False to Facts

We have grown accustomed to a philosophical use of the word “fact”, mainly from the writings of Wittgenstein and Russell in their logical atomism period. Roughly, this is the idea of a fact as a “combination of objects”—a sort of complex of objects, properties, and relations. This conception is what allows Wittgenstein to write, “The world is the totality of facts, not things” and “The world divides into facts”. Call this an “ontological use” of “fact”: facts are entities belonging to ontology in that they are constituents of reality existing independently of human knowledge. This use has become entrenched, even though there has always been difficulty explaining what a fact in this sense is (it’s certainly not a “combination of objects”). Some have thought that facts are just true propositions: this is still an ontological use (allegedly) though it rejects the idea of a separate realm of facts alongside propositions. The concept is then used in a variety of areas to formulate philosophical doctrines, or to reject them (see below). But it is not at all clear that the notion can bear this kind of weight, or that it is even coherent. Is there an ontological sense to the word “fact”? What does the word really mean? Is this what it means in ordinary speech? As always, the dictionary provides a salutary starting point: the Concise OED gives us “a thing that is indisputably the case” and “information used as evidence or as part of a report”; the Shorter OED has “A thing known for certain to have occurred or known to be true; a datum of experience”. These definitions are clearly epistemic not ontological: a fact is something indisputable, evidential, certainly known, a datum of experience. In this sense a fact is opposed to an interpretation, a theory, a hypothesis—not a falsehood. Interpretations, theories, and hypotheses can be true and yet not facts, i.e., things indisputable, evidential, datum-like. If we ask “What are the facts of the case?” we mean to be asking after the evidence, the known information—not the truth-value of theories about the case. It would be quite wrong to say that the world is the totality of facts in this sense, since reality contains more than what we indisputably know. We could say that the world is the totality of objects and their properties and relations, but that goes well beyond what is known to have occurred etc. In the dictionary definition, then, logical atomism is simply false, even bizarre. So, is there another sense? Or has the word been snatched from ordinary language and stretched beyond endurance, giving an illusion of intelligibility? I fear the latter is the case. A fact is a truth that is known, to put it simply—and nothing else. That is why all attempts to explain the ontological sense have failed miserably: facts are not a type a type of entity; they are just truths that stand in a certain relation to knowledge. We could say they are known truths, known events, known states of affairs, known instances of objects having properties. But they are not a category of being that exists independently of any reference to knowledge: that is a philosophical fiction. True, we sometimes speak of “unknown facts”, but this odd locution appears to mean either “unknown to a certain person or group” or “unknown truth”. In general, talk of facts is always talk of what is known or perceived or otherwise registered by the mind—as opposed to what is objectively real. The theory of evolution, say, was not a fact until it was placed beyond serious doubt, not when it was merely a theory awaiting confirmation (though perfectly true then). The words “fact” and “true” are not coextensive or correlative, let alone synonymous. One way to see this is to note that “true” is redundant in a way that “fact” is not: compare “The heliocentric theory of the universe is true” and “The heliocentric theory of the universe is a fact”, said at the time of Copernicus; the latter is not equivalent to “The sun is the center of the universe” but the former is. To state a fact, a proposition has to be both true and beyond dispute. It is distinctly odd to say “The facts might never be known by anyone” (for how can facts be so hidden?), but perfectly okay to say “The truth might never be known by anyone”. The former is like saying “The evidence might never be known by anyone”—or the information, data, signs, clues, and indications. For all these words suggest availability to knowledge, whereas “true” does not. And to say that someone believes it’s a fact that p is not at all like saying that someone believes that p. What has happened is that philosophers have slid from the legitimate epistemic use of “fact” to an ill-defined ontological use.[1] This slide has a bearing on three philosophical issues in which the word “fact” has been theoretically deployed. The first is the correspondence theory of truth: a statement is true if and only if it corresponds to the facts. But this has to be wrong given that facts are always known, since there are many truths that are not known. And what might true statements correspond to if not to facts? If facts won’t do, there is nothing for truths (whole propositions) to correspond to, and hence there is no correspondence. The theory sounds plausible when we know what the facts are (e.g., the fact that snow is white), but it founders when we don’t know. Second, people ask whether ethical statements are “fact-stating” as a way of asking whether ethics is concerned with objective truth. But if we hear the phrase “fact-stating” as implying “indisputably the case”, we will be biased against ethical objectivism, because ethics is an area of frequent dispute. What is really intended is the idea that ethical propositions state objective truths (whether disputed or not) not whether they say things that no one will dispute. Many kinds of statements can be objectively true without stating facts in the dictionary sense, so the “fact-stating” formulation is wide of the mark. Third, the positivists tended to view meaningfulness as equivalent to being fact-stating, but this makes verifiability sound like a decent criterion for meaningfulness only if you take “fact” in the ontological sense. Otherwise, you are merely saying that a statement states something known only if it is verifiable—a straight tautology. But if you interpret “fact” in the ontological sense, it looks as if failure to state a fact really is tantamount to lack of meaning. Once it is admitted that “fact” can only mean something known it looks hopelessly procrustean to insist that all meaningful statements must be verifiable. Tying meaning to facts in the (alleged) ontological sense doesn’t seem so arbitrarily exclusive, but tying meaning to facts in the epistemic sense looks preposterously narrow—the only meaningful statements will be ones already verified. So, the concept of a fact can lead us astray once we try to detach it from the sense it ordinarily has. It should really be banned from philosophical discourse when used in the way initiated by Wittgenstein and Russell. The only facts there are are facts that are known to be so—and they are only facts in relation to what is known. It is a kind of category mistake to speak of facts as if they are things that can exist independently of human knowledge, as objects, events, and properties can. Facts don’t constitute the world; they are what we indisputably know of the world, as opposed to speculate about or are ignorant of. For the description of reality, we must stick with the concepts of truth, proposition, object and property, event and process; there are no facts out there in addition to these.[2]

[1] One symptom of this is the problem of individuating facts: when are facts the same and when different? Does “The morning star is bright” report the same fact as “The evening star is bright”? Also: can facts be perceived (seen, touched)? Can they be referred to by singular terms? Can they cause anything? Are there general facts, negative facts, disjunctive facts? Do they weigh anything (objects do)? Do they have the same constituents as propositions? Are they denoted by sentences? Are there fictional facts? Are there possible facts or are all facts actual? Can facts be quantified over? Are there facts about facts? Were there facts before the world was created, such as the fact that the world does not yet exist?

[2] Use of the word “fact” in an ontological sense encourages idealism, because it suggests the idea, always tempting, that unknown reality is necessarily like known reality. We know what known facts are like and then we project this conception onto reality as a whole, but then we are thinking of the unknown by analogy with the known. If the world consists of facts and facts are epistemically defined, then so is the world—hence idealism. The cure is not to contrive a non-epistemic notion of fact (a hopeless task) but to abandon the idea that reality consists of facts many of which transcend our knowledge. Reality does not consist of data, information, evidence, etc.

An A Priori Order

In his Private Notebooks 1914-1916 Wittgenstein writes: “The great question around which everything I write revolves is: Is there an a priori order in the world, and if so, of what does it consist?” The question is a good one (and highly metaphysical). In the Tractatus we read what looks like a reply: “There is no a priori order of things”. (5.634) This appears to imply that the order of things is entirely a posteriori. What we know of the world is derived exclusively from experience; otherwise, our knowledge of the world is a total blank. How this is consistent with the opening propositions of the Tractatus is not explained: “The world is all that is the case” (1), “The world is the totality of facts, not of things” (1.1.), “The world is determined by the facts, and by their being all the facts” (1.11), and so on. For these propositions are presumably intended as a priori necessary truths: it isn’t that we merely empirically discover that the world is all that is the case (etc.), as we discover that it contains hydrogen and walruses; or that it is just contingent that the world is made up of facts, as it is contingent that it contains London and denim jeans. No, the idea is presumably that these are a priorinecessities arrived at by pure reason (perhaps they are even analytic truths, if that’s the way you think of necessity and the a priori). Wittgenstein doesn’t say as much, but it is a natural interpretation of his words. In any case, I wish to discuss Wittgenstein’s question in its own right, not issues of Wittgenstein exegesis. So: is there an a priori order in the world, in things, in reality? Are there things we can know about the world without examining it, without observing it, without living in it? Notice that we are not asking whether there is an a prioriorder in propositions or sentences or in mathematics; we are talking about the world as a whole, including (and especially) what might be called concrete reality—material bodies, biological organisms, the weather. We can readily allow that an a priori order exists in logic, language, and number theory (I don’t think Wittgenstein would disagree—hence his use of “world” and “things”), but it is another question whether an a priori order might hold in the ordinary world of physical and psychological things. The question concerns whether there is such an order in all of reality, including the things we know by ordinary sensory perception. It is surely clear that much of the world is not knowable a priori: most of we know of the world is known empirically. But is any of it knowable (or actually known) a priori, and here we include the world of empirical concrete objects? Imagine standing outside the world, having had no experience of the world: is there anything you can still know about its make-up? We can start with Wittgenstein’s own proclamations: can we know a priori that the world is everything that is the case, the totality of facts not things? These words are not pellucid, despite their resonance. Does the first sentence not mean “everything that is true”, which must then mean “every proposition that is true”? But the world is not made of propositions. The second sentence speaks of a “totality of facts”: but what kind of totality (a set, an aggregate, a list?), and what is a fact? Wittgenstein goes on to introduce the idea of a “combination of objects” as constituting a fact, but that is a faulty conception of facts, and the notion of object here is highly theoretical and abstract. Still, the general drift of Wittgenstein’s words does not seem mistaken: we know a priori that the world must contain certain ontological categories, however difficult it may be to characterize them. We might venture the suggestion that the world consists of particulars and universals, objects and properties, individuals and general characteristics. We don’t know a priori what kinds of entity fall into these categories, but we do know a priori that the world (any world) must divide into them. Thus, there is an a priori order of things, of what there is, of reality. Admittedly, we have to be careful how we formulate the precise nature of what there must be: we don’t want to say that the existence of matter is known a priori, or persisting objects, or even space and time. The world might consist of immaterial stuff, or only of events, or be devoid of space and time: we can’t rule these epistemic possibilities out a priori. Empirically, we know these not to be the case, but they don’t seem like things we could exclude a priori. We can’t just excogitate the existence of matter, enduring objects, and space and time from the concept of a world. So, we should limit ourselves to only the most schematic description of the order we have discerned. I think the best and most convenient way to do this is to say that the world must contain instantiation: any world must have the property that instantiation occurs in it—there must be individual instances of general features or forms or characters or types. This we know a priori—though of course in fact we are confronted with it at every waking moment of our lives (and even in dreams). You can’t make a world without installing instantiation in it. The empty world in which nothing instantiates anything is not a world at all. Nor can particular things exist without instantiation, since every individual must have properties of some kind, as a matter of metaphysical necessity (no individuals without facts about them). But is that all we can know a priori about the world? Not quite, perhaps, because we can add logical laws if we accept that they apply to objects and properties and not just to propositions. We can say that no property can be instantiated and not instantiated in the same object at the same time, and that an object either instantiates a given property or it does not, and that objects and properties are always self-identical. So, any world obeys logical laws; and we know this a priori. We can also say that properties, being general, are shareable (even if not actually shared): this is in the nature of properties (hence the term “universals”). And if they are shareable there ought to be generalizations about them—laws of some sort. More adventurously, if instantiation must be present, then predication cannot be far behind: that is, if language and thought exist in the world, then they must contain predication, because instantiation is just the worldly counterpart to predication. We therefore know something about how the world must be represented in language and thought (if they exist in the world), given its ontological structure of instantiation. We can’t know a priori that they do exist in the world, but if they do they have to contain predication. Still, this is thin gruel compared to the vast number of things that are known a posteriori about the world: it is purely structural rather than substantial. But what is remarkable is that anything can be known a priori about the world: how can we know the form of the world without even looking at it? And isn’t it amazing that the world divides into an a prioripart and an a posteriori part? That is, each object partakes of an a priori order and an a posteriori order. Why should this be—why shouldn’t it all be a posteriori? Not all of an object’s nature has to be known by means of the senses, though most of it is. We have a kind of global cosmic rationalism: everything has a nature that is partially knowable by pure reason. Wittgenstein never asks whether there is an a posteriori order in the world, because it is obvious that there is, but it is not at all obvious that there is also an a priori order—and that these coexist. There is something startling about the fact that the world harbors an a priori order. We don’t learn from experience that instantiation (and associated structures) is a fact about the world; it just follows from being a world of any kind. No doubt this is a rather mysterious type of knowledge (not that a posteriori knowledge is entirely unmysterious), but it evidently exists and is philosophically important.[1]

[1] Wittgenstein was evidently a metaphysician in the grand style in his Tractatus days (ironically given his influence on the logical positivists), but nothing of this survives in his Investigations years. I have always thought this is a pity because he has a genuine talent for metaphysics. It is really not possible to be interested in logic without straying into metaphysical territory. The Investigations is pretty much devoid of logic.

Linguistic Phenomenology

J.L. Austin described his method as “linguistic phenomenology”. It is highly likely that this is an allusion to Husserl’s phenomenology: Husserl’s work was well known in Oxford in Austin’s time and Gilbert Ryle had a special interest in Husserl (he was a colleague of Austin’s). A cheeky allusion, perhaps, but an allusion nonetheless. However, there is no evidence that I know of that Austin ever studied Husserl, and if he did it made no discernible impact on him. Is Austin’s linguistic phenomenology at all like Husserl’s phenomenology? Do any of Husserl’s concepts show up in Austin’s work? Is Austin’s linguistic phenomenology phenomenological? Husserl’s phenomenology emphasizes consciousness, the first-person perspective, intentionality, noesis and noema, the epoche, suspension of the natural standpoint, and other concepts of the phenomenological tradition (going back to Descartes, Hume, Brentano, and others).[1] No trace of this is to be found in Austin’s writings; he is not doing the phenomenology of language at all, i.e., the way language manifests itself in consciousness. The notion of consciousness does not appear in Austin’s work on language. Nor does the notion of intentionality. There is no suspending of the natural standpoint in order to focus on the contents of consciousness as such. What Austin appears to mean by “phenomenology” is simply careful description as opposed to premature theory and spurious explanation, which is not a natural or normal interpretation of that word. Austin is surveying the phenomena of language but not doing the phenomenology of language. One would think that he might, after appropriating Husserl’s use of the word, make a point of investigating what Husserl has to say, but apparently, he did not (or it didn’t have any impact). In point of fact, Husserl does a good deal of linguistic phenomenology in his sense, investigating meanings, predication, propositions, indexicality, and other linguistic phenomena. But none of this seems to have influenced Austin’s work. Pity: it would have been interesting if Austin had produced a comparative study of himself and Husserl; and it might have enriched his own efforts. Actually, there is really no phenomenology in the analytic tradition devoted to language: what we call philosophy of language includes no phenomenology of language in Husserl’s sense. Frege discusses thought, but his philosophy of language makes no reference to consciousness, or even intentionality. Russell talks a lot about sense-data, but he never tries to describe what it is like to speak or understand speech. Wittgenstein is all logic and no consciousness in the Tractatus; and the Investigations, though avowedly descriptive, never examines language from the first-person point of view, preferring to look at it from the outside. Nor do we find any phenomenology in Carnap, Quine, Davidson, Strawson, Kripke, Montague, Searle, Kaplan, and numerous others. You might think that for this tradition meaning and consciousness have nothing to do with each other.[2] It is noteworthy that Husserl studied with Wundt as well as Brentano: so, he was steeped in introspectionist methodology, as opposed to the behaviorism that influenced later thinkers. In any case, there is no such thing as linguistic phenomenology in the analytic tradition. And this despite the fact that Husserl intended to found a school of philosophy that could claim to be a rigorous science (he was also a mathematician and physicist)—as analytic philosophy in some of its forms also aspired to be. What Husserl called “transcendental phenomenology” is not that different methodologically from the school of thought known as “ordinary language philosophy”, in that both seek to study the human creature as abstracted from the world of natural science: that is, consciousness as it is in itself independently of the natural world and language as a system of communication divorced from the reality outside of it—consciousness as such and speech as such. Ordinary language philosophy also had its epoche. Even speech act theory, which stresses the psychological aspects of speech behavior (Grice in particular), avoided consciousness and its inherent structures; so, real phenomenology never found a place in analytic philosophy of language. It is also interesting that what we might call “linguistic existentialism” never took off (except perhaps subliminally): I mean the application of concepts drawn from existentialism to speech. Sartre’s views derive from Husserl and deliver a distinctive approach to human existence, which might be carried over to language. Thus, is there any room for the contrast between Being and Nothingness and the idea of radical freedom in the philosophy of language? Is meaning an example of the intentionality of consciousness with its negative essence? Certainly, a Sartrean view of intentionality is opposed to a Fregean theory of sense and reference, since the conscious mental act has no nature not conferred by its reference—its essence is nothingness (compare the idea of direct reference, which also dispenses with mediating senses). And the idea that speech acts are instances of free action comports well with certain strains of thought concerning the nature of human action in general (I am thinking of Chomsky’s views on linguistic performance). So, we can see parallels between ideas found in existentialism and ideas familiar from the (so-called) analytic tradition. Sartre’s conception of consciousness as nothingness is redolent of Mill’s view of names (also Russell and Kripke), and his view of action as radically unconstrained is like Chomsky’s view of performance (and other libertarian views familiar from the analytic tradition). Accordingly, we can envisage linguistic existentialism as we can envisage linguistic phenomenology; and both have their counterparts in strands of the intellectual tradition more familiar to most of us (the “analytic”).[3] Austin could easily have become a phenomenologist and existentialist! He just needed to open up about consciousness and read some Husserl and Sartre. I suspect it was a deep-seated hostility to Descartes in the English-speaking world that inhibited that delightful rapprochement. Might it be because Descartes was a Frenchman, est-ce possible?[4]

[1] In case readers are wondering, my first published paper was on Mach and Husserl (1972).

[2] In all the literature on proper names and definite descriptions is there anything on the phenomenology of using names and descriptions? Don’t different kinds of meaning inhabit consciousness differently?

[3] I don’t believe there is any serious intellectual distinction between what are called “analytic” and “Continental” philosophy. The difference is purely geographical and linguistic (i.e., what language the philosophy is written in).

[4] I’m joking, of course, but I do find the hostility to Descartes hard to understand. Ryle had it in spades, and Wittgenstein nourished it, and Quine embodied it.  There is a clear line leading from Descartes through Locke, Berkeley and Hume, up through Russell, Brentano, and Husserl; but the twentieth century saw a determined effort to expunge his legacy from the canon. It’s quite puzzling but should not be underestimated. I do wonder whether, if he were an Englishman, the hostility would be quite so marked. No Englishman wants to be labeled a Cartesian. Something similar is true of Sartre.

Data in Philosophy

Every academic subject requires a source of data. Without data a subject cannot thrive, survive, or even exist. History requires written documents of the past. Archeology requires preserved artifacts. Microbiology requires data from microscopes. Anatomy requires dissections. Atomic physics requires data from supercolliders. Astronomy requires light readings from telescopes. Zoology requires observations from the field. Psychology requires behavioral data from experiments. Chemistry requires records of chemical reactions. Paleontology requires fossils. Certain sorts of data fit the subject in question and allow it to develop and grow. Practitioners have to learn how to collect and interpret data. Theories have to be tailored to data. Some potential subjects don’t exist because there is no data to support and nourish them—there is no celestial zoology or anthropology. Data is a sine qua non. But what kind of data is suitable for philosophy? It can’t be without data of any kind, because then it would not be at all. Here matters turn controversial: it is a contentious question. Some say that the data are human experiences introspectively apprehended (phenomenology). Some say the data are intuitions about concepts (intuitionism). Some say the data are the results of thought experiments (conceptual analysts). Some say the data are the results of the several sciences (scientism). Some say the data are surveys of people’s opinions (experimental philosophy). Some say the data derive from the study of ordinary language (the linguistic turn). Some have said (though rarely today) that religious revelation affords the data of philosophical reflection (fundamentalism). Some (especially today) have even insisted that the data of the brain are the appropriate data for philosophy (neuro-philosophy). There is no consensus on what kind of data are right for philosophy; indeed, there is vigorous opposition to some of these schools. Some people prefer a pluralist perspective (I include myself in this category), looking for data anywhere they can find it (though I draw the line at religion and brain science). Psychology is the closest analogue to philosophy: since its inception as a science debate has raged as to what the right source of data should be—introspective data, behavioral data, cerebral data, computer data. This is why psychology has never been able to free itself of the taint of philosophy (its progenitor and prototype). But the real problem in philosophy is not lack of consensus about what data are kosher; it is that no data (so called) really add up to data—a solid foundation with a clear connection to philosophical theories. The problem is data paucity, data inadequacy, data ineptitude. In the case of astronomy, we are lucky that light from the distant universe reaches planet Earth relatively unimpeded (but those pesky clouds!), because we have little else to go on, and light data is quite impoverished. Imagine if we had to rely on sounds from outer space, or the occasional visit from a meteor, or sensations of heat: astronomy would hardly be possible. It would exist in a primitive state, barely making any progress, sorely lacking in consensus—a bit like philosophy, in fact. We want to know the answers to philosophical questions—the problem of free will, the mind-body problem, the objectivity or otherwise of ethics—but our data base provides little guidance. We examine our language, our inner experience, the relevant sciences, the results of thought experiments—and we just don’t make much progress. The data underdetermine the theories; worse, they don’t seem to make proper contact with the theories. J.L. Austin thought that if he just assembled enough data from the use of ordinary language the proper theory would reveal itself, but that hope was forlorn—the correct theory remained elusive. Nor does knowing all the science help much. Et cetera. We are like astronomers reduced to listening and feeling. In short, the data of philosophy are a pretty sorry bunch when it comes to solving the problems that they are wheeled in to solve (they may be rich and fascinating in themselves). They can never be cited as settling an issue, as the data of the sciences often are. The putative data are hardly dataat all, only hints or clues or bits of decoration. I don’t mean to denigrate philosophy in saying this—the questions are real and we have to go on what we have to go on—but the relation between data and theory is depressingly tenuous. We are operating in an evidential void, in effect. Nothing in our data base can decideanything, except perhaps some elementary conceptual questions (e.g., whether knowledge is true justified belief). Sure, we have arguments, some of them very good, and objections, some quite devastating, but we don’t have anything analogous to the data of other academic subjects. If we had no data at all, things would be a lot worse; it would be like trying to do extra-terrestrial zoology or artifact-free archeology. We do have something to go on—in fact, quite a lot—but it just isn’t all that probative, all that datum-like. It’s a bit like trying to learn about the sun just by feeling the warmth of its rays. Or not like that, because at least the sun’s rays are systematically connected to its actual nature. It’s more that the data—say, ordinary language—is disconnected from the nature of the thing that puzzles us—free will, consciousness, moral rightness. Words will only tell us so much about the things they refer to (imagine an ordinary language chemistry!). The alleged data don’t make close enough contact with the thing that puzzles us. Our subject matter seems out of reach of the data we marshal to shed light on it. What we want to do is gaze directly at the facts and see what their nature is, but that isn’t feasible, so we resort to remote “data”. We can’t tell what numbers are simply by looking at them, so we resort to indirect support for our hunches deriving from loosely associated phenomena—numerals, mental acts of calculation, feelings in the belly (the number 2 feels real). We are data deprived; data bereft. Our supposed data is hardly worthy of the name. The OED defines “datum” as “a piece of information” and “an assumption or premise from which inferences may be drawn”: but the data of philosophy don’t inform us of the philosophical truth, nor do they constitute premises from which philosophical truths can be deduced (or even induced). At best, they may be suggestive or corroborative, things we can brandish in philosophical debate, but they can’t be put into a graph or entered in a ledger as statistical facts raising the probability of a philosophical hypothesis (there is no philosophical t-test). That is why there is so much disagreement about the nature of philosophical data: it’s because nothing we have at our disposal constitutes data in the normal sense (if it did, we would all welcome it with open arms). Philosophy is not a data-driven science (or even a data-driven humanity like history or Russian literature). So, we hover in a strange no-man’s land methodologically: we are not data-independent (whatever that might be) but we are also not data-driven—we are only data-stimulated, data-inspired. We find various facts philosophically suggestive, but these facts don’t convert into something solidly datum-like. Hence the permanent possibility of philosophical ignorance (like the plight of the would-be astronomers deprived of light as a source of information). No data, no knowledge.[1]

[1] Philosophers don’t proceed by first presenting their data and then drawing inferences from it. They start with a theory that appeals to them and then hunt around for some data that seems to support it—from language, from science, from introspection. The data is never viewed as definitive or even probability-raising (in the statistical sense). Nor do they accumulate more data over time like a typical scientist, hoping to edge a theory into the acceptance zone. The role of data in philosophy is quite different, more like a seduction than a deduction—designed to entice not overwhelm. The social psychology of the use of data in philosophy would be worth studying.