If philosophy consists of conceptual analysis, is it thereby debarred from being a science? I argue that it is not and that philosophy so conceived is a science. The argument takes the form of careful attention to the meaning of “science,” “experiment,” “empirical,” and related words. Philosophy is a formal science. This does not mean it is not part of the humanities. The role of observation in other kinds of science is investigated. There is more methodological homogeneity in the various sciences, including philosophy, than has been recognized, despite some clear differences. Seeing this helps restore philosophy to its rightful place in the academic firmament.
Keywords: metaphilosophy, science, experiment, intuition, observation, concepts
The Science of Philosophy
What is the nature of philosophy? Two views have been influential. One view is that philosophy is “continuous with science”–a kind of proto science or a commentary on the sciences or a synthesis of them.According to this view, philosophy is an empirical discipline, though more removed from the data than typical science: it is not different in kind from physics, chemistry and biology. Thus the subject of philosophy comes under the general heading of “science” because of its methodological similarity to the received sciences. Historically, “philosophy” once contained the sciences, which eventually broke off from it, and it is still a kind of science-in-waiting–pupal science, as it were. The second view is that philosophy is quite unlike empirical science, both in methodology and subject matter: it is an a prioridiscipline, removed from observation and experiment. According to this view, philosophy is to be contrasted with empirical science, and is often regarded as properly one of the “humanities”. In its purest form, the second view takes philosophy to consist of conceptual analysis aimed at establishing a priorinecessary truths—the antithesis of empirical science. Thus philosophy is held not to be a branch of science, having its own distinctive nature as a field of enquiry.
I hew to the second view: philosophy is conceptual analysis (in a suitably broad sense). I won’t be defending that view here; I will presuppose it.My question is whether it is correct to withhold the designation “science” from philosophy as so conceived: is it consistent to hold that philosophy consists of conceptual analysis andthat it is a science? I shall argue that these are compatible propositions; and I shall further contend that philosophy is a science—indeed, that it can be rightly described as an empirical experimental natural science. These may seem like surprising claims, but actually they spring from obvious linguistic facts. Thus philosophy, in my view, consists of the a priorianalysis of concepts and it is alsoan empirical experimental natural science—with no tension between these two traits. Moreover, all this is trivially true, once we attend to the linguistic facts.
What is a science? Better: what does the word “science” mean? Here we naturally reach for the dictionary. Consulting the Concise Oxford English Dictionary(Eleventh Edition), we find two definitions: 1, “the intellectual and practical activity encompassing the systematic study of the structure and behaviour of the physical and natural world through observation and experiment”; 2, “a systematically organized body of knowledge on any subject”. The word comes from the Latin scientia“knowledge”, from scire“know”. For “scientific” we find: 1, “relating to or based on science”; 2, “systematic; methodical”. These definitions are worth careful study. The first definition of “science” would apply to the analysis of concepts, construed as constituents of the natural world, up till the phrase “through observation and experiment” (but see below where I point out that even this last word can apply to conceptual analysis). Presumably this definition is deemed by the dictionary editors to cover only part of the accepted meaning of the term, because it excludes what are traditionally called the “formal sciences”—arithmetic, geometry, mathematical logic, abstract computer science, game theory, information theory, general systems theory, and the like. These sciences are not empirical in the usual sense—they are not based on observation and experiment—but they are sufficiently rigorous, organized, and systematic to qualify as sciences. So it is not the status of a discipline as empiricalthat makes it a science—that adjective applies also to history and geography—but whether the subject achieves the right level of internal rigor and systematic organization.We can say right away, then, that philosophy will not be disqualified from falling into the category “science” simply because it is not empirical: “a prioriscience” is not a contradiction in terms, as “formal science” is not. Philosophy properly belongs with the formal sciences, not the empirical sciences, but it is no less a science for that (nor is it any less “scientific”).
But the dictionary will only take us so far here. We need to identify the marks of genuine science and see whether philosophy shares these marks—and we must not bias our discussion by favoring such sciences as physics and chemistry. I take it that what distinguishes a discourse as scientific are such traits as these: rigor, clarity, literalness, organization, generality (laws or general principles), technicality, explicitness, public criteria of evaluation, refutability, hypothesis testing, expansion of common sense (with the possibility of undermining common sense), inaccessibility to the layman, theory construction, symbolic articulation, axiomatic formulation, learned journals, rigorous and lengthy education, professional societies, and a sense of apartness from naïve opinion.Thus mathematics, as much as physics, is inherently difficult to understand, arduous to learn, rigorous, technical, jargon-filled, highly general and abstract, based on objective epistemic procedures (proof not experiment), specialized, professionalized, and so on. Given all this, it would be arbitrary to deny that mathematics is a science, and historically it has been so classified (as in Gauss’s famous remark “mathematics is the queen of the sciences”). Only misplaced ideology would insist that physics is a science but mathematics is not–being non-empirical is not to the point.Thus the attribute of being “based on observation” is not a necessary condition for being a science (nor is it a sufficient condition, or else random remarks about what’s going on around you would be science).
Now it seems to me clear that contemporary academic philosophy, as it is practiced in typical university philosophy departments, has exactly the marks I have just recited: it is technical, rigorous, jargon-filled, difficult to master, expansive of common sense (or at variance with it), explicitly articulated, and so on. Someone might balk at the attribution of refutability and theory construction, holding that nothing like the physicist’s experimental testing and empirical theory construction applies to philosophical claims. But again, we must not bias the discussion by presupposing dubious paradigms of the scientific. The obvious fact is that philosophers revel in the construction and refutation of arguments of highly explicit and articulated kinds, often symbolically formulated; and when they propose a conceptual analysis they offer necessary and sufficient conditions that may easily be refuted by ingenious counterexample. We philosophers are often wrong, and demonstrably so. We propose theories of concepts (for example, the causal theory of perception) and our theories may be falsified. Also, the presence within philosophy of formal logic, philosophy of logic, philosophy of language, philosophy of physics, philosophy of biology, philosophy of psychology, and even abstract “theories of justice”, all testify to the affinity of philosophy with science—even if the philosopher is just in the business of analyzing concepts (both scientific and lay). My own view is that philosophy can be aptly described as “the logic of concepts”, and I take it that logic is a formal science in good standing—so philosophy is logical science, as I conceive of philosophy.
Not only is contemporary analytical philosophy a science, we can also fairly claim that philosophy was the firstscience—long before the formation of physics, chemistry and so on as special sciences. Philosophy attained the status of science well before physics ever did. To see this, just compare the “physics” of the pre-Socratics with the metaphysics of Aristotle: the former is oracular, poetic, and unsystematic, as much myth as fact, while the latter is carefully worked out, rigorous, and systematic—scientific. Geometry has some claim to be the original science (a formal science), but Aristotle’s syllogistic logic can reasonably claim to be one of the earliest forms of scientific study—when physics and chemistry were just glints in the eye of ancient thinkers. The sciences later split off from philosophy, but philosophy was already a science—metaphysics was a science before physics was (and Aristotle’s metaphysics holds up better today than his physics). To the unprejudiced eye, physics owes its present scientific status to a combination of sound philosophy and advanced mathematics—that is, to the achievements of the formal sciences. Without a solid methodological philosophy (itself arrived at a priori), and without the achievements of mathematics, especially geometry and calculus (themselves also arrived at a priori), physics would not be the imposing scientific edifice it is today.Empirical observation is only part of the story (see below for more on this). One might even say that philosophy, at its origins, was the consummatescience, especially symbolic logic—the model and ideal for other disciplines. It took physics and chemistry a long time to catch up.
Now I must deal with two objections to my classificatory picture. First, if we classify philosophy as a science, do we not deny that it belongs to the “humanities”, and isn’t philosophy largely concerned with the human? Here again the dictionary provides a useful starting point: for “humanities” the OEDgives “learning concerned with human culture, especially literature, history, art, music, and philosophy”. The obvious reply to this natural objection is that science and the humanities, so defined, are not mutually exclusive: in principle, there can be a science of human culture. Aren’t anthropology, sociology, social psychology, and economics precisely sciences of human culture in a broad sense? Suppose we agree that the human conceptual scheme is a part of human culture (not part of human biology); then studying human concepts is by definition the province of the humanities. But why can’t this study itself be a science? Isn’t the study of concepts by cognitive science a science—and yet we are agreeing that human concepts are part of human culture? There is simply no logical opposition here. Some kinds of study of written texts attain the status of science (word frequency counts and so on), so language studies can qualify as science too. And isn’t linguistics a science—yet with language part of “human culture”? It isn’t the subject matter that counts but the style, methods, and results. But—and this is even more obvious—it is simply false that philosophy is (exclusively) about human culture, even if its method is agreed to be conceptual analysis. For large tracts of philosophy concern the non-human world: the essential nature of space, time and matter, causation, necessity, probability and laws, consciousness, choice, perception, knowledge, justification and truth. Some of these subject areas indeed concern the psychological, but that doesn’t make them an aspect of “human culture”: they may rather be aspects of the biological world, and may also apply to non-human animals.But anyway many of these areas of study are not even about psychological matters, so philosophy isn’t one of the “humanities”, as the dictionary (quite reasonably) defines the term. It is about the whole world not the specifically human world. Some parts of philosophy do deal with aspects of human culture—aesthetics, philosophy of law, social theory, maybe some of ethics—but many do not. I would not myself want to describe human concepts as part of human culture (though they are part of human nature) despite their being psychological entities; so I don’t think that conceptual analysis is ipso factoa study of a component of human culture. At any rate, there is no good objection here to counting philosophy as a science in good standing.
The second objection I want to consider is more serious. You might agree that the distinctive method of philosophy does not preclude its being a science, but you might also fasten onto another aspect of the dictionary definition of “science”—that a science is an organized system of knowledge. And here the objection is apt to be that philosophy cannot boast an established set of results—a body of accepted knowledge, agreed upon by all, and neatly set out in the philosophy textbooks. For where are the philosophical factsto be set beside the facts of physics, chemistry and biology? Philosophy, it will be said, does not make the kind of progressmade in the sciences, including the formal sciences; it just isn’t as epistemically solid. The very idea of philosophical knowledgeis an oxymoron, a sheer fantasy.
Now there is much to be said about this kind of objection to the scientific standing of philosophy, but I shall try to be brief. First, we must not underestimate how much knowledge philosophy has actually acquired, of a quite straightforward sort, mainly in the way of establishing certain distinctions: that is, philosophers have clarified certain important distinctions that were previously blurred and unrecognized—and this is real cognitive progress (use and mention, type and token, particular and general, name and quantifier, necessary and contingent, fact and value, knowledge and belief, analytic and synthetic, sense and reference, character and content, implication and implicature, etc).Second, philosophy has discovered and articulated the various theoretical optionsthat are available in any given problematic area, even if it has not actually settled which options are the true ones: it has mapped out the philosophical geography—and this too is genuine cognitive progress. Knowing what these options are, and appreciating their strengths and weaknesses, is a large part of what makes philosophy appealing to many of us—we hadn’t thought of them before we came to the subject, and they add to our store of knowledge (they also expand our imagination).But still, it may be retorted, don’t the sciences do more than merely articulate the theoretical options–don’t they decidewhich are correct and which incorrect?
Well, that depends. The further from direct observation the science becomes the harder it is to produce consensus and conclusive verification. Contemporary quantum physics is an obvious case in point: many options and no agreed way to settle which is right. So is theoretical physics not a science? Biology cannot decide how life originated on earth, though some options have been sketched out; so is biology not a science? Psychology is notoriously beset by disagreements, sometimes fundamental, but it would be a stern linguistic policeman who denied the label “science” to psychology (ditto economics and sociology). No science is immune to controversy and polarized opinion, once you get beyond the lower reaches of the discipline. But there is a more telling point to be made, concerning difficult and easy science. Suppose I establish a new field of study called “mysteryonics” in which all the hardest questions of science are to be pursued—from physics, biology, psychology, etc. No easy questions are allowed in mysteryonics! Then, clearly, this subject will make little solid factual progress, compared (say) to botany or geography, precisely becauseit is about the most difficult questions of all. We might say that mysteryonics encompasses the “hard sciences”, with “hard” now connoting sheer intellectual difficulty. Should we say that this subject is not science at all? That seems to limit the word unduly to easy science (we might call this field “easyatrics”). Surely science can come in degrees of difficulty, and it is invidious to withhold the label from the tougher areas. But then isn’t the word “philosophy” really a name for a subject that deals with reallydifficult questions (though sometimes with easy ones)? So the degree of difficulty of philosophy shouldn’t be interpreted as an inherent lack of scientific status. Speculative problematic hard-to-do science is still science—and it may be necessary and unavoidable science (for any intellectually honest and adventurous enquirer). So long as the questions are real and the standards of investigation are rigorous, we can still claim the title “science”. If physics and chemistry had proved harder to do than has emerged historically, would they not then be sciences? And just because it is relatively easy to establish particular historical or geographical facts doesn’t make these studies into science (or the best science). Degree of difficulty is beside the point. We should certainly not let crude outdated positivism dictate how the term “science” is to be applied (“to be scientific is to be conclusively verifiable”).
I am resisting the idea that certain sciences constitute paradigms for what a science must be—the sciences regularly deemed “empirical”. In particular, I am rejecting the notion that there is any necessary link between the concept of a science and the concept of observation: some sciences are observational and some are not (I will be returning to this). We must avoid the fallacy of the misplaced paradigm, here as elsewhere—which is often the science of physics. There is simply nothing in the meaning of the word “science” to entail that a science must be based on observation—that is, perception by means of the (outer) senses of objects and events in the scientist’s physical environment. So in saying that philosophy is a science I am certainly not saying that philosophy is based on observation, like physics. I am saying that despitenot being based on observation philosophy is a science—and has been for a long time. Nor is it my view that philosophy must be refashioned in order to become like a science: it is alreadya science. This is true even of the part called “ethics”, which could be called “axiological science” (the phrase “moral science” already exists). There can be a science of value, i.e. a systematic, rigorous study, involving generalizations and demonstrations.If philosophy consists of conceptual analysis, then it is (in Kantian terminology) an “ampliative science” not an “augmentative science”, i.e. it produces analytic truths and not synthetic truths. But again, there is no good reason to insist that augmentation of knowledge is definitive of science, so long as the other features are present. If mathematics is likewise ampliative, not augmentative, that is no reason to deny that it is a (formal) science.
It might now be conceded that there is no linguistic impropriety in applying the word “science” to the discipline of philosophy, but maintained that certain key distinctions still exist between philosophy and other sciences—distinctions that blunt the force of the application of that word. Thus it will be said that philosophy is neither an experimental science nor an empirical science nor a natural science. These characteristics are usually taken to connote epistemic virtues, in which case philosophy lacks the virtues that are typical of other sciences. It is a science in name only, it may be said, lacking the traits that constitute the distinctive virtues of sciences in general. Is this line of thinking justified? I shall now argue that it is not.
Is conceptual analysis experimental? The OEDdefines “experiment” as follows: “a scientific procedure undertaken to make a discovery, test a hypothesis, or demonstrate a known fact”, from Latin experiri“try”. I take the primary sense here to be that of testing a hypothesis; so does the activity of conceptual analysis test hypotheses? I suggest that it does: for this procedure involves the production of a hypothesis about the analysis of a concept—specifically, a set of putative necessary and sufficient conditions—which is then tested by means of “thought experiments”. For example, we might propose the hypothesis that knowledge is simply true belief and test that hypothesis by imagining cases in which a subject has true beliefs and asking ourselves whether he has knowledge—and we might go on to produce a counterexample in the shape of a subject who has a true belief that is completely unjustified. We might then amend the hypothesis to maintain that knowledge is true justified belief and proceed as before—finding that these conditions are not sufficient either (we come across Gettier cases). We have conducted an “experiment” in our mind to test the analytic hypothesis we have conjectured to be true. It consists of conceiving possible cases in which the conditions are met and then asking ourselves whether our understanding of the concept indicates that the concept applies in these cases. True, we did not make any perceptual observations in performing such a thought experiment, but the dictionary does not specify that experiments must be conducted by deploying the senses perceptually. The heart of the definition is trying out a conjecture in an open-minded spirit and coming up with a confirmation or a refutation. We certainly did not presuppose an answer to our question before undertaking the procedure: we let the procedure decide the question only after it had been completed. The hypothesis makes certain predictions about our intuitions in possible cases and then we check to see if our intuitions confirm or disconfirm the hypothesis. The intuitions function as evidence for or against the hypothesis, and they are grounded in our actual grasp of the concept in question. So it is not wrong to describe the method of conceptual analysis as an “experimental” procedure: this is quite literally true, as the dictionary defines the word (and as we normally understand it). Maybe uttering the sentence “Conceptual analysis is experimental science” will typically have conversational implicatures that are false, conjuring up images of physical dissections of concepts and Bunsen burners of the mind; but in its literal content it is a perfectly true statement (it has no false logical implications). Thought experiments are indeed experiments, just as we would naively suppose from the phrase. We have here a procedure that arrives at truth in a non-question-begging way by consulting a body of data (“intuitions”) generated especially for the purpose.
I can imagine an opponent sputtering that these so-called “experiments” are not conducted in a laboratory, like real scientific experiments, with suitable equipment: there is no such thing as a “philosophy lab”! But is that true? Here is how the dictionary defines the word “laboratory”: “a room or building equipped for scientific experiments, research or teaching”, from the Latin laborare“to labour”. So the core notion refers to a space expressly set aside for performing scientific experiments, as opposed to a space set aside for day-to-day living or throwing parties or darts matches. But don’t philosophers have rooms especially set aside for research purposes—spaces in which they philosophically labor? We call them “studies” or “offices” or “seminar rooms”. I shuffle into my study in order to do philosophical research work and in that room I often carry out thought experiments to test analytic hypotheses; so am I not spending time in my “philosophy lab”? The implicatures of saying this in certain contexts will doubtless include suggestions of white coats and expensive equipment, but there is nothing literally false in the proposition itself: I am simply laboring to make philosophical discoveries in the space set aside for making such discoveries. There are many kinds of laboratory, varying in their contents according to the subject in question. If I am pressed to specify what equipment I use in my lab, I might reply that I require a chair and desk with suitable writing materials and some peace and quiet—these are the tools of research that I employ (physicists, more grandly, have their massive particle accelerators). I work in a lab performing thought experiments—and my most precious tool is my brain.
My interlocutor might at this point reluctantly agree that I am guilty of no outright linguistic solecism in describing conceptual analysis as “experimental” but insist that such experiments hardly qualify as “empirical”. Here the question becomes trickier, because “empirical” can mean several things. Let us again turn to the dictionary for some initial guidance: “empirical” is defined as “based on, concerned with, or verifiable by observation or experience rather than theory or pure logic”, from Greek empeiria“experience”. Now we can agree that conceptual analysis does not proceed by observation, but what about experience? During a thought experiment don’t we have certain experiences—what might be called “conceptual experiences”? Conscious cognition is one variety of experience in a suitably wide sense; not all experience is sense experience.When I judge that I have imagined a case of true belief that is not knowledge do I not report a certain conscious experience that I had? The conscious episodes called “intuitions” are just a type of cognitive experience—part of my total phenomenology at that time. Similarly, we may have mathematical experiences, as when a proof is appreciated and accepted. In the case of conceptual analysis, the intuitions play an evidential role, so we can say that they constitute empirical (“experiential”) evidence in this broad sense. We certainly did not proceed from pure theory or logic without regard for any new cognitive input—we were open to new cognitive experiences. Thought experiments are thus rightly described as “empirical” in a perfectly good sense of the word. Indeed, we can even describe them as a posterioriin the sense that they establish their results only aftercertain experiences have been obtained; they are not dogmatically held quite independently of what experience may bring. It is only after the analytic investigation has been completed that a conceptual hypothesis is accepted; it is not presupposed at the start.
It now appears that philosophy and physics are alike in being experimental empirical sciences, cleaving strictly to what these words literally mean. So what distinguishes them? Here we might appeal to the notion of perception: physics relies on perception of things but philosophy (conceived as conceptual analysis) does not. Surely that distinction is rock solid! Not quite: the dictionary must again give us pause. Under “perceive” we find “become aware or conscious of”, from Latin percipere“seize, understand”. There is nothing here restricting perception to the five (or more) senses: intellectual perception is a type of perception too. Thus I can be said to “perceive” that true belief is not sufficient for knowledge by conducting an appropriate thought experiment. Here “perceive” is synonymous with “apprehend”, which has both sensory and intellectual forms. Thus philosophical investigation involves perception in this capacious sense: I often become aware or conscious of something while engaged in philosophical thought. I “see” that something is so.
But we can easily recast the thought behind this suggested differentiaby bringing in the senses explicitly: philosophy does not depend on perception by means of the senses. Even here we must tread carefully, since one traditional view is that we have an “inner sense” capable of sensing what lies within—in the place where concepts lurk. According to that view, I do sense the make-up of my concepts, because I use my inner sense to gain insight into them. The obvious amendment here is to qualify “sense” by “outer”: philosophy, unlike physics, is not based evidentially on the deliverances of outersense. The dictionary provides a useful gloss on this philosophical notion: “sense” is defined as “faculty by which the body perceives an external stimulus; one of the faculties of sight, smell, hearing, taste, and touch”, from Latin sentire“feel”. And that notion of sense certainly excludes conceptual analysis, since I don’t use my bodily senses of sight, touch, smell, etc to excavate the content of my concepts. If we use “observation” to capture the use of the senses in this narrow sense, then we can correctly say (at last!) that physics is based on observation and philosophy is not. But notice that we can still claim that philosophy has the qualities (in both the descriptive and normative senses) of empirical experimental science: the methodological gap is therefore not as wide as we might initially have thought.
Is philosophy a natural science? I think there are two senses in which it is. First, it is not a supernaturalscience (which looks like a contradiction): it doesn’t deal with the supernatural, unlike theology. In this sense the formal sciences are also natural sciences—because not supernatural. But philosophy differs from the formal sciences in that it can also be about objects that exist in nature, as numbers and the like do not. Metaphysics deals with the nature of material objects and causation, say, and philosophy of mind deals with consciousness and choice, and philosophy of language deals with human natural languages. The subject matter consists of things that exist in the natural world, not the “abstract world”. It is therefore entirely appropriate to say that philosophy is a natural science (concepts themselves are perfectly natural entities in both senses of the word).
We can even describe both philosophical and physical knowledge in causal terms. As the physicist causally interacts with his subject matter of physical objects to gain knowledge of their properties, so the philosopher causally interacts with his subject matter of concepts to gain knowledge of theirproperties. That is, concepts are psychological entities that we investigate in conceptual analysis, and these entities play a causal role in producing knowledge about themselves.When I come to know that knowledge is true justified belief, say, it is my concept of knowledge that causally controls the process of analytic belief formation; I arrive at this analysis becauseof the concept of knowledge that I possess. So, again, at a deeper level we find an affinity between the knowledge systems of physics and philosophy, not stark difference. The acknowledged difference, relating to the use of the (outer) senses, comes to seem relatively minor, not a mark of clear superiority on the part of physics. We can see a clear methodological parity.
Still, it may be retorted, there is that clear distinction, and observation is clearly an epistemic virtue; so isn’t physics a better(empirical experimental natural) science than philosophy? And given the honorific force of the word “science” we can see why someone might want to apply it preferentially to sciences like physics, as opposed to sciences like philosophy (the formal sciences). In reply to this, I propose to make a more radical suggestion: physics is notinherently an observation-based science, and conceptual analysis is notinherently observation-independent. We can, in fact, invert the epistemic basis of the two types of science. This is actually not so very difficult, on reflection. Consider first a brain in a vat, observing nothing: all its sensory experiences are hallucinatory. This individual might, however, be an aspiring physicist. Let us stipulate that this physicist-in-a-vat has experiences as ofbeing in a physics lab and performing experiments, but never is really so situated: the vat supervisors feed in these hallucinations, making sure to provide her with data that are in fact correct. They simulate the course of experience of an actual perceiving human physicist conducting actual experiments, but no real observationsare ever made.On the basis of these hallucinatory experiences, our aspiring physicist-in-a-vat might well come to entertain some physical theories; and since the data don’t mislead her, and she is a scientific genius, she comes up with true physical theories—first following in the path of Newton and somewhat later Einstein. I suggest that she is engaged in the science of physics, even though she never makes any actual observations (only apparent observations). In other words, veridical perception of experimental results is not a necessary condition of doing physical science. Human physics is based on observation (so long as we are not actually brains in a vat!), but this is a contingent not a necessary truth. All the procedures of inference and theory construction are the same for the physicist-in-a-vat as for the physicist-in-a-lab, so it would be quite unwarranted to declare the latter a genuine physical scientist and the former not.
If you are worried that the vat physicist case at least still involves sensory experience, then consider a further case: all the evidential knowledge possessed by the typical physicist is fed into the genetic make-up of a hypothetical physicist, so that she knows all the physical data innately. There are no sensory experiences (even hallucinatory ones) as of a meter reading such and such, but just basic beliefs about what meters read in such and such conditions.But we can suppose that there is no innate knowledge of the correct theory that explains all this innately known data; that will require scientific intelligence of a high order. So the would-be physicist here needs to construct theories to explain the data written into her genes; and if she succeeds in doing that then I submit that she is engaging in physical science. Yet she never makes any observations. After all, doesn’t God have scientific knowledge of the physical universe, and yet he makes no sensory observations, not having any bodily senses. That’s just the way wedo it, given the limits of what we know before interacting with the world through our sense organs; it is not essential to the very existence of scientific knowledge of physical reality. Science and sense experience are not inextricably linked.
The same point can be made in a different way. As things are, we learn about the brain by observing it: we open up the head and take a look, applying various observational techniques. But is this essential to knowledge of the brain? What if an aspiring brain scientist, perhaps more ingenious than we humans are, undertook to learn about the brain by means purely of introspection? He introspects his states of mind, recording their laws and ways, and tries to infer what is going on in the underlying reality of the body (we can suppose that he has never seen a brain). Why should he not be able to come up with hypotheses about what the organ is like? He might conjecture that it has a cellular structure (other organs in the body have been observed to be cellular), and that it exhibits localization of function, and even that mental processes are powered by electricity (many other biological processes are and it fits the introspective data nicely, what with the rapidity of mental processes and the like). He might with sufficient ingenuity come up with a theory very much like our observation-based theory, yet he never observes brains at all, proceeding entirely from introspective data plus some ancillary knowledge of the natural world. I submit that he is doing neuroscience, despite the absence of an observational foundation. Thus there can be science without observation (though not without evidence—but philosophy has evidence too, i.e. conceptual intuitions).
Now my interlocutor is itching to make his final devastating objection, viz. intuitions are not evidence at all but just subjective hunches and prejudices! I won’t attempt here to reply fully to this kind of objection, having done so elsewhere,but I will make one point that completely undermines this entire line of objection to the enterprise of conceptual analysis per se: namely, there is really no reason that conceptual analysis mustproceed in a non-observational first-person style (though I see nothing wrong with such a procedure); we can, instead, opt for third-person observational conceptual analysis. There are at least two ways of doing this. One is simply to investigate the concepts of others by eliciting their judgments about possible cases (“Would you describe the following case as an example of knowledge?”). This is the survey method much employed by the social sciences: questionnaires, statistical analysis, and so on. It is a method well suited to discovering the content of other people’s concepts when that is your main interest—as with anthropological investigation. But it is still conceptual analysis, i.e. discovering the necessary and sufficient conditions for the application of a concept. And it is straightforwardly observational, as much as any other survey of opinion. But a second, less orthodox, method might involve delving into the brain mechanisms underlying concepts: by discovering the neural correlates of a concept it might be possible to find out what other concepts the given concept embeds. Thus the concept of knowledge might have a neural correlate that contains as a part the neural correlate of the concept of belief or justification; this would be evidence, of a sort, that one concept contains another. It would be difficult research to carry out, and rather indirect, but it would surely be observational—and it would result in information about conceptual constituency. So there is nothing inherently non-observational about conceptual analysis. Such brain information could certainly supplement ordinary first-person inquiry into concepts. If our first-person conceptual judgments were highly unreliable for some reason, this might be a sounder way to proceed. At any rate, it is not logically ruled out. Such an inquiry would proceed from sensory observational knowledge, by contrast with the hypothetical methods of doing physics and neuroscience sketched above. To those who champion observation as the defining mark of the scientific, I ask whether they would agree that conceptual analysis would be methodologically superior to theoretical physics in the scenarios here imagined. Somehow I doubt it—which shows that the presence of observation is not so critical to solid science as some people seem to suppose. We don’t derive intellectual prestige inversion as a straightforward corollary of observational inversion. I myself think it is highly invidious and implausible to place so much emphasis on observation as determining what is sound respectable science. This is a legacy of positivism we can well do without.
This brings us to the amorphous but unavoidable question of science and epistemic virtue. The positivists made testability the central epistemic virtue of any theory, and any field of inquiry. And by testability they meant testability by means of sensory observation. The more observationally testable a proposition is the better it is. If a proposition or theory is not testable, or very hard to test, that is a demerit of the proposition or theory. Testability is regarded as theepistemic virtue. This produces a highly distorted picture of epistemic virtue. There are certainly many other epistemic virtues–such as generality, depth, interest, importance, profundity, objectivity, impartiality. Not only is testability not clearly correlated with these virtues, it also seems inverselycorrelated with them. The more testable a theory is the less general and profound it is apt to be. The reason is that human knowledge, particularly scientific knowledge, aspires to transcend the human viewpoint and human limitations—to describe the world as it exists independently of the human perspective—but testability directly reflects the nature and limits of human faculties. To be testable is to be testable by humans. This means a maximally testable theory is one that must cling to how the world appearsto us—the world as accessible to human faculties. If the faculties are the human senses, then testability is restricted to aspects of the world to which those senses are sensitive. The more a theory can be tested by the use of the senses the more it will be limited to appearances and the less to reality beyond appearances. A maximally testable theory is therefore apt to be trivial, as with a “theory” that lists the colors of objects in one’s immediate environment or gives the weight of every person in a particular town. Once a theory attempts to penetrate the local appearances, as with microphysics or cosmology, the harder to test it becomes. The most interesting theory is likely to be the one that is leasttestable. It might not even be (humanly) testable at all—yet very interesting nonetheless, and even true. We see this situation played out in contemporary theoretical physics, where the main theoretical options seem virtually incapable of experimental or observational demonstration (string theory, the many-worlds hypothesis, and so on). Big philosophical theories are notoriously difficult to establish and test, but they can be extremely interesting—such as Plato’s theory of forms or possible worlds metaphysics or panpsychism. Philosophy is more like difficult science than easy science—more like theoretical physics than taxonomic botany. We don’t think of botany as the queen of the sciences simply because its propositions (some of them) can be easily tested; we understand that testability is just one epistemic virtue among many. Criticizing philosophy because of its relative lack of observational testability therefore reveals a mistaken picture of what epistemic virtue consist in. And, of course, testability is a discipline-relative concept, with the formal sciences differing from the natural sciences in respect of how they are tested. Nor should it be forgotten that philosophical propositions are often quite straightforwardly refuted.
Are there viable conceptions of philosophy according to which it is clearly not a science? Don’t say that normative studies fail to be a science because science deals only in facts not values: that fails to envisage the possibility of “moral science”, i.e. a scientific value theory. Such an axiological science is so by virtue of its rigor, system and organization, as compared with naïve common sense, and certainly many practitioners have sought to develop a science of morals (for example, Bentham’s quantitative utilitarianism and Kant’s deductive deontology). To be sure, some parts of philosophy, as it exists today, might well not meet high standards of rigor and hence fail to qualify—perhaps “the philosophy of sex and love” would be an example. Nor need we assert that nothing can be of intellectual value that is not properly scientific: thus the humanities of literary studies and cultural history, or even marriage counseling and horse whispering (whatever exactly that is). My position is certainly not that science is the only form of worthwhile cognitive activity.It is just that philosophy as it exists today, and has existed for quite some time, is aptly described as a science, with all the virtues that attach to that particular form of inquiry. I suppose this may be contested by people characterizing themselves as “Wittgensteinians”: they may see a sharp contrast between the activity of philosophy and anything deserving to be called a “science”. But three points may be made about this. First, philosophy as I conceive it is a sui generisscience, not to be assimilated to the so-called “natural sciences” of physics, chemistry and biology (and these differ among themselves too). I am emphatically not taking physics as my scientific paradigm (I might even take philosophy as my paradigm of the scientific). Second, Wittgenstein’s later “therapeutic” conception of philosophy is really an extreme and minority position, fitting ill with vast tracts of the subject, early and late (which Wittgenstein seemed willing to dismiss completely). Third, it is not so clear that no trace of the scientific, in my capacious sense, survives in Wittgenstein’s work. The Tractatusis certainly a scientific treatise in my sense (as is Russell and Whitehead’s Principia Mathematica,on which the Tractatusis modeled); and even the Investigationscan be construed as a kind of treatise in linguistic science, with its naturalistic surveys of the different forms of language and inquiry into natural language “grammar”. The Investigationsis quite rigorous and systematic in its way; it is certainly not just a collection of vague poetic pronouncements and gnomic exhortations, or unrelated apercus. It is a piece of analytical philosophy, after all, possessing the kind of rigor and organization characteristic of such philosophy (it is nothing like the writings of, say, Henry David Thoreau or the utterances of Eastern mystics). And Wittgenstein was himself originally a scientist. We might reasonably construe him as resisting the emblematic pull of the natural sciences, interpreted narrowly, not as repudiating the word “science” altogether as properly applicable to philosophy.
It seems to me, then, that the standard conceptions of philosophy—continuous with (empirical) science, a prioriconceptual analysis (concerned with de reessence), and even Wittgenstein’s later philosophy—are all compatible with the idea that philosophy is aptly described as a science. And if you look at what is actually done professionally today in university philosophy departments the impression is overwhelming, no matter what meta-philosophy you may favor. Nor is there any cogent reason that I can see to resist the label. Then why is it not so regarded? The reasons are no doubt many: misplaced paradigms of the scientific, traditional university classifications, mistaken ideas of value as something inherently “subjective” and hence “unscientific”, a lingering association with religion and the “spiritual”. But surely part of the reason is the word“philosophy”: its etymology, history, and popular connotations. For how can a general “love of wisdom” count as a particular science with its own specific subject matter and methods? The name of the subject accordingly blinds people as to its real nature. This is why I have suggested elsewhere that we would do better to re-name the subject in order to reflect its true status as a science.Just as the other sciences have shed their earlier label as species of “philosophy”, so we philosophers should shed our traditional (and misleading) name. Unfortunately, there is no convenient alternative name already in existence, so we need to invent a new one or adapt a word already in use. The best I can come up with is “ontics”, which for various reasons strikes me as preferable to other possible choices (“ontology” is already in established use as a name for a partof philosophy). It sounds a bit like “physics” and a bit like “ethics”, and is intended to express the concern of our subject with general questions of being. I won’t try to defend this linguistic choice here. My point is that if you sympathize with my thesis that philosophy is really best viewed as a science, then you might well want to have a name for the subject that reflects that position—as “philosophy” plainly does not. Of course, we could keep both names in use, at least for a hundred years or so, in order to acknowledge the past and avoid bafflement. But having the name “ontics” to hand would dispel a lot of misunderstanding about what kind of subject philosophy is and also do justice to its status as a branch of scientific learning. The title of this paper might then be recast as “The Science of Ontics”, which carries no whiff of oxymoron. Psychologists once decided to rename their subject “behavioral science” because they felt this label better reflected the nature of the discipline they practiced, and the newfangled “cognitive science” has much the same point (is academic psychology really the study of the “psyche”?). I am making a similar proposal: “ontical science” is simply more accurate and descriptive, less misleading. Using this term in conjunction with the traditional label will foster a better understanding of the field so named, and eventually the label “philosophy” may fall out of common use. No doubt there was a period in which the study of matter and energy was called both “natural philosophy” and “physics”, as the transition from one term to the other was made; I advocate such a transitional period for the field now called “philosophy”—with “ontics” the term that will eventually be preferred.If philosophy is indeed a science, to be set beside other scientific subjects, then it needs a name that fits its real nature.
We associate this type of view with Quine, but Russell espoused it also. Perhaps we should add that both philosophers were prepared to jettison such parts of traditional philosophy as could not be so subsumed: what was discontinuous with science in the inherited corpus of philosophy should be consigned to the flames. In this they shared the predilections of the pruning positivists.
These are not the only conceivable metaphilosophies: one might hold that some philosophy consists of synthetic a prioripropositions, in which case conceptual analysisdoes not exhaust the field; or one might favor a purely therapeutic view of philosophy in the style of the later Wittgenstein. But the two metaphilosophies I have mentioned are the most popular.
For a defense of this position see my Truth By Analysis: Games, Names, and Philosophy(Oxford University Press: New York, 2011). The position is nowhere near as narrow as we have been taught to think, once we have a properly inclusive conception of analysis.
Note that the dictionary editors require that a body of knowledge should be “systematically organized” before it qualifies as a science, so that episodic history or disconnected geography will not count. Of course, it is necessary to say more about what this systematic organization amounts to in order to have a more precise definition—which I supply in the text. What is crucial in the dictionary definition is that “science” can refer to knowledge “on any subject”—so long as the knowledge is of the right systematic kind.
Someone might say that “empirical science” is pleonastic, since all science is by definition “empirical”. But this is semantically implausible, because “mathematical science” is surely not contradictory. If someone were to insist that as theyuse the word “science” all science is by definition empirical, I would respond as follows. Let us introduce the word “schmience” to refer to any discipline that has all the marks of science except being empirical: then mathematics and logic will count as schmience, as will philosophy in my estimation. Now since “schmience” is not so easy to pronounce, let us modify it to “science”: then we can say mathematics and philosophy are science in the sense defined. The point of any classificatory scheme is to capture salient similarities, even where differences exist, and I am suggesting that mathematics and philosophy share important traits with disciplines already described as “science”. As instances in which the word is naturally employed by philosophers to characterize their discipline, let me cite Galen Strawson and Edmund Husserl (neither of whom shares the “naturalistic” view of philosophy). Strawson writes: “Philosophy is one of the great sciences of reality” and goes on to list its similarities with the so-called natural sciences, despite being a priori: Real Materialism and Other Essays(Oxford University Press: Oxford, 2008), p. 1. Husserl published a famous essay in 1911 entitled “Philosophy as Rigorous Science”, extolling the virtues of his phenomenology and contrasting it with the prevailing naturalism of the day; and his own metaphilosophy was scrupulously a priori. Neither of these philosophers takes the word “science” to apply only to empirical disciplines like physics and chemistry, and they are guilty of no semantic solecism.
I am not here trying to provide a definition of science, still less specify a clear-cut “demarcation criterion” in the manner of Popper and others; I am simply listing the salient marks of science as we ordinarily understand the term. This is enough to recognize that astrology, say, will not qualify as a science, mainly through lack of truth and justification, and neither will simple geography, through lack of laws and generality (among other things). If I were pressed to settle on the core notion here, I would suggest the presence of highly general laws or principles—of which philosophy can boast many. But it is probably best not to insist on a rigorous definition of “science” in terms of noncircular necessary and sufficient conditions.
The case of geometry is instructive: it was traditionally regarded as both a prioriand as giving the truth about real space—does the former imply that it cannot count as science? But then isn’t it also a rigorous description of one piece of physical reality, viz. space? Some propositions of physics itself have an a prioricharacter, as with Newton’s laws of motion or Descartes’ definition of matter. The only kind of view of mathematics that could disqualify it from being a type of science is that it is merely a game with symbols to which the notions of truth and falsity do not apply—that is, that we cannot knowmathematical propositions (since strictly there are none). Science presumably requires at a minimum that some actual propositions be actually known (but notice that on some instrumentalist views of physics we don’t have real truth and falsity either).
See my Truth By Analysis(cited in note 3), esp. chap. 7. Philosophy can thus be described as “the science of entailment”. If you feel that some of philosophy is too impressionistic and inchoate to count as science, then by all means amend my claim to read that a great dealof philosophy qualifies as science (and is everythingtalked about in physics and biology departments properly “scientific”?).
This needs more discussion, but I think my point is clear enough: it is not the mere presence of observation in physics that makes it so impressive and successful, but rather its mathematical articulation and abstract generality—which have a quite different source. Also, the very emphasis on observation is not the resultof observation but instead reflects a distinctive epistemological position—that knowledge of reality is best gained that way rather than by means of revelation or inherited authority.
Obviously, I am rejecting a “pan-cultural” view of reality—that it is all “social construction” or some such. I am supposing that philosophy deals with reality as such—the real, objective article. I take this to be compatible with the thesis that methodologically philosophy proceeds by conceptual analysis: in philosophy we analyze reality conceptually (see the book cited in note 3).
Even if some people see fit to reject some of these distinctions for one reason or another, it cannot be denied that they have clarified previously murky ideas and paved the way for superior ways of thinking.
One type of knowledge delivered by philosophy is knowledge of knowledge—and of ignorance. We learn the scope and limits of knowledge—what is doubtful or unproven or merely groundlessly accepted. This is real knowledge, not available to those who refuse to study the subject; according to Socrates, it is knowledge of a particularly valuable kind. What philosophy does not provide is knowledge of particular empirical matters of fact—but why is that so marvelous? Philosophy produces its own kindof knowledge.
The whole positivist emphasis on verification distorts our view of the essential character of science—especially if we try to reduce theoretical propositions to something called “empirical content”. But this is by now an old story.
There is not just moral value to be considered but also norms from non-moral domains, e.g. logic and epistemology. In logic we are certainly concerned with the normative question of how to reason, but that doesn’t disqualify logic from being a science. And does the use of logical norms in reasoning within physics mean that physics is not a science either? A logical system simply isa science of (logical) norms. Normative science takes its place as one type of science among others.
The old distinction between the inductive and deductive sciences is useful here. Mathematics and philosophy are deductive sciences, being concerned essentially with entailment, and hence proceeding by proof and argument; while physics, chemistry and biology are inductive sciences, being concerned with deriving laws of nature from particular observations (perhaps using inference to the best explanation). The genus is science and the species are the inductive and deductive sciences.
Compare conceptual analysis with chemical analysis. A chemist might conjecture that water is composed of H2O and tests this hypothesis by contriving suitable chemical combinations (she already knows that oxygen or hydrogen by themselves are not sufficient to produce water). Just so, a philosopher might conjecture that knowledge is true justified belief (he already knows that belief and truth separately are not sufficient to produce knowledge). The questions in both cases are fundamentally mereological. The chemist uses empirical observation, the philosopher uses intuitions about possible cases—but the type of question is the same, i.e. what constitutes what. There are factsabout what constitutes our concepts and it is possible to ascertain what these facts are, just as there are ascertainable facts about what constitutes water (see Truth By Analysis, cited in note 3).
We have emotional and conative experiences, as well as sense experiences, and the exercise of our rational faculty also involves distinctive modes of experience. If we refuse to apply the word “experience” here, then what word shall we use instead? Clearly there are conscious goings-on of some sort.
We can truly say that we acquire informationabout concepts by interactingwith them in the course of conceptual analysis—as we acquire information about material things by interacting with them in the course of empirical observation. Concepts are real mental entities that we can gain knowledge about by directing our attention to them in the process of conceptual analysis, thus deriving necessary and sufficient conditions for their application. The mode of interaction here is admittedly not by means of the outer senses, but so what?
The case is not essentially different from gaining knowledge of our feelings, sensations and thoughts by means of introspection; conceptual analysis just digs a little deeper into the structure of our thoughts and other propositional attitudes.
One might wonder whether the vat supervisors make actual observations as a basis for the information they feed into the non-observing brain-in-a-vat, so that there is an ultimate observational basis for the knowledge acquired by the latter individual. The case would then be just like a testimony case. But we can get around this objection by stipulating that the supervisors do not acquire their knowledge of physics by perceptual observation: they might have their physical knowledge innately or have non-sensory godlike faculties or be equipped with a kind of super blindsight (i.e. they have no perceptual experiences at all, though they do gain information about the world by interacting with objects in their environment). All we need to do is eliminate the role of ordinary veridical sense experience from the epistemic picture, and this seems easily done. After all, some philosophers believe that perceptual experiences play no evidential role anyway.
I am here relying on a basic principle about beliefs, viz. no belief is necessarilycaused by a perceptual experience. Any belief actually based on a sense experience couldhave arisen from some other cause—either by inference from another belief or as a basic innate belief. It is only contingently true that beliefs are caused by sense experiences, though in the human case this mode of causation is very common. The truth of this principle already refutes empiricism, since no belief has any intrinsic“empirical content”, i.e. a set of sense experiences that are entailed by having that belief. A conceivable believing subject could have the same beliefs as us and yet have no sense experiences at all, according to the principle.
Let me be clear: I am not saying that observation cannot be evidence, only that not all evidence is observational, i.e. based on experience generated by outer sense. Of course, we could stipulate a new sense of “observational” to mean just “whatever is evidentially basic”, whether this is sense experience or something else; but then we are affirming only that science is necessarily based on evidence. The conceptual analyst believes in evidence too, in the shape of conceptual intuitions, so no epistemic distinction has yet been identified. Once “observation” is detached from sense experience no deep epistemic distinction between physics and philosophy exists, because both are evidence-based enterprises. We can all agree that rational belief requires the existence of reasonsfor belief, trivially so, where these reasons might or might not be (or involve) sense experiences.
This is what Bernard Williams called “the absolute conception” in Descartes: The Project of Pure Enquiry(Harmondsworth, Penguin Books, 1978). See also Thomas Nagel, The View From Nowhere(Oxford University Press, New York: 1989). How human beings test a theory, say by directing their eyes at a measuring instrument, is actually at variance with what the content of the theory aspires to be, i.e. independent of the human viewpoint.
This point is really quite obvious, but it is often ignored: it is not that philosophical claims are somehow too wishy-washy to be falsifiable. I think myself that philosophy can boast an epistemic superiority compared to physics, because of its extreme generality, depth, and transparency. I discuss the epistemic limitations of physics by contrast with both philosophy and psychology in “Two Types of Science”, in my Basic Structures of Reality: Essays in Meta-Physics(Oxford University Press, New York: 2011).
I want to make room for poetry and literature as valuable sources of knowledge, but they are clearly not science. I might even be prepared to uphold the cognitive value of music and dance. I even think “life tips” can be worth listening to. It is just that some areas of human discourse are aptly described as “science” and some are not—with philosophy falling into the former category and poetry not. This is purely a question of descriptive accuracy, not some kind of misplaced “science worship”.
If it is countered that Wittgenstein opposed explanations in philosophy and the search for causal laws and generalizations, then we can note that not all science is explanatory and causal: some science is modestly descriptive and taxonomic, as with much of biology (and mathematics is hardly explanatory and causal). Nor does the emphasis on intellectual therapy preclude a scientific foundation, since such therapy may well proceed from a scientific basis—such was the claim made on behalf of psychoanalytic therapy. Wittgenstein’s therapeutic efforts proceed from a naturalistic description of the many forms of language and from a quite specific conception of the nature of meaning. I see no reason to withhold the label “science” from his brand of linguistic investigation, any more than from other forms of linguistic study (Austin, Chomsky, Grice et al).
The professional name “scientist” was introduced as recently as 1833, by William Whewell; before that we had “natural philosopher” or “man of science”. In a similar spirit we philosophers could rename ourselves “onticists”, if we are persuaded that “philosopher” is not an apt name for what we do. Apparently there was a good deal of discussion in the Royal Society regarding the merits of the name “scientist”; we can envisage just such debates about the proper labeling of the people now called “philosophers”. I think “onticist” has quite a nice ring once you get used to it.