Metaphysical ecumenicalism and Moore’s proof

Abstract

You have hands, but does it follow that there’s an external material world? Moore thought so. However, we argue that this is a mistake. We defend the Ecumenical View, on which ordinary object terms like “hands” are metaphysically ecumenical, akin to the way that terms like “table” are physically ecumenical: just as there are wooden, metal, or plastic tables, so too there can be material, virtual, or immaterial hands. Moore’s position, however, is metaphysically sectarian: the semantics of “hands” requires a materialist metaphysics. Moore’s proof fails not because it displays bad epistemology but because of its problematic commitment to sectarian metaphysics. We conclude with some explorative thoughts on how the Ecumenical View bears on common sense epistemology and a possible equivocation in our understanding of the premises of Moore’s argument.

1. Introduction

The canonical presentation of Moore’s proof of an external world is as follows:

• P1: Here are two hands.

• P2: If here are two hands, then there is an external world.

Therefore,

• C: There is an external world. ¹

At least since its publication, the proof has troubled philosophers. As Barry Stroud comments: “we immediately feel that Moore’s proof is inadequate” (Stroud 1984: 86). Many philosophers agree. It has been described as “clearly unsatisfactory” (Weatherall 2017: 119), “intellectually unsatisfying” (Carter 2012: 115), and “an obviously annoying failure” (Coliva 2008: 235). At the very least it “sounds funny” (Pryor 2004: 349). ² Like other philosophers, we think that Moore’s proof fails.

Call P1 the “epistemological premise” and P2 the “semantic premise.” Interestingly, while the epistemological premise has received the most attention in the commentary on Moore’s proof, the semantic premise receives the lion’s share of attention in Moore’s paper. Our analysis of the failure of Moore’s argument turns on the idea that there are subtle problems in understanding and interpreting the word “hands” and other ordinary object terms (e.g., “tree” and “desk”). ³

By way of preview, ordinary object terms display a certain degree of metaphysical flexibility: the applicability of the term can be unaffected by certain settled attitudes or discoveries bearing on the referent’s metaphysics (e.g., the fundamental natures of the things that satisfy the terms). We shall refer to this understanding of the meaning of ordinary object terms that emphasizes flexibility as the view that such terms are metaphysically ecumenical. Some understanding of this notion can be made by analogy with physical ecumenicalism: the concept “table” is (at least) physically ecumenical in that it allows that there are many physical variations in the construction of tables: tables may be made of wood, plastic, metal, glass, etc. Scaling-up, the metaphysically ecumenical view says that certain variations in metaphysical structure, for example, materialist, idealist, virtualist, or panpsychist views about ordinary object composition, are consistent with describing something using the term “table.”

As we plan to show in some detail below, the view that ordinary object terms are (to some degree ⁴ ) metaphysically ecumenical is inconsistent with Moore’s view that they are best understood as essentially embellished with a specific metaphysics, for example, that “hand” refers only to material objects. We will refer to appeals to specific metaphysics “metaphysically sectarian” views in contrast to the broader non-sectarian understanding offered by the ecumenical view.

So, part of our brief is that the plausibility of the ecumenical understanding of “hands,” at least in some contexts, offers a serious challenge to P2 of Moore’s argument. We will also try to address some of the aforementioned unease readers feel about Moore’s argument. Our proposed explanation for this unease is that there is a subtle equivocation in how to understand Moore’s use of “hands.” The epistemological premise seems more plausible when read using the ecumenical sense of “hand,” whereas the semantic premise requires the truth of Moore’s version of a sectarian conception of “hand.” The uncertainty of how to understand “hands” makes the argument appear more successful than it is. Many philosophers have noted that it is a point in favor of any analysis of Moore’s argument if it can explain why (P1) is so readily granted, while (C) seems far more problematic. Our analysis of Moore’s argument makes progress toward this goal.

In the next section (§2), we’ll explore Moore’s argument in more detail before advancing our criticisms of the semantic premise (§3) drawing on linguistic presupposition and entailment tests, along with reflection on (hypothetical) usage, before turning to the subtle equivocation in the argument (§4). In §5, we respond to objections from common sense epistemology, x-phi, and those who would advance a sectarian stipulation of “hands.”

2. The argument for the semantic premise

As noted, Moore spends most of “Proof of an External World” defending the semantic premise: approximately 70% of the paper is devoted to it. That’s why it’s surprising that it has received comparatively little attention in the literature. ⁵ In this section, we’ll unpack and evaluate Moore’s argument for the semantic premise (P2). As we’ll see, Moore explores a variety of weaker and stronger conceptions of “external things” that might be implied by “I have hands” and settles on a strong conception. This is important for our diagnosis of the inadequacy of Moore’s proof because it foregrounds our discussion of the metaphysically ecumenical conception of ordinary object terms.

Now a disclaimer. Although we engage in some interpretive exegesis of Moore’s text here, the primary goal is to map the ways in which Mooreans might defend the semantic premise, as Mooreans (or neo-Mooreans) tend to focus their attention exclusively on the epistemological premise (see, e.g., Pritchard 2007, Pryor 2004, Sosa 2007). ⁶ The semantic premise, however, is more complex and puzzling than epistemologists have tended to treat it.

Moore’s employs three different conceptions of “external things”:

Kant famously worried that the existence of external things hadn’t been proved. Moore’s thought was that proving that external things exist in some sense would not be sufficient to defeat Kant’s worry. So, one of Moore’s major goals in “Proof of an External World” was to highlight not only why his “proof” is a genuine proof of an external world but why the sense in which he has proven that there is an external world is precisely the sense that ought to satisfy Kantians. ⁸

What’s the difference between spatial presentation, spatial occupation, and mind-independence, then? Here, we’ll explain how Moore distinguished between them.

2.1. Spatial presentation

Consider seeing a chair and then experiencing a negative after-image of the chair as well. For Moore, both the chair and the negative after-image are presented spatially. The negative after-image of the chair is not like a visually imagined chair, which is not presented as being in any direction relative to one’s own. It doesn’t seem to be in front of one, beside one, under one, etc. (although one might imagine it as being seen from a certain direction, such as by imagining seeing the back of the chair only). A negative after-image, however, appears directionally. Likewise – albeit less clearly than with after-images – a pain might be felt in some part of one’s body and is thereby spatially presented without having an identifiable spatio-temporal location; it is not a “thing to be met with in space” in Moore’s sense. ⁹

2.2. Spatial occupation

Things to be met with in space are not only capable of being presented spatially (as having spatial and directional properties) but are also capable of being perceived by others. Moore thought that although both the chair and the chair after-image are presented spatially, of the two only the chair can be met with in space because no one else could perceive the chair after-image that he perceives: “If both were (the chair and the after-image were perceivable), it would follow that somebody else might see the very same two images which I see,” which he thinks isn’t possible (Moore 1939 [2013]: 153). That is, while both things appear to have spatial properties – spatial properties are perceptually represented – only the chair can be perceived by others. This modal property of <being possibly perceived by others> is what differentiates spatio-temporal occupation from spatio-temporal presentation. ¹⁰

Moore thought that the spatial occupation sense of “external things” is co-extensive with “physical” and “material object”: “all things of the sort that philosophers have been used to calling ‘physical objects’, ‘material things’, or ‘bodies’ obviously come under this head,” he said (Moore 1939 [2013]: 130). However, he thought of the spatial occupation sense of externality as being broader than “physical object” because of minor objects like shadows, which have spatio-temporal location but aren’t thought of as being “physical.” As Moore put it: “The phrase ‘things which are to be met with in space’ can … be naturally understood as having a very wide meaning” (ibid).

With these distinctions in play, Moore argues that from the fact that a plant exists or that a soap-bubble exists, for example, it follows that there are things to be met with in space. He thus accepts the following entailment:

Ordinary Objects Are Spatial

∃x x is a plant => Spatio-temporal Location(x)

Crucially, he says that “you will not require also to give a separate proof that from the proposition that there are plants it does follow that there are things to be met with in space” (Moore 1939 [2013]: 138). Mutatis mutandis for other ordinary objects. So, the thought is that it is just part of what “plant” and other ordinary object terms mean that they are things to be met with in space (i.e., have spatio-temporal properties). So, Moore seems to have thought that Ordinary Objects Are Spatial is analytic. Crucially, for Moore, this doesn’t yet show that “plant” refers to something external in any other sense.

Moore’s argument for the entailment is unclear, however. At some points, he suggests that he is just stipulating it: “I am so using ‘things to be met with in space’ that, in the case of each of these kinds of ‘things’, from the proposition that there are ‘things’ of that kind it follows that there are things to be met with in space” (Moore 1939 [2013]: 137). This is a stipulative entailment. One is of course free to stipulate the opposite as well.

However, given the fact that he takes great pains to distinguish spatio-temporal presentation from spatio-temporal occupation, we think it best to understand it not as a mere stipulation but an argument from cases and conceptual analysis: once we appreciate the property differences between plants and plant after-images, for example, this furnishes us with a conception of spatio-temporal presentation (i.e., things presented in space) and spatio-temporal occupation (i.e., things to be met with in space) that can be applied to other cases. We know by experience that plant after-images aren’t the sort of things that are spatio-temporally located. Rather, they are merely spatio-temporarily presented. ¹¹

Interestingly, Moore denies that this kind of argument suffices for the semantic premise. While he says that, with regard to “things to be met with in space,” that no proof is necessary for Ordinary Objects Are Spatial – that the entailment is (apparently) obvious – that “with regard to the phrase ‘things external to our minds’, …. the case is different” (Moore 1939 [2013]: 138). Indeed, he says that it’s possible that a plant might be met in space but not be external to our minds and that “external to our minds” is not a mere synonym of “to be met with in space” (ibid). ¹²

2.3. Mind-independence

So, we need a better understanding of Moore’s distinction between things to be met with in space and a thing’s externality. To this end, Moore reintroduces an argument he gave in his earlier (1925) work, “A Defence of Common sense,” where he says that “physical facts” and “mental facts” are distinguished by their logical independence (Moore 1925 [2013]: 50). The thought here is that statements that feature (perceptual) mental-verbs like “I saw,” “I heard,” or “I imagined” are all such that they presuppose that someone had an experience. From “I saw my hands,” it follows that I had some experience at that time. In contrast, statements which lack mental-verbs like “I am 4’11” or “there’s a tree in the quad” lack such implications. We can capture Moore’s view about what “external to our minds” means as follows:

Mind-independence: x is external to S’s mind only if the proposition that x exists doesn’t entail that there is someone, S, who had experiences. ¹³

Moore’s argument for the semantic premise draws on this principle. His thought was that things to be met with in space are external to our minds in this sense that they all respect Mind-independence. He argues for this in two steps. First, he employs a self-contradiction test on ordinary object terms: that from accepting the proposition that there is a soap-bubble here – which is something to be met with in space – that one would not contradict themselves if they added: “it was here before I perceived it” or “it might stay here even if no one perceives it” (Moore 1939 [2013]: 144). ¹⁴ It would not be like asserting “there’s a bachelor here,” adding “there’s no unmarried man here,” which would be contradictory.

The second step in his argument is that the logical independence between “there’s a soap-bubble here” and the corresponding mental-statement “someone had an experience as of a soap-bubble” is what distinguishes something’s being a real soap-bubble from its being a merely hallucinated soap-bubble or similar:

This seems to me to be part of what is meant by saying that it is a real soap-bubble, as distinguished, for instance, from an hallucination of a soap-bubble (Moore 1939 [2013]: 144).

So, on Moore’s view being an ordinary object partly consists in its obeying Mind-independence. As he puts it, a thing “would not be a soap-bubble unless its existence at any given time were logically independent of my perception of it at that time” (ibid). If the soap-bubble could exist only if it were perceived at some time, then this would compromise our distinction between something’s being a real soap-bubble and a merely hallucinated or imagined soap-bubble.

As we can see, then, Moore’s argument for the semantic premise draws on semantic intuitions about ordinary object terms: crucially, about their referents’ logical independence from psychological phenomena. We can summarize Moore’s argument for the semantic premise as follows:

Sub-Semantic Argument

• (P2a) We can conceptually distinguish between hands and mere hallucinations of hands only if the referent of “hand” obeys Mind-independence.

• (P2b) We can conceptually distinguish between hands and mere hallucinations of hands.

Therefore,

• (P2c) The referent of “hand” obeys Mind-independence.

Given relatively innocuous assumptions about reference, (P2c) implies the semantic premise, that if one has two hands, then there are external things. Mutatis mutandis for other ordinary object concepts.

Unsurprisingly, then, Moore’s argument for the semantic premise draws on semantic intuitions about ordinary object terms. As we noted in the introduction, Moore spent most of his lecture on the semantic premise, with his defense of the epistemological premise, (P1), coming down to two key points. First, that one doesn’t need to prove “I have hands” to know it, and, second, that it would be “absurd” to deny that he knows it (see Moore 1939 [2013]: 146). The extant literature on Mooreanism defends the premise in other ways: by appeal to dogmatism (Pryor 2004), safety theories (Sosa 1999); (Pritchard 2005), evidentialism (Steup 2011), or disjunctivism (Pritchard 2012). Fortunately, the ecumenical view doesn’t require us to disagree with Moore or Mooreans here. As we see it, we can have justification to believe “here are two hands.” ¹⁵ And while some philosophers have taken issue with Moore’s conception of proof (De 2020), as well as its rational persuasiveness, granting that it is a proof (Pryor 2004), our issue lies exclusively with the semantic premise. Metaphysical ecumenicalists seek to preserve the rationality and truth of our ordinary beliefs, “I have two hands,” “there are chairs,” and the like without metaphysical sectarianism. ¹⁶

As a final issue, one might ask whether Moore is really allied to a “materialist” view of ordinary objects, on which ordinary material objects are mind-independent. Materialism doesn’t say simply that ordinary objects have spatial location, but that such spatially located things are mind-independent. This caveat is important. Berkeley, after all, never denied that ordinary objects exist in space, but he did deny that they are mind-independent. He called that combination “materialism” about ordinary objects. ¹⁷ A more ecumenical view about ordinary objects would say that even if they are mind-independent – they aren’t logically entailed by our mental states – they might nevertheless lack material substance; they might be functional kinds, digital, or some hitherto unknown kind of stuff. ¹⁸

There are textual hints that Moore allies himself to materialism about ordinary objects. This doesn’t mean that he accepted metaphysical materialism, the thesis that all objects are physical or material, only the restricted view that ordinary objects – tables, chairs, plants, animals – are mind-independent bodies that exist in space. In his Replies, Moore tells us that it “seems quite obvious to me that we so use “I hear a rat,” that at least part of what we mean by this is that we are hearing a physical sound” (Moore 1968, 645). Moore’s emphasis on the physicality of the object is important. It suggests that assuming rats are a kind of ordinary object – that his view here is meant to generalize – that he thinks it’s part of what an ordinary object term means that it is a physical thing. And given what he argued in “Proof of an External World,” physical things are necessarily mind-independent.

Moreover, in his reply to John Wisdom, Moore clarifies that “some philosophers have used ‘material thing’ in such a sense that from ‘There are no material things’ there does follow ‘There are no human hands’; and it was only this usage of ‘There are no material things’ that I meant to say the proposition then expressed by these words can be proved false in the way I gave [in Proof of an External World]” (Moore 1968, 670). Hence, Moore seems to have in mind a materialism about ordinary objects (so understood) in mind in “Proof of an External World.”

Moore also addresses the point, apparently made by Wittgenstein, that no philosopher who denied the existence of matter or physical things ever meant to deny the existence of ordinary objects (like pants or hands). Moore’s response is telling. He says, “I think the statement is simply false”; that “they have meant to assert that no such proposition as that pants exist is true” (ibid). This is some further evidence that Moore accepts materialism about ordinary objects, as we understand it.

3. Four arguments against the semantic premise

In this section, we will offer four arguments against Moore’s semantic premise.

3.1. Usage

The first argument against the semantic premise turns on how the term “hands” is used by several authors. As we hope to show in more detail below, Moore’s argument for the semantic premise requires his sectarian understanding. A consequence of this is that Moore must say that any use of the word “hands” invoking a non-materialist metaphysics is false, or the usage must be understood in some non-literal fashion. Our argument is that this is at odds with how the term is often used. (Below we will suggest a weaker version of this conclusion that also undermines the semantic premise). By “non-materialist” we mean one which denies one or both of that ordinary objects are physical and mind-independent.

Consider that Berkeley appears to use the word “hands” at variance with Moore’s preferred understanding:

I don’t argue against the existence of any one thing that we can take in, either by sense or reflection. I don’t in the least question that the things I see with my eyes and touch with my hands do exist, really exist. The only thing whose existence I deny is what philosophers call “matter” or “corporeal substance.” And in denying this I do no harm to the rest of mankind – that is, to people other than philosophers – because they will never miss it. The atheist indeed will lose the rhetorical help he gets from an empty name, “matter,” which he uses to support his impiety; and the philosophers may find that they have lost a great opportunity for word spinning and disputation. ¹⁹

If Moore is correct that part of the meaning of the word “hands” is that they are material, then Berkeley has said something that is analytically false. It would be akin to Berkeley saying “I do not deny the existence of bachelors, I merely deny that they are unmarried.” If Moore is right, it is a wonder that Berkeley is not taken to task for being mistaken about the meaning of “hands” and other ordinary terms.

Consider also Charles Dickens’ A Christmas Carol, when Scrooge asks: “I am in the presence of the Ghost of Christmas Yet To Come?” The Spirit answered not, but pointed onward with its hand.” ²⁰ Again, if Moore is correct, it seems that Dickens has written a statement that is analytically false. Similar sorts of things might be said about how to refer to the prehensile appendages of the characters in the Matrix movies by the Wachowski sisters. (We will discuss this possibility in more detail below.)

A more recent example comes from an interview with Mark Zuckerberg who freely used “hands” and “his hands” to talk about his hands qua avatar in the metaverse. ²¹ Here, if we interpret Zuckerberg as employing a materialist sectarian understanding of “hands,” his sentence “I was waving my hands” is false. It forces the listener to reinterpret Zuckerberg so that what he meant was his avatar’s hand-like appendages. If, however, we interpret Zuckerberg as employing an ecumenical understanding of “hands,” he was speaking literally (and truly). The context provided the listener with knowledge – the knowledge of which kind of hands Zuckerberg was referring to – so that no reinterpretation of Zuckerberg is necessary. In turn, it’s not necessary to ascribe analytic error either, nor a metaphorical use of the term. ²² A similar point applies to the examples from Berkeley, Dickens, and the Wachowski sisters. ²³

3.2. The metaphysical Big Reveal

Our second argument considers how speakers might react if they discovered that materialism about ordinary objects, the idea that ordinary objects are material or physical – where this implies being spatial and mind-independent – is wrong. Consider:

The Big Reveal: Billions of people sat glued to their TVs and computers, tracking what looked to be the nuclear annihilation of civilization: every nuclear country on Earth had launched its entire nuclear arsenals. Miraculously, all the missiles were destroyed in midair with no nuclear fallout. Shortly thereafter, a giant figure appeared to each person on Earth and said: “I’m your one true God. Stop fucking around and play nice. Oh, and by the way, Berkeley was right about everything: if I even blink, much of the universe will disappear forever.” Lawyers for Big Bad Company were delighted with this news as they had studied Moore’s argument in college. Big Bad Company was in the middle of a legal battle when God revealed herself. The plaintiff, an employee at Big Bad Company, alleged that the company was criminally negligent for the loss of his hands in factory machinery that was not well-maintained for cost-cutting reasons. The lawyers attempted to defend Big Bad Company by using modus tollens on P2. (P2: If here are two hands, then there is an external world). The lawyers argued that since there is no material external world, the plaintiff never had any hands to begin with. So, the plaintiff could not have lost his hands in an industrial accident. Hence, the case should be dismissed.

We imagine that no one will be tempted by the lawyers’ argument. We analyze its failure in terms of the implausibility of the assumption that the Big Reveal shows that people never had hands. Rather, it seems a more plausible reaction is that people learned that their hands were made of immaterial stuff all along. In other words, the Big Reveal doesn’t reveal that people had false beliefs that they had hands, but that they had false beliefs about the composition of their hands.

Of course, this use of “hands” is incompatible with Moore’s proffered sectarian understanding of the term.

3.3. Neo’s hands

Our third argument is similar to the second, except that it asks what we might say when there are multiple (putative) candidates for the word “hands.” This line of thought is helped by the thought that the belief that macroscopic material objects are the typical cause of our sense experience is defeasible. Movies like The 13 ^th Floor and The Matrix are premised on this: viewers are led to see how the protagonists might come to the revelation that they live in a virtual world. Consider then the plight of Neo: the protagonist in the Matrix movies series. Neo was immersed in a virtual reality (“The Matrix”) for the first 35 or so years of his life, before being freed by those who would become his co-conspirators. In the movie, most humans are plugged into the Matrix for their whole lives, unaware that their bodies exist outside the Matrix in pods designed to support the biological functions of the body. Now think about Neo before he escaped from the Matrix. How many hands does he have while he is plugged into the Matrix? Here are three different answers:

1. (I) Neo has two hands – the two hands attached to his physical body in the pod. Yes, it is true that Neo has never seen or used these hands, but this just shows that one can have hands and not know much about them. The Matrix is entirely virtual, so there are no body parts in the Matrix, including hands. Body parts are essentially comprised of material stuff.

2. (II) Neo has four hands – the two hands he uses in the Matrix, and two hands attached to his physical body in the pod. Yes, Neo knows only of the two hands he has in the Matrix, but he’ll soon learn that he has two more hands.

3. (III) Neo has two hands – the two hands he uses in the Matrix. Yes, it is true that his brain is part of a material body in the pod, and the body in the pod has two hands. But Neo has never seen nor used these hands.

To understand three possible answers, it will help to introduce some distinctions:

• M: “Hands” refers to the prehensile part of the upper limb of humans and other species, and “hands” refers to material objects that obey Mind-independence. (Hereafter “mhands.”)

• V: “Hands” refers to the prehensile part of the upper limb of humans and other species, and “hands” refers to virtual objects. (Hereafter “vhands.”)

Both mhands are vhands are sectarian understandings of the term “hands” in the sense that the terms are semantically embellished by a particular metaphysics. This is in contrast with the aforementioned ecumenical understanding of “hands” that is not essentially linked to a particular metaphysics:

• E: “Hands” refers to the prehensile part of the upper limb of humans and other species, and “hands” refers to objects that may or may not be material, or virtual, or immaterial, etc. (Hereafter “ehands.”)

So, answering (I) appears to require “hands” is understood as mhands, answering (II) appears to require “hands” is understood as ehands, and answering (III) appears to require understanding “hands” as vhands. It is clear that the premises of Moore’s argument require that (I) is true, that is, that “hands” is understood as mhands; otherwise, his argument is invalid. (More on this below.)

Over the last several years we asked people which of the three answers they find most compelling. Among several hundred non-philosophers (this group includes philosophy students, family, and friends), (II) and (III) are roughly equal in popularity, with very little support for (I) in this group. Among the small group of philosophers (fewer than 20) asked, all three options had about equal support among our limited survey group. Here, semantic intuitions clearly vary.

Our informal data strongly suggests that “hands” is not used univocally. We conjecture that the same non-univocal results would be found if these scenarios were subject to empirically rigorous testing. Thus, we allow that if such testing was undertaken by psychologists, linguists, or x-phi philosophers, we would have much stronger evidence about how “the folk” and how philosophers understand the meaning of “hands.” Furthermore, we allow that this testing might reveal that our hypothesis about how subjects use the term “hands” might be disconfirmed. In other words, we are “mere” theoreticians: we leave it to experimentalists to subject this claim to more rigorous testing. This sort of division of labor is common in some areas of inquiry, for example, theoretical physicists make conjectures, and experimental physicists confirm or infirm the conjectures. Still, we think it is highly unlikely that such investigations will favor Moore’s view, and as we argue below, linguistic tests for presupposition and entailment strongly support our verdict.

As noted above, if Moore’s argument is valid, then it must be the case that (I) offers the correct answer, so it seems that Moore will have to admit that there is a high error rate in the use of “hands.” We’ll discuss the relevance of this point below.

3.4. The “Yes, but”-test

The final argument against Moore’s semantic premise leverages a standard way of testing for lexical entailment and presupposition employed by linguists: the “Yes, but”-test (henceforth YBT) (Cummins and Kastos 2012; Amaral & Cummins 2015; Schwarz 2016). Speakers consider a question–answer pair, with the answer both affirming and denying what the question expresses (or some related content to be tested). Here’s the basic idea:

Q: Did Samanya go to university?

A: #Yes, but she didn’t.

The answer is clearly infelicitous. The basic idea is that the “yes” response commits the speaker to all of the entailments of the question – that is, “yes” commits the speaker to “Samanya went to university” – so following this up with “but she didn’t go to university” is contradictory.

YBT distinguishes entailments from other types of inferred meaning, such as implicatures and pragmatic presupposition. For example, consider:

Q: Did Samanya drink some of the beers?

A: Yes; indeed, she drank all of the beers.

In this case, there is scalar implicature. After all, compare A with A_A “Yes, but she didn’t drink some of the beers – she drank them all!”. Notice that A_A is semantically permitted while A_N “Yes, but she didn’t drink any of the beers” is not.

Now consider a priori analytical entailments, like that bachelor entails being an unmarried man. The YBT predicts the correct verdicts about these kinds of entailments too:

Q: Is the man Darren a bachelor?

A: #Yes, but he’s married. (Or: #Yes, in fact he’s married.)

A is clearly infelicitous. The YBT correctly predicts that “Darren is unmarried” is an entailment of “Darren is a bachelor.”

Indeed, the YBT gives us the correct prediction about “non-core” meaning because of contextual shifts as well. Certain phrases prime people to coerce a certain interpretation of an answer-sentence the concept is embedded within so that there’s no inconsistency between the question- and answer-sentence. This can happen, for example, because of shared information about the referent’s context. To see how this works, consider a situation in which Darren is married, and the best man among “stereotypical lads” at a bachelor party:

• Bachelor Party:

  • Q: Is Darren a bachelor?

  • A: No, but he is for the weekend.

This is not infelicitous. Phrases like “… for the weekend” coerce semantic selection, so that the relevant term is interpreted with its non-standard (i.e., non-core) meaning. In Bachelor Party, the key phrase “… for the weekend” enables a consistent interpretation of the answer-sentence even though flat denial contradicts the fact that the subject is a bachelor. The phenomenon is a general one, not specific to “bachelor.” The phrase “for the weekend” also coerces an interpretation of other question-terms, like:

• Accountant:

  • Q_a: Are you an accountant?

  • A_a: No, but I am for the weekend.

• Nurse:

  • Q_D: Are you a nurse?

  • A_D: No, but I am for the weekend.

Imagine that Darren is preparing his taxes over the weekend, or that he is nursing a family member who has a cold over the weekend. In these cases, we have consistent answer-sentences; crucially, absent the phrases, we wouldn’t.

We now want to test Moore’s view about “hands” and other ordinary object terms like “keys” or “house” using the YBT. Consider the following example:

Keys 1: You wake up in the morning, leave your home, and check your pockets to see whether you have your keys there, and you discover that they were in your pocket.

To the question: “Do you have keys?” The answer is clearly “yes.” Now let’s add the following twist:

Keys 2: Later the same day, you discover that you and everyone else you have ever known have always been a brain in a vat, hooked up to a complex simulation which directly stimulates the brain.

Now consider:

• Q: Do you have keys?

• A: ?Yes, but I’m a brain in a vat in a simulation.

There is no clear infelicity here, not like with:

• A1: #Yes, but I don’t have keys.

• A2: #Yes, but I don’t have anything which unlocks anything.

After all, one might assert A above “Yes, but I’m a brain-in-a-vat” all the while trying to open a door, rending the negative conjunct that follows difficult to understand. Infelicitous answers also permitted by our initial question Q include:

• A3: Yes, but they are not wholly metal.

• A4: Yes, but they are wholly plastic.

These are unusual, but not infelicitous. This gives us a clue with respect to their lexical semantics. The thought here is that, semantically, being <keys> allows them to be compositionally many different sorts of things: metal, plastic, wood, or even glass. This is what we referred to above as physical ecumenicalism. If, for example, the person wanted to get another set of keys, the fact that the new set is steel rather than, say, brass wouldn’t affect their status as keys. As the YBT suggests, the BIV answer-sentence doesn’t appear to contradict the question-sentence; it’s not like the relationship between Q-A1 and Q-A2. Compare these sentence pairs with others, like:

• Digital:

  • Q5: Do you have hands?

  • A5: Yes, but everything is digital.

  • A5!: Yes, in fact they’re completely digital.

• Immaterial:

  • Q6: Do you have hands?

  • A6: Yes, but everything is immaterial.

  • A6!: Yes, in fact they’re completely immaterial.

Here, we submit that there’s no infelicity. They are unconventional, but not contradictory; again, not at all like the other pairs. Two comments: first, we think that <hands> displays metaphysical ecumenicalism: that the concept displays a certain degree of tolerance with respect to metaphysical compositional variation. Physical hands, mind-dependent hands, indeed, Panpsychist hands-as-part-of-God are all hands. Finding out that your hands are, say, God-parts, or mind-dependent, doesn’t affect their status as hands. Put generally, many of our ordinary object concepts will predictably display a certain degree of metaphysical ecumenicalism.

The YBT arguments suggest that the metaphysical ecumenical understanding of “hands” is not revisionary; it is part of what we already mean. Importantly, we acknowledge that this is a (partly) empirical claim about how speakers use “hands” and allied terms, and so potentially vulnerable to empirical refutation. Of course, this is true of all empirical claims. Naturally, we expect any further investigation to corroborate the verdicts of YBT here, and welcome studies that include a more diverse pool of speakers. Also of interest would be cross-linguistic studies in the same way that cross-linguistic studies of “knowledge” in Gettier cases have proved interesting (Machery et al. 2017). Of special interest is cross-linguistic studies that look at how “hand” might be translated into languages used by those who hold non-materialistic views, for example, some subsets of Hinduism maintain a form of idealism somewhat akin to Berkeley’s, on which ordinary objects (among other things) are immaterial, mind-dependent things (Raju 2008 2013).

The strongest conclusion that might be drawn from these four arguments against the semantic premise is that people generally mean by “hands” ehands. This would have the implication that Moore (at best) employs a stipulative definition of “hands.” A weaker conclusion is that “hands” is ambiguous: it means mhands in some contexts and ehands in others. What fixes the reference might depend on a variety of contextual factors, like practical interests (imagine Moore is within the Matrix, and more interested in his vhands than mhands), or salient presupposition triggers (e.g., after an accident in which Moore loses his hands and gets a prosthetic – which needs replacing – in the sentence “Did you lose your hands again?”, “again” is a trigger for the speakers that “hand” refers to prosthetic non-biological hands). An even weaker conclusion is that Moore is correct about the proper usage of “hands,” but people very frequently use the term incorrectly to convey a different meaning. Obviously, it is not uncommon for people to use terms incorrectly. When someone says, “I literally died laughing,” it’s clear that charitably understood they’re using “literally” to mean something more akin to “metaphorically.” An uncharitable understanding of such an obvious misuse of the term is to wish what one said is literally true.

As we hope to show in the following section, each of these three conclusions is sufficient to challenge the cogency of Moore’s argument.

4. On a subtle equivocation

As intimated above, Moore requires “hands” to be understood as mhands throughout his argument. That is, as follows:

1. (A) Material Interpretation:

2. P1m. Here are two mhands.

3. P2m. If here are two mhands, then there is an external world.

4. Therefore

5. C: There is an external world.

The requirement that mhands be used in both premises can be seen from the following two versions of the argument:

1. (B) Ecumenical Interpretation:

2. P1e. Here are two ehands.

3. P2e. If here are two ehands, then there is an external world.

4. Therefore

5. C: There is an external world.

1. (C) Two-Term Interpretation

2. P1e. Here are two ehands.

3. P2m. If there are two mhands, then there is an external world.

4. Therefore

5. C: There is an external world.

Obviously, P2e is false, since there are different sectarian metaphysics compatible with ehands, hence the antecedent is not sufficient for the consequent. The two-term version employing P2m is invalid, since different terms appear in the conditional and non-conditional premises.

Both arguments might appear more successful than they are because of two different ways of equivocating. The (C) version applied to the canonical version of Moore’s argument says that “hands” in P1 is understood as ehands, and “hands” in P2 is understood as mhands, resulting in a subtly invalid argument. The (B) version says that “hands” means the same thing in both premises of in the canonical presentation, but what goes unnoticed is that the truth of P2 requires a different understanding of “hands.” In either case, the argument appears more successful than it is because of subtle differences in the meaning of “hands.”

There is also a fourth interpretation we should consider:

1. (D) Misunderstanding: The term “hands” properly means mhands, but speakers or hearers interpret each other as using the term to mean ehands in at least some contexts. Thus, speakers in some contexts (wrongly) understand Moore’s argument as (B) or (C).

The following is intended to show how something analogous to Misunderstanding might occur: Suppose during his worldwide lecture series on his Proof of the External World, Moore had a cold. He asked if anyone in the audience could offer him a Kleenex. Several dozen offers came immediately. Good fortune had come your way: Moore took up your offer and of course as an eager graduate student you jumped at the chance to ingratiate yourself with the great philosopher. After blowing his nose, Moore went on to offer an additional proof to the audience:

1. Moore’s Other Proof

2. P3: Here is a Kleenex.

3. P4: If here is a Kleenex, then the Kimberly-Clark company exists.

4. C2: Therefore, the Kimberly-Clark company exists.

Moore offered proof of P3 with great fanfare: he waved around what he claimed to be a Kleenex for all to see. He asked the audience whether anyone objected to P3 – there was no objection from the audience. He then argued for P4 by noting that the word “Kleenex” is a trademark belonging to the Kimberly-Clark company. Anything properly called a “Kleenex” must be made by the Kimberly-Clark company. With great pride he pronounced that he has proved that the Kimberly-Clark company exits. Moore’s Other Proof was met with a polite standing ovation, although there were whispers that the argument sounded funny, that it was unsatisfactory, or even an annoying failure.

Understandably, you did not have the heart (or nerve) to tell Moore that you hadn’t entirely understood what he meant when he asked for a Kleenex. You understood this as referring to any paper facial tissue, not necessarily one made by Kimberly-Clark. You had grabbed a handful of facial tissues from home before leaving for the lecture, and you strongly suspect you bought a cheaper generic brand. Your analysis of Moore’s Other Proof is as follows: P3 was readily granted by the audience because they understood “Kleenex” to refer to a facial tissue, not a specific brand of facial tissue. The audience also accepted Moore’s argument for P4 about the meaning of the word “Kleenex.” What makes the argument sound funny is an equivocation in how to understand “Kleenex.” Think of an “ecumenical Kleenex” (hereafter an “eKleenex”) as one that allows facial tissues made by different companies all to count as Kleenexes. Think of a “sectarian Kleenex” (hereafter a “sKleenex”) as referring to facial tissues made by the Kimberly-Clark company. So, if “Kleenex” in the argument is understood as an eKleenex in both P3 and P4, then P4 is false. If the argument is understood as a sKleenex, however, then P3 is likely false. To put the point slightly differently, P3 is much more plausible when interpreted as referring to eKleenexes and P4 is much more plausible when interpreted as referring to sKleenexes. Of course, the argument is invalid when disambiguated in this way. The reason Moore’s Other Proof looks successful is because the audience systematically misunderstood that “Kleenexes” refers to sKleenexes and never properly to eKleenexes.

We are now in a position to see why the (D) interpretation is a problem for Moore’s proof. If (D) is true, then the proper use of “hands” refers to mhands and never to vhands, or ehands, etc. However, if (D) is true, then subjects often misunderstand “hands” in (P1) as referring to ehands. If so, then in terms of subjects’ uptake, either subjects see P2 as false (along the lines suggested by (B)), or invalid (along the lines suggested by (C)).

For present purposes, we do not need to adjudicate between the (B), (C), and (D) interpretation. It is sufficient to note that if any of this set is correct, then Moore’s argument is not dialectically successful. We believe that our arguments support the claim that the likelihood of one of (B), (C), or (D) is true is higher than the likelihood that (A) is true.

Our argument also offers a straightforward answer as to why P1 seems more readily granted than C. Given some common assumptions about consistency of probability judgments, if the canonical version of Moore’s argument is correct, then it can’t be the case that we could consistently be more confident about the conjunction of the premises (P1 and P2) than C, since the argument is valid. As indicated above, we think that people would still believe they had hands even if it is discovered that Berkeley is correct, or if we find we are in the Matrix. This means that some interpretation other than the (A) interpretation must be correct, but it is much easier to see how one might be more confident in the premises than the conclusion of an invalid argument (C) or more confident in C when there is an obviously false premise (as with (B)). Finally, if (D) is true, then Moore is right about how “hands” is properly understood, but subjects often use a mistaken understanding of “hands” along the lines suggested by the (B) or (C) interpretation.

5. Some objections answered

In this section, we shall try to answer some possible objections.

5.1. More empirical studies

It might be objected that more rigorous empirical studies are needed about the meaning and language users’ uptake of “hand” before we reject Moore’s semantic premise with full confidence.

We agree. One of the upshots of this paper is that establishing the semantic premise of the argument is far more difficult than it first appears, as much of the literature takes it for granted, which we have at least shown is controversial and shouldn’t be treated as non-negotiable.

What’s more, we have intimated several lines of systematic empirical evidence that might bear on its truth:

1. (i) In the first argument against the semantic premise, we offered some preliminary “field evidence” when we discussed how Dickens and Zuckerberg appear to use the word “hands” in a way that conflicts with Moore’s semantic premise. We imagine that empirical “field linguistics” might look more systematically at evidence using linguistic databases to see how non-philosophers use the word “hands.”

2. (ii) The last three of the four arguments against the semantic premise may lend themselves to laboratory testing. The arguments might be operationalized in a way that could yield more systematic empirical evidence to test the plausibility of More’s semantic premise. Linguists or psychologists are probably the best qualified to carry out such studies. As noted above, conducting such investigations on a culturally and linguistically diverse sample might yield interesting results.

While we agree that more empirical evidence would be necessary to reject the semantic premise with full confidence, we still hold that the preponderance of evidence at present supports leaning toward the rejection of the semantic premise. Our credences should side against it.

Suppose those sympathetic to the “more empirical evidence” objection dig in here and say that more systematic evidence is required before we would be in a position to even lean toward the rejection of the semantic premise. At least for the sake of the dialectic, we can concede the point so long as the same high evidential standards are consistently maintained. In particular, absent the aforementioned systematic empirical studies, a consistent application of these high evidential standards requires not accepting Moore’s semantic premise. That is, these high evidential standards imply suspending judgment on Moore’s semantic premise without the relevant systematic empirical studies to move the evidential dials.

Of course, such a stance directly conflicts with Moore’s claim that we are in a position to know the premises of the argument. And so, the objection is not friendly to Moore’s proof: yes, it undermines our claim that the preponderance of evidence supports leaning against the truth of the semantic premise, but at the cost of undermining Moore’s proof anyway.

5.2. Stipulation

It might be thought that this entire discussion could be circumvented if Moore simply stipulated that what he means by “hands” is mhands. So stipulated, it doesn’t matter what folks think about the semantics of “hands.” In other words, the objection on Moore’s behalf is that he is attempting to make precise a polysemous term. Thus, in one fell swoop the objection does away with the problems suggested by (B), (C), and (D).

We have no objection to Moore (or Mooreans) stipulating in this way. And we allow that the canonical presentation of the argument is best stipulated to be the (A)-interpretation. The question is whether doing so will weaken the dialectical strength of his argument, since it calls attention to the fact that there are different ways of describing what we see when Moore waves when he is demonstrating the truth of P1. Moore’s stipulation allows him to say that he sees mhands, but Berkeley might object, appealing to his own stipulative definition, that Moore is waving ihands.

Similarly, those who favor the virtual world hypothesis using the stipulation tactic can say that they see vhands, and those who prefer to stay away from sectarianism say that they see ehands. The problem then is that opting for a stipulative definition opens the floodgates to many possible proofs of differing sectarian metaphysics and ecumenical metaphysics. Thus, this line of objection leads to what has been described in the literature as the “problem of too many proofs” (Walker 2020).

5.3. Common sense epistemology

Common sense philosophers argue that common sense has a distinctive epistemic standing. For example, Lemos (2004) thinks that if a philosophical theory is inconsistent with common sense, that’s some reason to reject the theory. (Kelly 2005) says something similar: “there are very substantial limits on how radical a change in our views philosophy might legitimately inspire” (Kelly 2005, 181). Soames (2003) argues that common sense is the proper starting point for philosophical inquiry; that common sense lacks “the sorts of claims that can be overturned by philosophical argument” (Soames 2003, 5). So, one might worry that our criticisms are incompatible with common sense, and given that our argument is incompatible with common sense, our argument must be unsound.

This is not the place to delve into metaphilosophical issues like the epistemological status of common sense. One such issue that would have to be addressed is the status of sectarian appeals to metaphysics. In particular, an appeal to common sense will have to address the fact that other cultures appear to have very different sectarian views about metaphysics.

For example, it’s not common to all cultures to think of tables or even one’s body as external in Moore’s sense. Raju argues that “Idealism is the dominant philosophy of India,” and, indeed, is embedded within the religious traditions of East Asia (Raju (1953), 212). Other cultures have views that look closer to panpsychism, and so on (Povinelli 1995). We do not need to settle any of this to note that if this line of objection is to go through, it must be the case that some very substantive notion of common sense is appealed to – one that makes appeal to the sectarian metaphysics of materialism about ordinary objects and external world realism as the correct views.

But let us suppose that the appeal to this sort of very substantive notion of common sense is correct, and so the conclusion of Moore’s argument must be correct. Still, we believe this is compatible with our main line of criticism. All that is required is that we reconstruct Moore’s argument as an enthymematic argument, where UP notes the unstated premise.

Enthymematic Version (EN)

• P1e. Here are two ehands.

• UP: Sectarian Common Sense Materialism (SCSM): things in our immediate environment that we normally see are material objects that obey Mind-Independence.

• P2’e. If here are two ehands, and SCSM is true, then there is an external world.

• Therefore

• C: There is an external world.

Notice that EN is consistent with Mooreans making an error about the meaning of “hands” and it permits the meaning of “hands” to take any number of different sectarian views about the metaphysics of composition. Mutatis mutandis for other, ordinary object terms that might be used in Moorean proofs. EN also explains the dissatisfaction with Moore’s argument in its canonical formulation: it is hard to see how the evidence for P1 – merely looking at Moore furiously waving his hands – could establish a metaphysically substantive and sectarian conclusion like C. The explanation implicit in EN is that SCSM is doing the heavy lifting in the argument. It explains why one might be more confident in the first epistemological premise (P1e) than in C. The reason is that, although the argument is valid, it is still possible that P1e is true while C is false. Thus, the common sense objection does not weaken our case.

Finally, we note that what philosophers typically mean by “is inconsistent with common sense” is when a theory entails that “there are tables” or “I have hands” and the like are false or unjustified (see, e.g., Merricks forthcoming). Revisionary metaphysics is characteristically metaphysics inconsistent with “common sense” in that sense. Fortunately, our view lacks those revisionary implications. The ecumenical view permits that “I have hands” can be true – crucially, part of the furniture of reality – and justifiably believed. ^{24 , 25}

6. Conclusion

Moore famously argued that one could prove that there is an external world from “here are two hands.” Neo-Mooreans likewise embrace the proof; whatever dialectical misgivings it might suffer in a contest with the radical skeptic, it nevertheless proves that there is an external world. However, the proof fails because it requires that ordinary object terms (like “hands”) are understood as metaphysically sectarian, but such terms are metaphysically ecumenical, or at least they are often (misunderstood) as ecumenical.