Nearly 30 years later, in The Construction of Social Reality, John Searle published a sustained riposte. In it, he claims that it is possible to distinguish brute and institutional facts. Institutional facts are those facts that are facts only through collective intentionality. According to Searle, these facts are only possible in light of `brute' facts--i.e. in light of some reality independent of all human agency. The appropriate analysis of institutional reality--of the socially constructed world--reveals that there are things which cannot be so constructed; that there are things which must exist independently of human intentionality. In what follows, I argue that Searle conflates two separate theses in his defense of External Realism, and that this leads him to beg the question against the anti-realist. One can acknowledge Searle's ER while also holding that Searle inverts the relation of dependence between brute and institutional facts. To put the response succinctly: we can call some fact `brute' only because of entrenched institutional facts. The very existence of `brute facts' can itself be regarded as an institutional artifact.

Searle claims that institutional facts have the logical form of constitutive rules (`x counts as y in context c'). The existence of money, to take one example, is analyzable in terms of a rule, or set of rules, which describe what money is. Thus, we might say that `this piece of green paper (x) counts as money (y) when it is made in the appropriate way, by the appropriate agency, and so on (in c).' As is obvious, this analysis itself contains further institutional facts: `a piece of paper' is not part of the furniture of the universe. Rather, it is the product of human needs and purposes. Thus, as Searle rightly acknowledges, a complete analysis of any institutional fact via constitutive rules is likely to yield iterations of rules: `this material counts as a piece of paper when it is used in thus and such a way' could thus take the place of the original x term, making the constitutive rule have the following structure: (x counts as y in c) counts as y in c.Foot note 1_1 At the end of such iterations, Searle contends, institutional facts bottom out in brute reality: there is some brute `x' upon which we impose some status `y.' Without an x upon which to impose some status function, there would be nothing out of which social facts might be constructed.

One problem with this view (and there are many) is that it begs the questions against the anti-realist. Searle has simply helped himself to a realism about brute facts in order to establish his account of social ontology.Foot note 1_2 Without establishing that there are objects independent of our systems of descriptions--and hence brute--it is open to the anti-foundationalist to contend that there simply are no brute facts.

At the end of Construction, Searle grants that it is time to pay his argumentative dues: he claims that he will provide an argument for a view he calls `External Realism' (hereafter, `ER'). The argument given is transcendental: it aims to articulate the presuppositions of some x which has been taken for granted (in this case, language-use). There are two things I want to claim regarding Searle's argument: 1) it works, and 2) it is irrelevant to determining whether or not there are brute facts. To do this, it will be useful to spend a moment considering the nature of transcendental arguments. This will allow us to see what the status of Searle's external realism is, and also why it is irrelevant to the discussion of brute facts.

There are several more or less loose ways of articulating the significance of the results of transcendental arguments. We can call the results of such arguments constitutive rules of experience, or discoverable analytic statements, or necessary and sufficient conditions of experience, or propositionally expressed pieces of a world-picture, or, finally, with the old wisdom, the conditions for the possibility of some x. All of these are but casting shadows: rather vague ways of spelling out that at which a transcendental argument aims--which is, to state it oversimply, what must be taken for granted in the employment of a concept.

This expression of transcendental arguments, however, begs for a prepositional phrase: must be taken for granted for what purpose? To state the answer boldly: transcendental arguments yield intelligibility conditions; they express what makes certain concepts we possess intelligible. A transcendental argument tells us what we must take for granted if some x is to remain intelligible. Because transcendental arguments begin with a concept already in our conceptual panoply, the result of a transcendental argument cannot be epistemic gain: we do not learn anything about the furniture of the universe from intelligibility conditions; rather, we clarify our conceptual situation.

To offer a quick example: the solipsist who articulates her philosophical view is subject to transcendental refutation in the following way. One first shows that the meanings of particular terms in a language are public. The notion of each word having a meaning for each individual singularly is, the argument would go, preposterous. This means that, in order for the solipsist to express her view, she has to rely on the meanings of the terms she uses. Because she relies on these meanings, she has already conceded that there is a `public'--namely, other language-users with whom she might speak. But this means precisely that, in order to say that I am the only thing that exists, I have to presuppose that I am not the only thing that exists. Solipsism turns out to be a performative contradiction on this view. The articulation of this position ignores a condition for the possibility of meaningful assertion.Foot note 1_3

At the end of The Construction of Social Reality, John Searle offers an argument against a caricature of anti-foundationalists.Foot note 1_4 He (rightly) claims that the argument he invokes is a transcendental one. It runs as follows:

1. The normal understanding of utterances in a public language requires that the utterances be understandable in the same way by any competent speaker and hearer of the language.

2. A large class of utterances purport to make reference to phenomena that exist outside of, and independently of, the speaker, the hearer, and their representations, and indeed, in some cases, independently of all representations.

3. Features 1 and 2 require that we understand the utterances of many of these sentences as having truth conditions that are independent of our representations. By purporting to make reference to public phenomena, phenomena that are ontologically and not merely epistemically objective, we presuppose that the truth or falsity of the statements is fixed by how the world is, independently of how we represent it.

4. But that presupposition amounts to the claim that there is a way things are that is independent of our representations, and that claim is just (one version of) external realism (CSR, 188).

The point of this argument is to show that even the phenomenal idealist and the social constructionist, to use Searle's terminology, must accept some (rather weak) version of external realism: namely, the view that there is something over and above our systems of categorization which gets classified. The method of the argument is straightforwardly transcendental: «We assume that a certain condition holds, and then try to show the presuppositions of that condition,» (CSR, 183). The condition which Searle takes for granted, and for which he owes thanks to Wittgenstein, is that successful communication does occur: whatever the problems of realism, we communicate with one another, and we make claims with truth values about things which are independent of us--which are in the world--whatever the status of these claims turns out to be. On Searle's view, in order for such a practice to be intelligible--in order for communication to occur when we describe objects (for example) in the world around us--we must take for granted that there is a way that the world is in virtue of which what we say can be either true or false.

But the crucial point here is not about the claims we make, but about the practices we share when making such claims. To understand what you are saying when you say, e.g., `My dog has fleas,' I must presuppose that your words are intelligible to me--that they are not merely private representations. If I could not be sure that you were talking about a thing (a dog) which existed in the world, and which I understood in the same way (or more or less the same way) as you do, I could not understand your utterance. But we do understand each other, and this suggests that we are presupposing that there is a world out there to which certain utterances refer, and that this world is independent of our private representations of the world:

Normal understanding requires sameness of understanding by both speaker and hearer, and sameness of understanding in these cases requires that utterances of the referring expressions purport to make reference to publicly accessible reality, to a reality that is ontologically objective. But the condition on public accessibility to the sorts of phenomena in these examples [e.g. `My dog has fleas,' `Hydrogen atoms each contain one electron,' `Mt. Everest has snow and ice near the summit'] is that the way that things are does not depend on your or my representations (CRS, 186).

I find this argument compelling, but do not think anything Searle says is incompatible with, say, Foucault's philosophical approach--or with Berger's sociology of knowledge, for that matter. What I am interested in is the status of the argument Searle has given. He claims to have shown that all of us, insofar as we communicate meaningfully with one another, must take some version of external realism for granted (namely, that the world exists independently of our representations). To put the point in the idiom of the old wisdom, external realism is a condition for the possibility of (much of) our language. The existence of a world that is independent of our representations of the world is that which makes intelligible a large class of utterances we make, as well as the fact that we understand one another in the normal course of things. But what has the argument shown? Seeing the limits of the scope of transcendental arguments is actually a virtue of Searle's account:

There is nothing epistemic about the arguments. I am not saying that in order to know the truth of our claims we have to presuppose realism. My argument is completely independent of questions of knowledge or even of truth. On my account, falsehood stands as much in need of the real world as does truth. The claim, to repeat, is about conditions of intelligibility, not about conditions of knowledge (CSR, 195).

So it is that transcendental arguments do not constitute epistemic gain, on Searle's view. Searle has not shown that the solipsist is wrong about the nature of reality. In fact, he has shown nothing whatsoever about reality, and hence the argument has not improved our epistemic position via revealing the truth of some set of statements. The point of Searle's argument, and of transcendental arguments generally, is to clarify what is involved in some feature of human experience--in this case, the meaningfulness of particular sorts of language-use. The irony of Searle's conclusion is that the `ism' attached to `real' is no longer an ism at all: «External Realism is thus not a thesis nor an hypothesis but the condition of having certain sorts of theses or hypotheses,» (CSR, 178).

Our question thus becomes: does the transcendental argument used to show that external realism is a necessary postulate also support the existence of brute facts? The answer, it seems to me, is necessarily negative. The claims made for ER are non-referential--that is, one can advocate Searle's ER without thinking that any particular claim actually refers to any particular object (or group of objects) in the world. We necessarily take for granted that our language refers to some observer-independent reality, but we are in no position to say of any particular utterance that it does in fact refer to this Welt an sich. To infer from the fact that our entire language must presuppose reference that a particular utterance refers to the world is simply fallacious. Indeed, this is the basic error we train undergraduates to recognize under the heading of `the fallacy of division'--attributing a predicate that is true of a set to particular members within that set.

To put this point another way: the claim that language is intrinsically referential is compatible with the claim that there are no brute facts. For this reason, one cannot infer the existence of such facts from the referential role of our language. The non-existence of brute facts and an intrinsically referential language are compatible, quite simply, because language is fundamentally a social enterprise. We live in a world populated by social objects. It is crucial that we are able to refer to these objects--indeed, if we could not, socialization would be impossible. But it does not follow from the fact that we refer to things that we take to be unchangeable that those things are in fact unchangeable. The fact that we regard some facts as brute does not make them so.

I now want to establish, in a bit more detail, why Searle's argument for ER cannot establish any of the brute facts he postulates. I will confess, at the outset, that the failure of Searle's argument seems to me so obvious that I have some hesitation in articulating it--the same hesitation one has whenever one is put in a position to articulate the obvious. Nevertheless, while there has been much criticism of the apparatus Searle employs in Construction, there has been little comment on the argument for ER, and why it fails to justify brute facts.Foot note 1_5

Having said that, I want to make clear that the failure of the argument is not a failure to show anything whatsoever. Searle's argument, as I boldly claimed above, works. The problem is that what Searle establishes as `external realism' has nothing at all to do with the brute facts he employs throughout Construction.

Searle claims that institutional facts depend on brute facts. There are two different theses here. The first--and the one Searle utilizes in the beginning of Construction--is the view that there are facts that are language-independent; that there are propositions, representing states-of-affairs that would still exist even if the proposition expressing them did not. As examples, we get «My dog has fleas,» «There is snow on top of Mt. Everest,» and so on. The second thesis--and the one which we find at the end of Construction--is that the formulation `x counts as y in context c' depends on their being an x upon which agents can impose a status function. In brief, institutional facts require a world of non-institutional items that get understood in a particular manner. This, in turn, is a condition for the possibility of language-use.

By conflating these two theses, Searle effectively begs the question against the anti-realist. The second version of ER is compatible with any anti-realist position, as it makes absolutely no substantive claims about the external world. Indeed, it seems simply to be a reproduction of the (Kantian) view that we necessarily presuppose a world independent of our representations, even if we can say nothing at all about this world. The problem with Searle, of course, is that throughout Construction he never tires of making claims about the content of the `real' world (invoking the first thesis articulated above). For the sake of clarity, let's distinguish the two theses as follows:

ER₁ The world consists of states-of-affairs, accurately describable in our current language, which do not depend on human agency in any way.

ER₂ In order to have a language, we must presuppose that there is a world, independent of human representations, which is capable of being conceptualized in myriad ways.

It is clear that these views are not necessarily incompatible, but it is equally clear that one might accept ER₂ while denying ER₁. Indeed, the plausibility of ER₂ for the anti-realist will depend on it being distinct from ER₁ (i.e. on its being without content). The view that there are actual, articulable brute facts robs ER₂ of its plausibility for the anti-realist. It does so precisely because it begs the question against the anti-realist.

Consider: You claim that light is best explained as a wave. I retort that it is best explained as a particle. Someone else claims we should adopt the notion of a `wavicle' in order to explicate light. We can all agree, regardless of our philosophical bearings, that there is an explanadum here--a set of data that we are trying to make sense of. Our disputes arise precisely because of the underdetermination of theory by data. No one need here deny that there is a world, independent of all of our representations, which we are trying to understand. This is what lurks at the core of ER₂.

Disagreements arise between realist and anti-realist when we begin to assess divergent recommendations for understanding the world in a particular way. Whereas the realist wants to claim, in line with ER₁, that there is one correct articulation of how things stand, the anti-realist doubts that there is some such definitive articulation. The anti-realist is sensitive to the fact that our descriptive vocabularies emerge in historical contexts, that our theories answer contingent questions, and that science and politics are never mutually exclusive. He thus doubts the truth of ER₁. He doubts that the states-of affairs that comprise the world are neatly articulable in a way that is independent of the circumstances (historical and otherwise) of the speaker.

So, as it stands, ER₂, which Searle establishes via transcendental argumentation, is accurate. It is similarly pointless, however, as no one denies it. ER₁, however, is a substantive view about the content of our ontological inventory. If Searle had shown that ER₁ was an accurate view of things, he would indeed have refuted the antirealist. But he has shown no such thing. ER₁ does not follow from ER₂; one cannot infer brute facts from the necessary postulation of a Kantian noumenal realm. If one could, Kant would never have needed to engage in his own Copernican Revolution, and the world would have managed to avoid Hegel.

To see precisely how egregious this philosophical maneuver is, consider the old adage that all facts are theory-laden. When Goodman claims that there are `ways of worldmaking,' he need not deny that we presuppose some world external to our representation. In fact, he doesn't deny this--he simply points out that there are `versions' of this world which correspond to different systems of description. To offer any description of the world, then, is to `make' the world into more than a dull substratum that we must presuppose in our everyday linguistic interaction. Of course, the systems of description we employ are not created ex nihilo. As Goodman puts it:

We start, on any occasion, with some old version or world that we have on hand and that we are stuck with until we have the determination and skill to remake it into a new one. Some of the felt stubbornness of fact is the grip of habit: our firm foundation is indeed stolid. Worldmaking begins with one version and ends with another.Foot note 1_6

What Goodman is here intimating, and what we know well from Hanson, Kuhn, Cassirer, and Foucault (among others), is that systems of description, like predicates, can become entrenched in institutional life. It is precisely this entrenchment that gives particular utterances the appearance of rigidity, and that allows us to solve skeptical riddles like `Grue.' Indeed, it is precisely the entrenchment of systems of description that makes sense of the term `paradigm' in Kuhnian philosophy of science.

But this suggests that ER1 is simply false, as the content of any so-called brute fact will depend on what system of description is the dominant one--on what predicates are entrenched in institutional life. Indeed, it is a merit of Berger's sociology of knowledge that it can explain why some facts appear to be `brute.' As Berger and Luckmann explain:

Reification is the apprehension of human phenomena as if they were things...The reified world is, by definition, a dehumanized world. It is experienced by man as a strange facticity, an opus alienum over which he has no control rather than as the opus proprium of his own productive activity (89).

This `reification' occurs as much with recognizably social roles as it does with the institution of scientific practice. Because we have been socialized through a language, the basic elements of this language are seen as bits of a grand ontological inventory--as the brute pieces of an external world which we did not create. While this is a perfectly acceptable (and perhaps even necessary) product of socialization, it is nevertheless not the occasion to engage in high-church metaphysics. Collectively regarding some proposition as articulating a brute fact is as much a social phenomenon as is the buying and selling of goods with green paper.

While Searle is right that institutional facts are based on ER₂ (i.e. on the presupposition of an external world), he fails to see that the relationship can be regarded as exactly reversed for ER₁: substantive assertoric claims about the world are based on institutional facts about the meanings of terms. The assertoric speech act of the form `x is the case' is based on a more primitive phenomenological fact: that we acknowledge constitutive rules of the form `x counts as y in c.'

Consider the following case: I make the non-trivial assertoric claim that `The morning star is the evening star.' This is a claim about the empirical world--it asserts that a certain state-of-affairs is the case. It can be verified or falsified. In brief, it has all the flags of an assertoric speech act. Now, notice that this assertion depends on the two terms in the identifications having some significance. After all, if the claim can be verified or falsified, it must be the case that the claim can be understood. If I were to claim that `Degroner is in fact garap,' no one could possibly verify what I had said, as no one could understand it. To make substantive claims about empirical reality requires that this reality is understood in approximately the same way by our interlocutors. But this does not show that any sort of robust realism is the case. In fact, it suggests rather the opposite: our assertoric claims about external reality depend crucially on our antecedent understanding of this reality. In this way, we might here speak of the primacy of phenomenology. Before I can make any substantive claims about the nature of some x, it must be the case that we collectively understand this x in a particular manner--that it is recognizable to us as an object of predication.

Given this line of argument, Searle can be accused of inverting the relation between constitutive rules and brute facts. Our everyday understanding of the world is littered with constitutive rules, and such rules are a condition for the possibility of making assertoric claims about empirical reality. Language itself suggests that `brute facts' only have their basis in institutional reality. If language is constituted by constitutive rules which are the product of collective intentionality (as Searle claims), then what will count as the `correct' description of extant phenomena will also depend on these rules. Using Nelson Goodman's terms, we could put the point in an epigram: the more entrenched the predicate, the more brute the fact. On this view, the analysis of institutional facts will not bottom out in brute facts, as brute facts can be shown to rely on institutional facts.

The above characterization of the relation between brute and institutional facts does not establish that the anti-realist position is a tenable one. It merely establishes that Searle's argument for ER has not shown that we must abandon anti-realism. Searle's social ontology does not establish that there are brute facts. Indeed, as I have hopefully made clear, that same ontology can be used to explain why we regard particular facts as brute, even when we should not. Given two divergent characterizations of the basis of brute facts, and little way to determine which is the more accurate, the lesson to be learned here might be that disputes about realism and antirealism ought to be abandoned--at least in our discussion of social ontology.

Peter Berger and Thomas Luckman, The Social Construction of Reality (Garden City: Doubleday, 1966).

Nelson Goodman, Ways of Worldmaking (New York: Hackett Publishing Company, 1978).

John Hund, «Searle's The Construction of Social Reality,» Philosophy of the Social Sciences, 28(1), 1998, 122-131.

John Searle, The Construction of Social Reality (New York: The Free Press, 1995).

[Foot Note 1_1]

It would also be necessary to provide some such rule for green, as well as any other social facts which appeared in the iterated rule.

[Foot Note 1_2]

Moreover, it isn't even clear that the analysis Searle offers requires realism about facts. This will become clear in what follows.

[Foot Note 1_3]

A silent solipsist, one might say, is the best kind. The above argument does not show that solipsism is wrong. It only shows that one cannot coherently express this position.

[Foot Note 1_4]

It is a caricature precisely because any anti-foundationalist could accept the (minimal) claims Searle defends at length in that text.

[Foot Note 1_5]

One notable exception is Hund. See op. cit.

[Foot Note 1_6]

P. 97, Ways of Worldmaking (New York: Hackett Publishing Company, 1978).