliefs (C2) are unrevisable by philosophy (P2, or even P1) because philosophy (or even science) rests upon these foundations and it would be self-undermining for philosophy (or even science) to undermine them.

J2: The “starting point” argument (or Dynamic Argument): beliefs we have from the start (C3) have a prima facie or default justification (P3), because we don’t need to be convinced into believing them: they’re already there.

J3: The “greater certainty” argument: beliefs that have a maximal or very high degree of subjective certainty or confidence (C1) cannot be undermined by philosophical argument (P2), because any philosophical argument designed to undermine them will have philosophical premises less certain than them.

J4: The “consensus gentium” argument: beliefs that are widely believed across time and across all societies (C4) have a prima facie or default justification (P3), because it is unlikely that so many people should be deceived.

J5: The “natural unavoidability” argument: we are entitled (P4) to retain the beliefs that are produced of necessity in virtue of the constitution of our nature (C5), because it is impossible for us to believe otherwise, so there is no point in trying to believe otherwise.

J6: The “natural pragmatic” argument: we are entitled (P4) to retain the beliefs that are produced of necessity in virtue of the constitution of our nature (C5), because we are not likely to win anything good in a contest against Nature herself.^{15, 16}

J7: The “meta-coherent” argument: we should take as data in the philosophical enquiry (P3, or perhaps even P2) the beliefs that ordinarily seem to us to constitute knowledge (C6) because we are justified by these second-order seemings to consider them to constitute knowledge (phenomenal conservatism), and what we know constitutes our evidence-base (E=K).

There is an important objection against my construal of the core project of common-sense philosophy, as a defence that some specific class of beliefs Cn has an epistemic privilege Pn that protects it against revisionism for reasons Jn—we might call this the “CPJ project”. Some recent (and prominent) defenders of the common-sense tradition seem to reject the CPJ project itself as an interpretation of common-sense philosophy:

Though the common sense tradition does hold that some common sense propositions are epistemically justified for us, it is not committed to the view that they are epistemically justified for us in virtue of their being common sense propositions or in virtue of their being deeply and widely held. It is not committed to the view that being widely and deeply believed confers, or is a source of, any positive epistemic status upon a proposition. […] They are not ‘methodists’ who begin with a criterion such as ‘Whatever is a common sense proposition is true or epistemically justified.’ (Lemos 2004, p. 6)

I emphasize that Moore makes no argument from any proposition's being commonsensical to that proposition's having any positive epistemic status. (Lycan 2001, pp. 47–48, 2019, p. 31)

I can see three interpretations of this rejection of the CPJ project, none of which seems to me to successfully undermine the idea that the CPJ scheme is the best way to classify the various common-sense strategies.

The first interpretation (of Lemos’ quote in particular) is that Lemos is here rejecting a certain implementation of CPJ, namely a strong and indefensible one according to which the category of “deeply and widely held” beliefs ($C1 \cap C4$) would have the strong status of being “true or epistemically justified”, i.e. justified full stop, justified ultima facie, even when confronted with the skeptic’s challenge—in other words justified in a way that no philosophical argument can possibly defeat (P2). Now, if that is Lemos’s claim, then, of course, it is completely compatible with more modest versions of the CPJ project.

The second interpretation might be that, according to Lemos—and perhaps also (Kelly 2005), the class of beliefs that have a special status are not common-sense beliefs but rather another class, which we should rather call “Moorean facts”: some common-sense beliefs (i.e. beliefs in $C1 \cap C4$) are also Moorean facts, and therefore the defence of Moorean facts in general will allow the common-sense philosopher to defend some common-sense beliefs against radical revisions. But these will be defended not in virtue of being common-sense beliefs but in virtue of being Moorean facts. If that is Lemos’ point, then it seems clear to me that it is still a version of the CPJ project, with the only terminological difference that the significant class C is called “Moorean facts” instead of “common-sense beliefs”—and I have no problem abandoning the terminology of “common-sense beliefs”.

The third interpretation (probably the most accurate one) is that Lemos and Lycan are really attacking the CPJ project itself, because the CPJ project is a form of “methodism”, and Chisholm (1982) has taught us that the common-sense response against skepticism was essentially “particularist”. This is probably the main reason why philosophers whom I would venture to call “neo-Chisholmian” are so reluctant about the CPJ project, while neo-Reidians like van Woudenberg (2020) or Peels (2020) endorse it without any qualms. But it seems to me highly problematic to consider as the common core of the common-sense tradition the distinction between methodism and particularism. The reason is that this distinction is an epistemological one, which is designed to respond to the specific problem of epistemological pessimism.¹⁷ But the common-sense tradition, as we have already seen, is trying to respond to radical revisionism more generally.¹⁸ And it is far from clear how particularism could help us responding to revisionary arguments about movement, time or free will.¹⁹ The focus on Chisholmian or neo-Chisholmian particularism is a result of the (accidental) focus on epistemological revisionism (skepticism) as opposed to revisionism more generally.²⁰ Once we try to characterize a strategy against radical revisionism in general, it is hard to see how we could possibly do so without identifying a class of beliefs C to which the strategy applies and a reason J why this class avoids radical revisionism. And it is hard to see how this class C could avoid radical revisionism without having some common epistemic privilege P, that is an epistemological feature that other beliefs don’t have (beliefs that are easily defeasible).

What I have tried to show in this section is that, even though the common-sense tradition shares a core intuition, namely the CPJ project, there is not enough clarity and unity in the tradition as a whole to identify the label “common-sense philosophy” with a specific argument—there are, rather, several different strands of arguments (at least J1 to J6) that could be called “common-sense arguments”. I will call this the ambiguity problem. And it seems clear that, given this problem, any contemporary philosopher who self-identifies as a common-sense philosopher has the task of clarifying her precise position. How might she do so? I can see three different approaches to the ambiguity problem.

The disjunctive/ecumenical approach consists in saying that we can consider as common-sense beliefs beliefs in any of the classes C1–C6, and that justifications J1–J6 all count as legitimate common-sense strategies. According to this strategy, Reidian defences or Moorean defences of common-sense beliefs, and perhaps also Wittgensteinian defences of hinge propositions, are all compatible, mutually reinforcing and collectively forming a wide consensus in favour of “common sense”. The advantage of this strategy is precisely to give the impression of a wide consensus, but I think it is clear that this “consensus” is in danger of being merely verbal, which might be sufficient for the purposes of advertising but not for philosophical purposes.²¹

The conjunctive approach consists in providing a maximally conjunctive definition of common sense (in response to the Classification Question): common-sense beliefs wouldn’t be members of class C1 or C2 or C3, but rather they would be “found where [such] groups of propositions overlap” (Van Woudenberg 2020, p. 183). For instance, according to van Woudenberg, common-sense beliefs should be characterized as beliefs that are not only “not based on scientific or scholarly research” (C3), but also “widely believed” (C4), “not believed on the basis of inference” (C2) and such that their “denials are pragmatically incoherent (C7). His class of common-sense beliefs, therefore, is the conjunction $C2 \cap C3 \cap C4 \cap C7$.²² The advantage of this strategy is that it seems to make the Justification task easier: for if you cannot justify protecting “common-sense beliefs” due to their inclusion in C2, you still have the possibility to justify it by their inclusion in C3 or C4. Or if none of these classes by itself justifies any particular protection, perhaps their conjunction will generate a positive interaction that will protect such beliefs against revision. The conjunctive approach seems to maximize our chances to find a successful response to JQ. The drawback is double. For one thing, the class that we are thereby constructing has very few chances to count as a natural kind, or even a relevant kind—it looks like a hodgepodge of independent conditions. And, more importantly, even if we succeed in defending this conjunctive class against revisions, there is a risk that we will never grasp distinctly the proper philosophical contribution of each part of the definition. In other words, there is a great risk of losing in clarity when we turn to the Justification task. This is why I favor personally the third approach.

The third approach is what I would call the isolation approach: I think it is important methodologically to isolate the different strategies of common sense in order to evaluate precisely and clearly the proper epistemological status of each class Cn separately. As a matter of fact, the specific strategy I will try to defend in this paper is the starting-point argument or Dynamic Argument J2, which relies only on C3 (and P3). I do not rule out the possibility that such an isolated strategy will be found insufficient, and that we will find it necessary to add elements from other classes and other strategies—actually, this is indeed what I will do eventually, complementing the Dynamic Argument with Moore’s greater certainty argument in Sect. 3—but this will be done only once the philosophical necessity of doing so has been clearly identified.

2 The Dynamic Strategy of Common Sense

For two earlier and much shorter presentations of the Dynamic strategy of Common Sense, see Guillon (2020a, sec. 2, 2020b, sec. 2).

The best way to understand the Dynamic Argument of common sense (J2) is to contrast it with the foundations argument (or synchronic argument, J1). Both lines of argument are present in the seminal work of Thomas Reid.

Here is a typical formulation of the foundations argument:

Such principles [of Common Sense] are older, and of more authority, than Philosophy: she rests upon them as her basis, not they upon her. If she could overturn them, she must be buried in their ruins. (Reid 1997, pt. I sec. 5)

The idea here is that our system of knowledge forms a (foundationalist) architecture of which the principles (or beliefs) of common sense constitute the foundations. In contemporary vocabulary, common-sense beliefs are here characterized as (properly) basic beliefs (C2). And all other beliefs, including philosophical beliefs, are “based” on these foundations through a series of inferences. Because of this special foundational position, common-sense beliefs cannot be overturned by philosophy, since philosophy would thereby undermine itself. We could represent this strategy with the following typically pyramidal schema:

Here is now a typical formulation of the starting-point argument (or Dynamic Argument), as applied to the question of Free Will:

This natural conviction of our acting freely […] ought to throw the whole burden of proof upon that side; for, by this, the side of liberty has what lawyers call a jus quaesitum, or a right of ancient possession, which ought to stand good till it be overturned. If it cannot be proved that we always act from necessity, there is no need of arguments on the other side to convince us that we are free agents. (Reid 2011, pt. IV sec. vi)

The belief in Free Will, according to this quote, has a special epistemic privilege due to its being “common sense”. But what does it mean for it to be “common sense”? Here the argument has nothing to do with the structure of our system of belief. For all we know, it might be that this belief is an inferential belief. The only important characteristic is that it’s a belief that’s already there before any philosophical inquiry. The notion of common sense we have here is not the synchronic notion of the foundations of our system of knowledge (at a given time t); it is rather the diachronic notion of a system of beliefs that precedes any revisions that would result from the philosophical (or scientific) inquiry. And this chronological position gives common sense (according to the Dynamic Argument) a privilege of prima facie justification (or justification by default), because anyone who challenges this starting point has to justify a change of beliefs, while someone who maintains this starting point isn’t advocating any particular change. We could represent this strategy with the following schema:

These two strategies of common sense are radically different. They start from completely different notions of what counts as common sense—the synchronic notion of the foundations (at any given time) vs the diachronic notion of the starting point. And they arrive at completely different epistemological privileges—foundations are indefeasible, on pains of self-undermining, while the starting point is, at least in principle, revisable.²⁴

Another difference that will be crucial is the notion of “justification” that is at stake in both arguments. In the foundationalist argument, what is important is the justification of individual beliefs (belief-states) by other beliefs (or perhaps by “experiences” or “seemings”) within a synchronic architecture. In the Dynamic Argument on the other hand, what is important is the justification of revisions of beliefs, or belief-changes instead of belief-states: what the dynamic common-sense philosopher wants to show, utlimately, is that radical revisions (radical changes of beliefs) are not justified. Even though there are obviously connections between the justification of belief-states (within a synchronic architecture) and the justification of belief-changes (in a diachronic process of revisions), it is clear that these two notions of justification are distinct, as they apply to distinct ontological categories. In order to avoid any confusion, I will call the synchronic kind of justification support and the diachronic kind of justification motivation: according to this terminology then, we will say that a certain belief (a belief-state) is supported or unsupported,²⁵ while a revision (a belief-change) is motivated or unmotivated.

With the distinction between support and motivation, we can see now that if we want to assess (or even to formulate carefully) the Dynamic Argument of common sense, we need to study the conditions under which a certain revision of beliefs is motivated or unmotivated. And even though most of the literature in the epistemology of justification has traditionally been dedicated to the conditions of support of belief-states, there are nonetheless some contemporary epistemologists—most importantly Isaac Levi (1980) and Gilbert Harman (1986)—who have studied specifically the motivation of belief-changes within a process of revisions or, in Harman’s terminology, of “changes in view” within a process of “reasoning”. These works constitute a field of inquiry which I will call “Dynamic Epistemology”. This focus on the dynamics of belief revisions has historical roots in classical pragmatism, especially in C.S. Peirce. Indeed, Peirce’s argument in favour of “Critical Common-sensism” is the closest historical version of what I call in this section the Dynamic Strategy of common sense.²⁶

What have we learnt from the work of contemporary Dynamic Epistemologists? There are of course many points of disagreements between different authors in the field, but there seems to be a consensus between Dynamic Epistemologists on a basic principle which I will call the Principle of Motivation (PM). Here is Levi’s formulation of PM:

(PM) “X should not modify his body of knowledge unless in doing so he improves it. Hence, even though X need not justify having h in his corpus of knowledge once he has accepted it, prior to doing so he will be under some obligation to justify adding h to his body of knowledge.” (Levi 1980, pp. 1–2)

And I will reformulate it as follows:

(PM) Every revision, i.e. every move from an epistemic system Sn to a later system Sn+1, must have a positive motivation.²⁷

The Principle of Motivation is what Joanna Lawson (in the context of a defence of common sense) has also called a principle of “doxastic inertia” (Lawson 2021, p. 194). It states that what is in need of a positive motivation (justification) is epistemic revision, not epistemic standstill. In the absence of a positive motivation to change your system of beliefs, not changing it is always “motivated” (justified, appropriate) by default. It has also been called a principle of “conservatism”, including by Gilbert Harman himself, for instance in the following quote which contains another clear formulation of PM:

In reasoning, you start where you are with your current beliefs, plans, and goals, and your current methods or procedures for modifying these plans and methods. It is not reasonable for you to make any changes in your starting points except to resolve tensions within them and to answer questions which you have reasons to answer. Any reasonable methodology has to be conservative in this sense. In a certain (methodological) sense, your initial beliefs, plans, goals, and methods have an immediate default or prima facie ‘justification’. They are the ‘foundations’ for your reasoning – foundations in the sense of starting points. (Harman 2010, p. 153)

Even though neither Levi nor Harman are common-sense philosophers, it should now be obvious how their Dynamic Epistemology provides a basis for a successful Dynamic Argument of common sense. For suppose we decide to use the phrase “common sense” to designate (by terminological stipulation) “our initial beliefs, plans, goals, and methods”, it immediately follows that, by Levi’s and Harman’s principle, common sense has “an immediate default or prima facie justification”, i.e. that we “need not justify having [common-sense beliefs] in our corpus of knowledge” in the first place, because they’re already there, while we need to justify revising them. This is exactly Reid’s point in the quote about the “right of ancient possession” of the belief in Free Will. And contemporary epistemology allows us to see that this Reidian argument relies on an extremely modest principle of Dynamic Epistemology.

We can now formulate more rigorously the Dynamic Argument:

(C3) Common-sense beliefs (for X) are beliefs that X has before he starts philosophical inquiry (beliefs in the initial System of beliefs S0).

(1) Any system of belief Sn of X is a successor of S0 (Common Sense) via a finite series of revisions R1, R2 ... Rn.

(PM) Every revision Rn, i.e. every move from an epistemic system Sn to a later system Sn+1, must be positively motivated.

(P3) Therefore, for every common-sense belief B of X, B will be appropriately maintained by X in any system Sn successor of S0, unless there is a positive motivation to suppress it at some stage of the reasoning between S0 and Sn.

Another way of presenting the Dynamic Argument is to say that it offers a principled reason (by the principle PM itself) for shifting the burden of proof against the radical revisionist: if the revisionist doesn’t provide a convincing argument for his side, there is no need for the common-sense philosopher to offer any argument for his own common-sense belief. We have seen that Reid offers a rationale of this kind against the Free Will revisionist (a rationale to the effect that there is no need of a positive argument for Free Will). It might seem that G.E.Moore differs here, since he seems to (try to) offer against the external-world revisionist a positive “proof of an external world”. But there is at least one possible interpretation of Moore, offered by Greco, Soames and Kelly, according to which this “proof” is in fact “tongue in cheek” (Greco 2002, p. 547) or “ironic” and only designed “to show that there is no need for such a proof in the first place.” (Soames 2003, p. 23) And the reason why there is no need for such a proof is that “[Moore’s] position regarding the propositions of common sense is that they constitute the starting point for philosophy.” (Soames 2003, p. 5). This is what (Kelly 2005, p. 182) has called “playing defense” against the skeptic, as opposed to “playing offense”. If this interpretation is correct (a historical question I will not examine here), then what I call the Dynamic Argument of common sense is also present in the works of Moore.

I would like to convey the idea that this particular common-sense strategy is extremely modest and could easily be universally consensual. After all, its unique substantial assumption is the Principle of Motivation, which is probably the most modest and consensual principle for anyone trying to spell out a theory of belief revisions (Dynamic Epistemology). But the best way to show how modest this strategy is is perhaps to respond to a couple of objections.

• (1) The unsupported objection: “You are saying that all beliefs in our initial system are justified by default, but this is a crazily optimistic assumption from a foundationalist point of view. After all, it is completely possible and even highly plausible that at least some of our initial beliefs are (in the initial system) without any justification at all (neither inferentially justified nor properly basic).”

  Response: This objection relies on the confusion between motivation (dynamic justification of belief-changes) and support (synchronic justification of belief-states). What the dynamic strategy says is not that all belief-states in the initial system are supported. It only says that the belief-change (the vacuous belief-change) of maintaining them is motivated (unless there is a positive motivation to suppress them). But this is compatible with saying that some (or perhaps all) of them are unsupported; and if the agent becomes aware at some point that some common-sense belief is unsupported, this will very plausibly count as a positive motivation to abandon it. Nothing I have said rules out the possibility of such a kind of motivated revision. Therefore a radical revisionist (or a skeptic) could try to overturn some common-sense belief by proving that this belief is unsupported: what the dynamic strategy says is only that the revisionist has the burden of proving that much; if his proof fails, and if it remains as much as possible that the common-sense belief is supported, then the agent will remain motivated in maintaining it.

• (2) The anti-conservatism objection: “The principle of motivation is also, in Harman’s own words, a principle of epistemic conservatism. It is well-known that conservatism would serve well the interests of a defence of common sense, but far from being a consensual principle, epistemic or doxastic conservatism—the view that X’s merely believing that p gives X some justification for p—is an extremely unpopular view, largely seen as a form of (viciously) circular reasoning.” (see Daly 2010, pp. 20–21).

  Response: This objection relies on a confusion between very different views that have all received, in different contexts, the label “epistemic conservatism”. Vahid (2004) has done an excellent job distinguishing some of these. To use a classification that is partly dependent on his work, we can distinguish: doxastic conservatism,²⁸phenomenal conservatism and perseverance conservatism. The view that is mentioned in the objection and is very unpopular (rightly so in my personal opinion) is doxastic conservatism, i.e. the view according to which the fact that X believes that p suffices to make it so that X’s belief-state that p has (at least some) support. The risk of vicious circularity is quite obvious in this view. Phenomenal conservatism, defended for instance in (Huemer 2007) is the less implausible view according to which X’s belief-state that p receives support not by X’s believing that p, but by its seeming to X that p. This latter strategy avoids the charge of circularity, but it has in common with doxastic conservatism the fact that it deals with the synchronic support of belief-states. In contrast, perseverance conservatism (the view defended by Harman and that is at the core of the Dynamic Argument) is only concerned with belief-changes. As we have seen in response to objection 1, this kind of “conservatism” is completely compatible with our common-sense beliefs being synchronically unsupported (contra doxastic and phenomenal conservatism). And this kind of conservatism is an extremely modest and consensual one.²⁹

• (3) The terrible starting-point objection: “Even if your ‘conservatism’ is only dynamic, it still gives a privilege to the starting-point, and this seems to presuppose that our starting-point is at least decently satisfactory (even if revisable). But it might be that our starting-point is terrible, and in that case, we would be better off taking a new start altogether rather than being conservatives.”

  Response: This objection wrongly assumes that the Dynamic Argument (or the endorsement of perseverance conservatism) presupposes a positive evaluation of the starting-point. The argument has no such presupposition. The only principle on which the argument relies is the Principle of Motivation, and this principle of Dynamic Epistemology is accepted even by philosophers who have a negative opinion about our starting-point. This is explicitly the case of Isaac Levi: Levi’s motivation for studying belief-changes (inquiry in his terminology) is precisely the fact that the original “pedigree” of our beliefs is pretty dark and that our best hope of arriving at knowledge is to have a good methodology for making progress in our revisions.

  Where all origins are dark, preoccupation with pedigree is self-defeating. We ought to look forward rather than backward and avoid fixation on origins. Epistemologists should heed similar advice. Whatever its origins, human knowledge is subject to change. [...] Epistemologists ought to care for the improvement of knowledge rather than its pedigree. (Levi 1980, p. 1)

  The reason why Levi accepts the Principle of Motivation (or inertia, or conservatism—but it should now become clearer that these terms convey misleading connotations) is not at all because he thinks that our starting-point is decently good, but because he wants to make sure that by changing it at least we improve our chances to gain knowledge. This (extremely modest) kind of “conservatism”, far from being opposed to the idea of revision and progress, is, according to Levi, the very condition of progress itself.³⁰

  To be sure, most and perhaps all authors of the common-sense tradition will add here that what they consider as common-sense beliefs do have a positive epistemic status or pedigree, because their class Cn is not just the class of all beliefs in S0 (C3), but rather a subset of S0—perhaps ($C1 \cap C3$) for instance. I am in fact sympathetic to this kind of combined strategy, as will become clear in Sect. 3. But my point here is only this: such a combined strategy is not necessary in order to offer our common-sense beliefs a default motivation to maintain them (P3). This default motivation is guaranteed merely by their being members of S0 (assuming the Principle of Motivation). Narrowing the set of common-sense beliefs (and focusing on some beliefs that do have a positive epistemic status) will be necessary only if we are trying to secure a stronger epistemic privilege—for instance (P2) instead of merely (P3). This is in fact what we will do in Sect. 3. But whether or not the later strategy of Sect. 3 is successful, my argument so far is that the Pure Dynamic Strategy alone is sufficient to establish (P3).

• (4) The different starting-points objection: “Your strategy of common sense presupposes that the epistemological starting point S0 (the system of beliefs and principles from which ‘we’ start) is something ‘we’ have in common. But this is completely false: different people, especially people of different cultures and different historical periods, have very different epistemological starting points. Therefore, your Dynamic Argument doesn’t defend anything that would be ‘common’ in common sense.”

  Response: In fact, I completely accept the point of this “objection”, and it allows me to clarify what I am up to with this Dynamic Argument—and what I am not trying to do. What I am trying to do is to isolate one line of argument present in the common-sense tradition and to evaluate it on its own (isolated) merits. I believe that the starting-point argument, in itself, has a lot of dialectical force (because it relies on a very modest assumption (PM), but it is absolutely true that in itself it only gives us a defence of the agent X’s starting point whether or not this starting point is common with some or all of the rest of humanity. So the phrase “common sense”, as I use it here, should not be understood as implying that it is common to some group of people, let alone the whole of humanity. According to the use of the phrase I am making here, there might be different “common senses” for different societies, and even different “common senses” (different starting points) for different individuals of the same society. If that is the case, then this Dynamic Strategy will give “epistemic privilege” or protection to different beliefs for different individuals. It might be argued that this diminished ambition is too weak to be a plausible interpretation of the “common-sense tradition”, for this tradition does try to capture and defend what is common to all humankind. My response is two-fold. First, even though the universality of common sense is probably playing an important argumentative role in Thomas Reid’s works, I am not sure at all that this is true for G.E. Moore’s defence of common sense, at least in Lycan’s interpretation which we will see in the next section.³¹ Second, and more importantly, even if a common-sense philosopher wants to try and defend something common to all humankind (a project with which I am sympathetic), she should be careful to distinguish the arguments and principles by which she can do so. She should be clear, in particular, about the fact that the Dynamic Argument, by itself, doesn’t bring with it any dimension of commonality or universality. If she wants to get this dimension, she will need to complement her argumentative strategy with further definitions or premises. In this paper, I will not try to go in this direction at all. Even after the Moorean complements of Sect. 3, the strategy of “common sense” here defended will be completely compatible with the idea that different individuals have completely different “common senses” (and correlatively, revisions that are “radical” for an individual X may not be “radical” for another individual Y).

With these clarifications in mind, we could reformulate the conclusion of the Dynamic Argument as follows:

(D) Any epistemic agent, whatever his starting point may be (good or bad, common or idiosyncratic), has no better option than to make progress from where he starts, ensuring that he really makes progress at every stage by requiring a positive motivation for the changes he makes.

This precise claim, I think, should strike anyone as extremely modest, and it is probably the strong core of the quasi-universal “post-Moorean consensus” mentioned by Soames, Armstrong, Fine or Kelly (see our Introduction). It is so modest and consensual that we find essentially the same line of reasoning in the works of an author like David Lewis, whose defence of concrete possible worlds would hardly classify him as a “common-sense philosopher”. Indeed, it is precisely in the discussion of the “incredulous stare” provoked by his modal realism that Lewis makes the following concessive points:

A worthwhile theory must be credible, and a credible theory must be conservative. It cannot gain, and it cannot deserve, credence if it disagreess with too much of what we thought before. And much of what we thought before was just common sense. […] Theoretical conservatism is the only sensible policy for theorists of limited powers, who are duly modest about what they could accomplish after a fresh start. Part of this conservatism is reluctance to accept theories that fly in the face of common sense. (Lewis 1986, sec. 2.8)

The modest methodological idea that we find here, and in most post-Moorean analytic philosophy, is that as theorists we have to start from some starting point and that furthermore our actual starting point is something that is given to us as a datum, for better or for worse, not something we may choose. We might of course choose to restart from somewhere else—that is: we can choose our restarting point, as for instance Descartes chose to restart from a set of beliefs as empty as possible (a tabula rasa, restarting from scratch). But the Motivation Principle tells us that if you restart from somewhere else, you are under an obligation to motivate your restarting point, and this motivation can only come from beliefs and principles of your actual starting point (the one you haven’t chosen).³²

One aspect that makes the Dynamic Argument particularly modest (and therefore easily consensual) is that it accepts the revisability of the starting point. Indeed, the starting point is so revisable that someone like Lewis can accept this methodology and motivate a revision as radical as the defence of concrete possible worlds! But this advantage in terms of modesty and consensus could be turned into an objection against the Dynamic Strategy itself: for if this strategy is compatible with any revision of common sense, as radical as one may want, then it doesn’t seem to help the core project of common-sense philosophy, which was to offer a protection against radical revisionism, as we saw in Sect. 1. This objection wouldn’t be against the Dynamic Argument as such (or against its conclusion), but it would be against counting it as being in any relevant sense a (successful) common-sense strategy.

I think this last objection is important and legitimate and this is precisely why I will argue in favour of a (Moorean) complement to the pure Dynamic Strategy in Sect. 3. It is true that the pure Dynamic Strategy offers little protection against radical revisionism: even though it sets the burden of proof against it, a radical revisionist can always accept the burden and offer a positive argument in favour of his radical revisionist claim.

Nonetheless, before I come to the complemented strategy in Sect. 3, I would like to emphasize the fact that this objection does not prove the pure Dynamic Strategy to be completely impotent. The reason is that, just by changing the burden of proof, the Dynamic Strategy will be sufficient to undermine some radical revisionist challenges. Suppose for instance that a revisionist about Free Will reasons as follows: “I can’t prove that Determinism is true, but you can’t offer a scientific proof either in favour of the existence of Free Will; therefore the only reasonable stance is to remain neutral about the existence of Free Will.” A Dynamic Strategy suffices to undermine this kind of argument (at least for anyone whose starting point contains the positive belief that she is free), because it sets the burden of proof against the opponent who raises doubt about Free Will: if he cannot offer a positive argument to the effect that Free Will is inexistant, or that the initial belief in Free Will is epistemically unsupported, then he doesn’t offer us a motivation to change our belief.³³ More generally, the Dynamic Strategy will be enough to undermine any revisionist strategy that presupposes that agnosticism is the default position on any philosophical question: if the Motivation Principle is true, then the default position (for X) is not agnosticism, but whatever X has in his epistemic starting point. So there is a range of radical revisionist challenges that the pure Dynamic Strategy suffices to tackle.

This is already some kind of “protection” against radical revisionism. But this is a relatively weak protection, because many historically important forms of radical revisionism are not based on such a presupposition, but rather on a positive argument in favour of the revisionist claim. These forms of radical revisionist challenge could accept to take on the burden of proof, and claim that they have positively established the motivation to revise common sense. This is for instance what David Lewis claims to have done for concrete possible worlds—in spite of any common-sensical “incredulous stare”. And this is what most (or all) philosophers of the common-sense tradition strongly doubt could ever be successfully accomplished. But if we want to claim that a radical revisionist cannot (or cannot easily) offer such a positive radical revisionist argument, we need more ressources than just the pure Dynamic Argument. As I will argue now, what we need is to complement the Dynamic Argument with the Moorean Greater Certainty Argument.

3 Complementing the Dynamic Strategy with Moore’s Greater Certainty Argument

We have said that, according to one interpretation, what Moore is really doing in response to the radical revisionist (in particular against the skeptic) is not offering a positive proof or argument; rather, his point is to show that there is no need of argument to establish common-sensical claims, and that the radical revisionist (not the common-sensist) has the burden of proof. This is, in essence, the pure Dynamic Strategy.

But Moore was trying to do more than that. He didn’t only want to show that the radical revisionist has the burden of proof and needs to offer a positive and successful motivation to revise common sense. Moore also wanted to show that the radical revisionist cannot (or cannot easily) succeed in offering such an argument. Moore was seeing the requirement of offering an anti-common-sense argument as a challenge the revisionist cannot meet (or is very unlikely to meet). Why was he so confident that the revisionist would not (likely) meet this challenge, i.e. that all revisionist arguments he would bring forward would be fallacies or failures? Because of his Greater Certainty Argument, which appears (though briefly) in several of his epistemological papers³⁴:

I think we may safely challenge any philosopher to bring forward any argument in favour either of the proposition that we do not know it [that this is a finger], or of the proposition that it is not true, which does not at some point rest upon some premiss which is beyond comparison, less certain, than the proposition which it is designed to attack. (Moore 1960, p. 228)

The attempt to prove by means of such a principle as Hume’s, that we cannot know of the existence of any material object, seems to me a characteristic instance of a sort of argument which is very common in philosophy: namely an attempt to prove that a given proposition is false, by means of a principle which is, in fact, much less certain than the proposition which is supposed to be proved false by its means. (Moore 1953, p. 143)

In this section, my purpose is to reconstruct Moore’s Greater Certainty Argument and to show that it is best understood from within the framework of Dynamic Epistemology and the Dynamic Argument of Sect. 2. Very similar reconstructions of Moore’s argument have been offered by Pollock and Cruz (1999, pp. 5–9), Lycan (2001, 2007, pp. 92–95, 2019, pp. 6–9) and Kelly (2005). For this discussion, I will use as an example of a radical revisionist argument Zeno’s Argument of the Dichotomy (ZAD), which goes as follows³⁵:

(Z1) Any movement from A to B (for instance the movement of the runner Atalanta from Athens to Babylon) is the traversing of the distance between A and B in a finite time.

(Z2) In order to traverse the distance between any X and Y, it is necessary to traverse the first half of that distance.

(ZIC1) Therefore, in order to traverse the distance between A and B, it is necessary to traverse the distances corresponding to the first 1/2 distance, but also the first 1/4, the first 1/8, the first 1/16, etc. ad infinitum, i.e. an infinite sum of finite distances.

(Z3) An infinite sum of finite distances is an infinite distance.

(Z4) It is impossible to traverse an infinite distance in a finite time.

(ZIC2) Therefore, any movement from A to B is impossible.

(ZC) Therefore, there is no movement.

This argument has four premises (Z1-4),³⁶ and tries to convince us of a conclusion (ZC) which is radically revisionist, since it goes against the common-sense belief that movement exists:

(ZM) There is movement.

If we formulate it within the framework of Dynamic Epistemology, Moore’s Greater Certainty Argument against revisionist arguments has two fundamental stages³⁷:

• stage 1: Reformulation of the revisionist argument as a conflict of beliefs.

• stage 2: The Dynamic Principle of Greater Certainty.

Stage 1, Reformulation of the revisionist argument as a conflict of beliefs. The first thing to notice, when one is confronted with a radical revisionist argument such as (ZAD), is that there is something misleading in its presentation itself—a bias of presentation which Dynamic Epistemology will immediately make clear. If we take it at face value, Zeno’s argument seems to invite us to become aware of the fact that we (initially) believe premises (Z1-4) —or at least are initially disposed to adopt these beliefs upon hearing the propositions explicitly formulated –, and then to make the process of reasoning that will lead us to produce the belief (ZC), because (Z1-4) logically imply (ZC). In other words, Zeno would have us think that our initial system of beliefs is something like S0 = {Z1, Z2, Z3, Z4}, and that because these four propositions validly imply (ZC) we should therefore make a revision by “expansion” (or addition) and come to believe S1 = {Z1, Z2, Z3, Z4, ZC}. But of course, this description of the dynamic situation is completely false, because in addition to (Z1-4) our initial epistemic system also contains the belief in the negation of (ZC), i.e. (ZM). In other words: S0 = {Z1, Z2, Z3, Z4, ZM}. Now, it is true that Zeno has managed to point out a contradiction within the different beliefs (or dispositions to believe) that we have in our initial system. And this is indeed exactly what he should do if he wants to force us to revise our starting point. But what he hasn’t done is to show how we should revise our contradictory initial system S0. For sure, we should abandon one of the beliefs that are in logical contradiction with each other. But which one should we revise? Zeno chose to start with (Z1-4) and have us revise (ZM), but another interlocutor could just as well have started with (Z1-3) plus (ZM) and have us revise (Z4) instead. As Pollock and Cruz clearly noted:

The validity of the argument does not establish which of these beliefs should be rejected, because we can convert the argument into an equally valid argument for the denial of any one of the premises. […] The argument establishes that we must reject one of these beliefs, but it does not tell us which we should reject. (Pollock and Cruz 1999, p. 7)

Or to use Lycan’s synthetic formulations: “That an argument is deductively valid is cheap” because “any argument can be turned on its head.” (Lycan 2001, p. 39, 2019, p. 7). Sophists like to take advantage of our initial contradictions and to choose arbitrarily the concessions they first ask us to make in order to lead us wherever they want; but this is no more than a sophist’s trick. Once we are aware that there is a conflict of beliefs (or of dispositions to believe) within our initial epistemic system, it’s clear that we have to revise our system in some way, but the sophist’s choice of premises does not offer any principle as to how we should revise it. Logic itself (the validity of the “argument”) doesn’t tell us which initial belief should be revised. As far as logic is concerned, we could just as easily make a “G.E. Moore shift”,³⁸ stick to our guns regarding (ZM) and abandon instead any one of Zeno’s premises. So the first observation that Moore invites us to make against the radical revisionist argument is that as far as logic is concerned the argument can just as well be turned on its head. But that is not all: if we stopped there, it would mean that the common sense reaction to the argument (the Moorean shift) is just as rational and justified as the revisionist’s conclusion, but no more rational or justified. Maybe we would be entitled to stick to common sense, but the skeptic would be no less entitled to revise it. In order to avoid this equality, we need a principle that tells us what we should do in situations of conflicts of beliefs. In other words, we need, as Kelly says, a “norm of belief revision” (Kelly 2005, sec. III) —in my vocabulary a “principle of Dynamic Epistemology”—especially designed for cases of conflicts of beliefs. And Moore’s formulations of the Greater Certainty Argument offer just that. This is the second stage of Moore’s response.³⁹

Stage 2, The Dynamic Principle of Greater Certainty (PGC). If we accept stage 1, our situation upon hearing the revisionist’s argument, for instance the (ZAD), is that we are now aware that we have a set of contradictory beliefs. A first, quite obvious, principle of dynamic epistemology, is that we cannot stay there, i.e. that this contradiction is a positive motivation to revise our system by abandoning (at least) one of the beliefs that generate the contradiction. I would call this the Dynamic Principle of Non-Contradiction. But this principle doesn’t tell us which belief in the contradictory set should be abandoned. And if we leave it at that, Zeno will insist on abandoning the belief in the existence of movement (ZM); or at least, he might maintain that this is “just as rational” as abandoning any of the other beliefs that form his premises. Moore’s response in order to block this possibility is that at least one of these premises is “much less certain than the proposition which is supposed to be proved false by its means” (Moore 1953, p. 143), and therefore Zeno’s attitude is positively irrational. In other words, according to Moore, when we are confronted with a contradictory set of beliefs (or dispositions to believe) {B1, B2, … Bn}, it is irrational to revise one belief Bi in that set if there is one other belief in the set that is strictly less certain than Bi. If we combine this principle with the Dynamic Principle of Non-Contradiction (which forces us to make some revision to the contradictory set), we get the following norm of belief revision:

(PGC) If an agent finds herself with a contradictory set of beliefs (or dispositions to believe) {B1, B2, … Bn}, she should revise her overall system of beliefs by abandoning the belief in that set that is the least certain.⁴⁰

This is the norm of (dynamic) rationality which Zeno and the radical revisionists are violating, according to Moore. And this is why radical revisionism in general is irrational. We will come back to the comparative notion of “certainty” which is at the center of this principle, because its possible ambiguity⁴¹ has been the source of objections. But before this, we need to offer a complete formulation of the new common-sense strategy based on (PGC).

The important thing to notice, in order to formulate the revised Dynamic Argument, is that it relies on a different conception of common sense. In the Moorean strategy, common-sense beliefs are not characterized only by the fact that we have them from the starting point. After all, in our example of (ZAD), Zeno’s premises (Z1-4) were also supposed to be in the starting point, and yet the Moorean does not worry about the possibility of revising them: he worries about the possibility of revising (ZM). In a certain sense, it is true that even the Moorean will be a “revisionist” when he revises (Z1), or (Z2), etc., but he will not consider himself as a radical revisionist, going against common sense. This shows that the Moorean will consider as common-sense beliefs those members of the starting point that have maximal or very high degree of certainty. His strategy is not intended to protect all our beliefs in the starting point, only those that have a maximal or very high degree of certainty. This is obviously a much smaller set of beliefs, but the epistemological protection that the Moorean will give them is stronger: the protection is not just that these beliefs have the burden of proof in their favor, but that it is impossible or very unlikely that any argument will ever force us rationally to revise them.⁴²

The full argument, which relies on this new definition of common sense ($C1 \cap C3$), plus the two principles of Dynamic Epistemology (PM and PGC) will take the following form:

($C1 \cap C3$) Common-sense beliefs (for X) are beliefs that X has before he starts philosophical inquiry (beliefs in the initial System of beliefs S0) and that have a maximal or very high degree of certainty (for X).

(1) Any system of beliefs Sn of X is a successor of S0 via a finite series of revisions R1, R2 ... Rn.

(PM) Every revision Rn, i.e. every move from an epistemic system Sn to a later system Sn+1, must be positively motivated.

(2) Therefore, for every common-sense belief B of X, B will be appropriately maintained by X in any system Sn successor of S0, unless there is a positive motivation to abandon it at some stage of the reasoning between S0 and Sn.

(3) A motivation to abandon a belief B for X would take the form of a contradiction between B and other beliefs (or dispositions to believe) that X has.

(PGC) If X finds himself with a contradictory set of beliefs (or dispositions to believe) {B1, B2, … Bn}, he should abandon the belief in that set that is the least certain (and retain the others).

(4) If some common-sense belief B (for X) happens to be a member of a contradictory set of beliefs (of X), it is very unlikely that it will be the least certain member of that set (because it has, by definition, a maximal or very high degree of certainty for X).

(P3*) Therefore, for every common-sense belief B of X, it is very unlikely that X will ever have (or have had) a motivation that makes it rational for him to revise B.

Even though this strategy is different from the pure Dynamic Argument (it concerns a smaller set of beliefs and affords them a stronger epistemological protection), nonetheless it has some important features in common with the Dynamic Argument which I would like to emphasize.

First, it is a fundamentally individual strategy: what counts as common sense ($C1 \cap C3$) for X may not be common sense for Y. And this is true not only because different people might have different starting points, but also because different people can have different degrees of certainty for one and the same proposition. So this strategy does not retain anything from the “common” in “common sense”⁴³ —even though it is the strategy of the paradigmatically “common-sense philosopher” G.E. Moore, and in my opinion the strong core of the whole common-sense tradition.⁴⁴

Second, this strategy does not completely rule out the possibility of a successful philosophical argument against some common-sense belief: it only says that this possibility is “very unlikely”. This is because nothing rules out the possibility that there exist in our starting point a set of beliefs {B1, B2, … Bn} all members of which have a maximal or very high degree of certainty. If the radical revisionist is able to show that we are indeed in such a situation, then he would force us indeed to abandon at least one common-sense belief. Whether that has ever happened is something that could be disputed, but it is at least the challenge that the radical revisionist has to meet, a challenge much harder to meet than the one posed by the pure Dynamic Argument.⁴⁵

Finally, I think that the revised Dynamic Argument, just like the pure Dynamic Argument, relies on principles so modest that they should be largely consensual, which probably explains the actual Post-Moorean consensus mentioned in the introduction.⁴⁶ The pure Dynamic Argument relied only on the Principle of Motivation, of which we have said that it was absolutely consensual among Dynamic Epistemologists. The revised argument relies furthermore on the Principle of Greater Certainty, another principle of Dynamic Epistemology which is (admittedly) less trivial but nonetheless strikes many as quite obvious: after all, what better revision could one make (when confronted with a contradictory set of beliefs) than to revise the least certain of the beliefs? But it is true that the notion of “certainty” in that principle leaves room for ambiguities that have generated objections. I will end this paper, then, with a response to these objections.

The main difficulty with the Principle of Greater Certainty has to do with its lack of conceptual precision. What is clear in Moore’s strategy is that he is inviting us to make a comparison of the various propositions that enter into a conflict, a comparison “along that dimension—whatever it is—which determines the relative vulnerability of any proposition” (Kelly 2005, p. 188). But what that dimension is varies importantly from one Moorean author to another. Moore himself always talks of greater or lesser “certainty”… but sometimes he seems to have in mind subjective certainty⁴⁷ and sometimes propositional or evidential certainty.⁴⁸ Pollock and Cruz made their interpretative decision in favor of subjective certainty and talk of “those beliefs in which we have greater confidence” (Pollock and Cruz 1999, p. 6), while David Lewis seems to have a more externalist and evidentialist notion in mind since he talks of Moorean facts as “those things that we know better” than the revisionist’s premises (Lewis 1996, p. 549). Lycan once talked of a “plausibility comparison” between propositions, and later expressed his preference for the vocabulary of “credibility” (Lycan 2019, p. 9, note 9). Obviously, some clarification is needed here if one is to defend (some version of) the Principle of Greater Certainty. The version I will defend here is subjective (or internalist) like Pollock and Cruz’s but will be designed to respond to some objections raised by Kelly against such versions.

The first thing I want to defend is the subjective or internalist character of the relevant norm of belief revision. The various versions of the Principle can be divided into internalist norms—whose conditions of application are directly accessible to the epistemic agent—and externalist norms—whose conditions of application are not directly accessible. An example of an externalist norm, in cases of conflict of beliefs, would be the Dynamic Truth Norm:

(TN) If X finds himself with a contradictory set of beliefs (or dispositions to believe) {B1, B2, … Bn}, he should abandon the belief(s) in that set that is (are) not true.

Other possible examples would be the Knowledge Norm, the Externalist Certainty Norm, etc. An example of an internalist norm would be the Dynamic Truth-Seeming Norm:

(TSN) If X finds himself with a contradictory set of beliefs (or dispositions to believe) {B1, B2, … Bn}, he should abandon the belief in that set that has the weakest seeming of being true.

Other possible examples would be the Plausibility Norm, the Credence Norm, etc.

The problem with externalist norms is that they are not directly applicable by the epistemic agent. For example, if an agent wants to apply (TN) in order to resolve a conflict of beliefs, she will have to start making an internalist evaluation about which of her beliefs in conflict has the best chances of being “not true”. In other words, in order to apply the externalist rule (TN), she will be in fact following a rule along the lines of (TSN) or some similar internalist rule. Here, it is the internalist norm, not the externalist one, that is offering a “recipe” or a “decision procedure” (to use Kelly’s (2005, p. 194) terminology). Why is it important for us to have a norm that is “directly applicable” by the epistemic agent? Because it is only on the basis of such norms that we can criticize or blame the agent (for instance the radical revisionist) for failing to apply the norm. Suppose for instance that Sarah has tried her best to apply (TN) in resolving a certain conflict of beliefs. Given the data and evidence accessible to her, she has abandoned the belief that had the highest chances of being not true. But as a matter of fact her data and evidence were extremely poor, and she has abandoned one of the true propositions in the conflicting set. Sarah, according to the externalist norm (TN), has made the wrong revision. But she is most plausibly excusable for having done the wrong revision. And the reason why she is excusable—not blameworthy or criticizable—can be explained by the fact that she has aptly followed (TSN) or some closely related internalist norm, which was her best chance to satisfy (TN). Now, the Moorean is someone who wants to criticize the radical revisionist for a certain (dynamic) fallacy—a revision that was inappropriate given the facts accessible to her. That’s why I say that the norm we are looking for in the present context is an internalist norm.⁴⁹ This rules out the Truth Norm, the Knowledge Norm (pace Lewis), or any version of the Certainty Norm interpreted along externalist lines.

If we now turn to internalist norms, the three main candidates are the Confidence Norm (defended by Pollock and Cruz), the Truth-Seeming Norm (which I take to be equivalent to the Plausibility Norm), and the Credibility Norm. I’ll start with Pollock and Cruz’s Confidence Norm, and will end with the one I think has the best chances of being the right one, namely the Credibility Norm.

As said before, Pollock and Cruz’s Confidence Norm is to be understood as equivalent to the notion of (degree of) subjective certainty. It can be formulated as follows:

(CN) If X finds himself with a contradictory set of beliefs (or dispositions to believe) {B1, B2, … Bn}, he should abandon the belief in that set which he believes with the least confidence.

Kelly (2005, pp. 191–194) makes two important objections against this kind of norm. First, there is an ambiguity as to the time at which the confidence comparison ought to be made, namely “prior to becoming aware of the conflict” or “once one becomes aware of the conflict”. This is important because it might be that the awareness of the conflict itself produces a change in our degrees of confidence or subjective certainty. This might happen, for example, if the degrees of certainty are not defined in an atomistic way (where each belief has it’s own degree of certainty independently from all other beliefs) but rather holistically (where the degree of certainty depends on the place of the belief in the overall epistemic system).⁵⁰ It seems clear that, in case the awareness of the conflict creates by itself a change in the agent’s confidence in his belief, then what is relevant is the degree of confidence once the agent is aware of the conflict. In order to take into account this possibility, we could revise (CN) into (CN*):

(CN*) If X finds himself with a contradictory set of beliefs (or dispositions to believe) {B1, B2, … Bn}, he should abandon the belief in that set which he believes with the least confidence once he becomes aware of the conflict.

But as soon as we try making this correction, we discover the second objection formulated by Kelly, which is that at the time at which the agent is aware of the conflict and deliberating as to what he should believe next, the question of which beliefs he has and with which degree of confidence is precisely what he is deliberating about. It is the result of his deliberation and therefore cannot be the data with which he deliberates.

I am not sure whether this objection is really inescapable for the defender of the Confidence Norm. It seems to presuppose that during the time of the epistemic deliberation the agent has to be suspending his beliefs in all the propositions of the conflicting set. But perhaps he doesn’t (and shouldn’t). Perhaps he goes on believing them all (though they are contradictory) with various degrees of credence (all above 0.5) and is just deliberating which one should move below 0.5 at the end of the deliberation.⁵¹ Be that as it may, even if we suppose that the agent suspends his conflicting beliefs during the time of the epistemic deliberation, there is still one psychologically accessible factor that can be evaluated at that time, namely the strength of seeming with which the various propositions seem to him to be true. Therefore, Kelly’s second objection could be seen as a good reason to move from the Confidence Norm to the Truth-Seeming Norm (now more specifically applied at the time of the conflict and deliberation):

(TSN*) If X finds himself with a contradictory set of beliefs (or dispositions to believe) {B1, B2, … Bn}, he should abandon the belief in that set that has the weakest seeming of being true once he becomes aware of the conflict.

Against this kind of norm (discussed under the label of “plausibility”), Kelly offers another objection (Kelly 2005, p. 189). The problem, according to Kelly, is that seemings (or plausibility) are compatible with known falsehood. This is the case in (known) illusions of perception (such as the Müller-Lyer illusion), or in cases of deceptive a priori seemings (such as the seeming of Frege’s Unrestricted Comprehension Principle). Suppose for instance that I have an extremely strong visual seeming that the Müller-Lyer lines are of unequal lengths, and I have two other seemings (weaker ones) that logically imply that they are of equal lengths. Furthermore, suppose I am aware of the Müller-Lyer illusion. In this situation, obviously, I shouldn’t follow the strongest seeming (and abandon one of the weakest), since the strongest seeming is undermined by my knowledge of the Müller-Lyer illusion. So it is not always the case that I should follow (TSN*).

I think this objection by Kelly successfully undermines the Truth-Seeming Norm but that it doesn’t rule out the possibility of finding another internalist norm of belief revision. After all, when I decide to distrust my seeming that the Müller-Lyer lines are of different lengths, I am not doing this due to some internally inaccessible criterion: I am doing it because I have the (second-order or meta-cognitive) seeming that this (first-order) seeming is illusory. Because of this second-order seeming, the content of the first-order seeming (“that the lines are of unequal lengths”) doesn’t seem to me all-things-considered belief-worthy. In contrast, the content of the two other first-order seemings (which generate the contradiction) do seem to me all-things-considered belief-worthy. In other words, the cases of illusions do prove that “a truth-seeming that p is compatible with known falsehood”, but they do not prove that “an all-things-considered seeming of belief-worthiness is compatible with known falsehood”. Therefore, this new epistemic dimension (all-things-considered seeming of belief-worthiness) seems to be what we are looking for when we deliberate about which member of an inconsistent set we should abandon. I will call this dimension “Credibility”. The resulting norm would be the following:

(CrN*) If X finds himself with a contradictory set of beliefs (or dispositions to believe) {B1, B2, … Bn}, he should abandon the belief in that set that has the weakest all-things-considered seeming of belief-worthiness once he becomes aware of the conflict.

The irrationality of the radical revisionist, according to (CrN*), is that he abandons a belief that seems (to himself!) all-things-considered more belief-worthy on the basis of some beliefs that seem (to himself!) all-things-considered less belief-worthy.

An obvious objection to this proposal is the following: if this is the norm we have to follow in resolving conflicts of beliefs, then the radical revisionist can easily satisfy it, and in most cases, arguably does! Zeno, for instance, can easily claim that, even though the proposition that there is movement seems initially true and belief-worthy, yet, once he considers the argument of the dichotomy, his all-things-considered meta-cognitive seeming is that the seeming that there is movement is not belief-worthy after all (or at least not more belief-worthy than the premises of the argument). “That’s why—says Zeno—I abandon this belief: because it ceases to seem to me all-things-considered belief-worthy once I consider the problem.” To a certain extent, the kind of individualist approach to common sense that I am defending here makes this kind of possibility unavoidable: since different people may have different degrees of seemings for the same propositions, this entails that what will be radical revisionism for one will not be radical revisionism for another. So it is possible that the radical revisionist, instead of being someone making an irrational revision of the more certain by the less certain, is in fact someone whose degrees of certainty are very different from the Moorean’s. But brute individual differences in seemings is not the most plausible or common situation: the most plausible or common situation is that the radical revisionist has an all-things-considered low seeming of belief-worthiness for the common-sense proposition because he is under the illusion that he has to “follow the argument wherever it leads” and that the argument “leads” in the definite direction from the premises to the conclusion. In other words, the radical revisionist I am now imagining does revise his beliefs in accordance with his all-things-considered seemings of belief-worthiness, but these seemings are demonstrably erroneous because they are the result of a misunderstanding of the situation—the lack of understanding that “the validity of the argument does not establish which of these beliefs should be rejected.” (Pollock and Cruz 1999, p. 7) In this kind of cases, the Moorean can still criticize the radical revisionist for a principled reason. Not because he hasn’t followed the Credibility Norm (CrN*)—in fact he has—but because he has fallen prey to a logical illusion (which I called earlier “the sophist’s trick”). In other words, the norm he hasn’t followed is the following:

(CrN**) If X finds himself with a contradictory set of beliefs (or dispositions to believe) {B1, B2, … Bn}, he should abandon the belief in that set that has the weakest all-things-considered seeming of belief-worthiness once he becomes aware of the conflict and once he understands that it is just a conflict of contradictory beliefs, with no privilege of the order of presentation proposed by the revisionist.

This will be my best bet for a precise formulation of the Principle of Greater Certainty. Any revision of our initial beliefs that violates this revised Credibility Norm is, very plausibly, irrational. And for all common-sense beliefs (i.e. beliefs that are members of our initial system and with a maximal or very high degree of credibility⁵²), it is very likely that arguments proposed to overturn them will commit such a violation. If this is true, then we can show, with the assumption of only two very modest principles—the Principle of Motivation and the Principle of Greater Certainty (properly understood) —that common-sense beliefs have the following epistemological privilege:

(P3*) for every common-sense belief B of X, it is very unlikely that X will ever have (or have had) a motivation that makes it rational for him to revise B.

4 Conclusion

The common-sense strategy we arrive at, at the end of this paper, is very close to being G.E. Moore’s own version of commonsensism—or an interpretation thereof. What I’ve tried to do is to arrive at this solution starting from the general intuition of the common-sense tradition (the “CPJ project”) and to situate it within the possible trends of common-sense strategies. In particular, I think the Moorean kind of commonsensism is best understood as being a development of what I call the “Dynamic Argument” or the “starting-point Argument” (sometimes referred to in the literature as the argument from “conservatism”, though we have seen that this label has misleading connotations, as it seems to imply a positive evaluation of the starting point, while the Dynamic Argument requires nothing of the sort). It seems to me that the strategy of “isolating” the various argumentative trends allows us to appreciate better what precise role each one is playing. In particular, even in the Moorean strategy, it is useful to distinguish the properly dynamic element (the pure Dynamic Argument, or the Principle of Motivation) from the comparative element (the Principle of Greater Certainty).

I do not rule out the possibility that adding more elements to the strategy would be fruitful—for instance the consideration of the commonality of (some parts of?) our starting point, or their naturalness. This will have to be explored in further work. But the two elements that constitute the present strategy seem to me to be the core that any common-sense philosophy should have, and I’ve tried to show that this core relies on extremely modest principles. This probably explains why Moorean commonsensism has been seen by many prominent analytic philosophers as a reasonable consensus in meta-philosophy.