CC BY
26
Эпистемология и философия науки 2019. Т. 56. № 2. С. 46-52 УДК 101.2
Epistemology & Philosophy of Science 2019, vol. 56, no. 2, pp. 46-52 DOI: 10.5840/eps201956228
R.
keply to dennett, knobe, kuznetsov, and stoljar on philosophical methodology
Timothy Williamson - PhD, Wykeham Professor of Logic. Oxford University, New College.
Holywell St., Oxford OX1 3BN, UK.
e-mail: Timothy.williamson@ philosophy.ox.ac.uk
The paper replies to replies by Dennett, Knobe, Kuznetsov, and Stoljar to the author's 'Armchair Philosophy'. Keywords: Armchair philosophy, intuitions, a priori knowledge, abduction, model-building, dependency structures.
Ответ оппонентам
по поводу философской методологии
Тимоти Уильямсон - доктор философии, профессор. Оксфордский университет. Holywell St., Оксфорд OX1 3BN, Великобритания. e-mail: Timothy.williamson@ philosophy.ox.ac.uk
В данной статье автор отвечает на критические замечания оппонентов (Д. Деннета, Дж. Ноуба, А.В. Кузнецова и Д. Столжара). Ключевые слова: кабинетная философия, интуиции, априорное знание, абдукция, построение моделей, структуры зависимости
Daniel Dennett, Joshua Knobe, Anton Kuznetsov, and Daniel Stoljar have made thoughtful responses to my position piece on armchair philosophy, identifying many points of agreement and some of disagreement. This reply deals mainly with the latter.
1. 'Philosophical Intuitions'
Knobe's title is 'Philosophical Intuitions are Surprisingly Robust Across Demographic Differences'. He writes that 'the aim of experimental philosophy [...] is to find the truth about people's intuitions'. He takes for granted that a central issue can be neutrally articulated in the question: how reliable is 'a method that relies on intuitions'? According to Kuznetsov, my view in The Philosophy of Philosophy is ('Roughly speaking') 'that there are special philosophical intuitions that support philosophical inquiry' which 'are, to some extent, universal and need special philosophical training'. In their pieces, neither Knobe nor Kuznetsov makes any attempt to explain what they mean by an 'intuition', or by describing one as 'philosophical'. Dennett characterizes 'naïve naïve axiomatic auto-anthropology' as 'thinking that the royal road to truth is to attempt to axiomatize, with your companions,
46
© Timothy Williamson
your shared intuitions', though he is careful not to ascribe that methodology to me. He too does not say what an 'intuition' is. Stoljar is the only one of the four not to use the 'i'-word.
In 'Armchair Philosophy', I simply avoided the 'i'-word. Given the limitations of space, I preferred not to use any of it explaining my reasons for avoidance. Since my 2004 article 'Philosophical "Intuitions" and Scepticism about Judgment' (the clue is in the scare quotes), I have been arguing that the debate about the reliability of 'philosophical intuitions' is ill-posed, because the extension of the quoted phrase is quite unclear. The point is not just that there are borderline cases; we cannot eliminate all vagueness from our vocabulary, and at the margins is usually does little harm. With the term 'intuition, it is much worse: most human judgments are in the disputed territory. Let me explain.
Psychologists distinguish between 'intuitive' and 'reflective' judgments. Roughly, reflective judgments are those based on conscious reasoning; intuitive judgments are those not based on conscious reasoning (for simplicity, I concentrate on judgments, but the distinction can be extended to inhibited inclinations to judgment and the like). Some philosophers use the word 'intuition' with explicit reference to the psychologists' distinction. An example is Jennifer Nagel's excellent paper 'Intuitions and Experiments: A Defense of the Case Method in Epis-temology' (2012). However, as Nagel emphasizes, one consequence of so defining the term is that normal perceptual judgments (and many others) count as intuitions. Thus relying on normal perceptual judgments would count as relying on intuitions. That is not what the metaphilo -sophical debate was supposed to be about. Indeed, in that sense of the term, avoiding reliance on intuitions is not an option. For all judgments based on conscious reasoning rely on judgments not based on conscious reasoning. For instance, when you do a complex arithmetical calculation in your head, your final answer is based on conscious reasoning, but you did not go through an infinite regress of conscious reasoning: at some point in the calculation you made a judgment not based on conscious reasoning.
Can one finesse the problem for philosophical purposes by stipulating that 'intuitions' are based neither on conscious reasoning nor on perception? That too would wrong-foot the metaphilosophical debate. For our judgments about thought experiments are typically made by using offline, in imagination, the very cognitive capacities we use online, in perception. For example, the proposed stipulation would allow us to sidestep reliance on intuitions in Gettier cases by making judgments based on perception of real-life Gettier cases. We observe someone at 3 o'clock setting his watch by a clock that happened to have stopped at 3 o'clock, and judge that he does not know that it is 3 o'clock. Our judgment that he lacks knowledge is not an 'intuition' in the stipulated sense, since it is based on perception, but critics of the case method in epistemology will
be just as uneasy about it as they are about the verdict on the corresponding thought experiment - as I have put to the test by tricking audiences at my lectures into real Gettier cases. Thus the proposed restriction misconstrues the metaphilosophical debate.
In my paper 'How deep is the distinction between a priori and a posteriori knowledge?' (2013), I used this easy exchangeability between online and offline judgements to argue that the distinction between the a priori and the a posteriori is epistemologically superficial. Kuznetsov uses the traditional distinction to characterize my account of armchair philosophy. That is bound to be misleading, given how little I think of the traditional distinction.
As for the problem of defining 'intuition', an alternative strategy is to concede that ordinary non-reflective judgments based on perception are intuitions, but deny that they are philosophical intuitions. That too is unpromising. For what is distinctively philosophical about the judgment 'He doesn't know that it's 3 o'clock'? 'Know' is one of the commonest verbs in the English language. If such an everyday judgment counts as philosophical, it is hard to guess what would count as unphilo-sophical. Virtually any judgment can be used in a counterexample to some suitably wrong-headed philosophical theory.
To vary the example, for most adults the judgment '2+2 = 4' counts as intuitive in the psychologists' sense, since they do not base it on conscious reasoning. They also do not base it on sense perception. Moreover, '2+2 = 4' is philosophical in the sense that many philosophers of mathematics rely on the truth of such arithmetical equations in their arguments. I have certainly heard experimental philosophers define 'philosophical intuition' in a way that makes '2+2 = 4' a philosophical intuition. When the method of relying on'philosophical intuitions' is debated, are elementary arithmetical equations to be included?
The moral is this: do not use the word 'intuition' in debates on philosophical methodology unless you have properly clarified what you mean by it. Such clarification requires, at a minimum, answering the questions raised over the past few paragraphs.
2. Abductive Philosophy
In 'Armchair Philosophy', I characterized a broadly abductive methodology for philosophy. To emphasize that this need not give philosophy the character of a natural science, I cited the example of foundational inquiry within mathematics. Kuznetsov objects: 'Mathematics and philosophy are significantly different - the ontology of formal systems is known without a trace: we know all the basic laws of these systems'. But that is not true of foundational mathematics. As Kurt Godel and Paul
Cohen proved, neither Cantor's Continuum Hypothesis (CH) nor its negation is derivable from standard set theory (given the consistency of the theory). If CH is true, it is a basic law of set theory. If CH is false, its negation is a basic law of set theory. Either way, there is a basic law of which we are ignorant. Of course, on some views there are many set-theoretic universes, with CH holding in some and failing in others. Then the more basic framework is that in which we investigate the space of all set-theoretic universes. But then we do not know all the basic laws of that more general framework, for reasons connected with Godel's incompleteness theorems. Although there are many obvious differences between mathematics and philosophy, whether our knowledge has limits is not one of them.
Dennett's main concern with philosophers' use of an abductive methodology is that if they take intuitions as the input, the abductively derived outputs will be no more reliable than the inputs - unless the outputs are recycled as a theory about the content of the implicit folk theory which generated the intuitions, not as a theory about whatever the intuitions themselves are about. The radical unclarity of 'intuition' discussed in §1 clouds that concern too. Dennett mentions David Lewis in connection with an 'intuition'-based abductive methodology, but Lewis spoke of 'intuitions' just as our opinions, in describing something like the method of reflective equilibrium in philosophy, with no intention to exclude natural scientific opinions.
Dennett seems a little unfair to advocates of an 'intuition'-based abductive methodology when he describes them as 'taking their intuition-pumped consensus as a sure path to the "real nature" of whatever they were talking about'. His words'a sure path' suggest that they expect something like certainty from their methodology. But many of them would settle for a much weaker epistemic status, such as high rational credence. Dennett also flirts with a reading of a passage I quote from Austin as 'a complacent assurance that the time-honored, well-honed home truths of the manifest image are the last word on anything', but in that discussion Austin explicitly proposes that ordinary language should just be the first word on some things; he offers no candidate for the last word.
In my view, the conception of philosophical methodology as directed towards reflective equilibrium suffers from the usual defects of internalist and coherentist epistemology. It ignores crucial questions about where our evidence comes from. To discuss the methodology of natural science as directed towards reflective equilibrium without mentioning our interactions with the external world through observation and experiment would, rather blatantly, be to miss half the picture. Although the omission is less obvious when philosophical methodology is described in terms of reflective equilibrium, it is still there. Our knowledge of the world includes many findings of natural science; it also includes much else
besides. In principle, our evidence base for abduction in philosophy comprises all of that knowledge. In practice, parts coming from natural science are highly relevant to some philosophical questions; to ignore them would be foolish. But, again in practice, not all philosophical questions are like that. For example, the findings of natural science often have no distinctive relevance to abductive arguments for first principles of logic or mathematics, though there is no ban in principle on appeal to them even there. Sometimes, common sense knowledge is enough; sometimes, high-powered mathematical knowledge is needed. When things go well, we acquire knowledge (not just high rational credence) in the form of the abductive conclusions. It does not follow that the conclusions are the last word on anything. That something is known does not imply that no one is allowed to question it.
3. Models and Dependency Structures
In 'Armchair Philosophy', I proposed that philosophy, like much of natural science, often makes progress by constructing better models of matters of interest, rather than by discovering new universal laws of those matters. Of course, models in philosophy are usually not geared to making testable quantitative predictions, but the same applies to some models in natural science. For example, a model of evolution with three-sex rather than two-sex reproduction need not aim at making quantitative predictions: instead, its purpose may be to help explain why three-sex reproduction tends not to occur. Similarly, the purpose of models in philosophy tends to be explanation, not prediction. Kuznetsov seems to have an overly predictive conception of models when he writes 'Model building in science relies on empirical results and is mediated by them'.
Daniel Stoljar agrees that the conception of progress as the discovery of new universal laws is far too narrow for both philosophy and natural science, but he argues that it is for a more general reason as well: 'progress in both science and philosophy consists in the provision of better information about dependency structures'. Such structures may involve relations of either causal or constitutive dependence.
I was certainly not suggesting that discovering new universal laws and constructing better models are the only forms that progress in either philosophy or natural science can take. Nor have I anything against progress in either case by providing better information about dependency structures. However, I do not see what is so special about dependency structures. Progress in philosophy or natural science might be made by providing better information about almost any general kind of relational structure, whether they involve dependency relations or relations of some other sort.
Dependency relations typically involve an ordering, irreflexive (x does not depend on itself), asymmetric (if x depends on y, then y does not depend on x), and transitive (if x depends on y, and y depends on z, then x depends on z). But many relations of philosophical and natural scientific interest are not dependency relations. Logical relations, such as entailment, are an example. That p entails q tells us nothing about whether p depends on q, or q depends on p, or neither. For a start, the entailment may be mutual. Of course, we can rig up an irreflexive, asymmetric, and transitive relation of one-way entailment, where p one-way entails q just in case p entails q but q does not entail p. But it still implies nothing about dependency. For example, 'This is red and square' one-way entails 'This is red', where the temptation is to say that the entailer depends on the entailed, but 'This is red' one-way entails 'This is red or square', where the temptation is to say that the entailed depends on the entailer. Nevertheless, better information about entailment is often highly explanatory, in both philosophy and natural science. Something similar goes for mereological relations: to say that x is a proper part of y is not yet to say whether x depends on y, or y depends on x, or neither. Yet better information about parthood can be explanatory. In philosophy, better information about the existence, identity, and distinctness of things can also be explanatorily crucial, yet it is not naturally understood as information about a dependency structure.
The significance of progress by building better models is not that it is the only alternative to progress by discovering new laws, but that it is a different, widespread, and theoretically very powerful form of progress, distinctive of advanced natural science and, as it turns out, advanced philosophy too. How much progress in advanced natural science really consists of finding out more about dependency structures?
Список литературы / References
Dennett, 2019 - Dennett, D. "Philosophy or Auto-Anthropology?", Epistemology & Philosophy of Science, 2019, vol. 56, no. 2, pp. 26-28.
Knobe, 2019 - Knobe, J. "Philosophical Intuitions Are Surprisingly Robust Across Demographic Differences", Epistemology & Philosophy of Science, 2019, vol. 56, no. 2, pp. 29-36.
Kuznetsov, 2019 - Kuznetsov, A. "Armchair Science And Armchair Philosophy", Epistemology & Philosophy of Science, 2019, vol. 56, no. 2, pp. 43-45.
Nagel, 2012 - Nagel, J. "Intuitions And Experiments: A Defense Of The Case Method In Epistemology", Philosophy and Phenomenological Research, 2012, vol. 85, pp. 495-527.
Stoljar, 2019 - Stoljar, D. "Williamson On Laws And Progress In Philosophy", Epistemology & Philosophy of Science, 2019, vol. 56, no. 2, pp. 37-42.
Williamson, 2004 - Williamson, T. "Philosophical Intuitions and Scepticism About Judgment', Dialectica, 2004, vol. 58, pp. 109-153.
Williamson, 2007 - Williamson, T. The Philosophy of Philosophy. Oxford: Wiley-Blackwell, 2007, 346 pp.
Williamson, 20l3 - Williamson, T. 'How Deep Is The Distinction Between A Priori And A Posteriori Knowledge?', in: A. Casullo & J. Thurow (eds.). The A Priori in Philosophy. Oxford: Oxford University Press, 2013, pp. 291-312.
Williamson, 2018 - Williamson, T. Doing Philosophy: From Common Curiosity to Logical Reasoning. Oxford: Oxford University Press, 2018, 176 pp.
Williamson, 2019 - Williamson, T. "Armchair Philosophy", Epistemology & Philosophy of Science, 2019, vol. 56, no. 2, pp. 19-25.
