Aristotelian rhapsody: did Aristotle pick his categories as they came his way?

ABSTRACT

In the first Critique, Kant raises two objections against Aristotle’s categories. Kant’s concern, in the first instance, is whether Aristotle generated all categories that there are and if he did not generate any spurious categories. However, for Kant, this is only a symptom of the second – deeper – flaw in Aristotle’s thinking. According to Kant, Aristotle generated his categories ‘on no common principle.’ This paper develops the two Kantian objections, offers an overview of Brentano's (1862. Von der Mannigfachen Bedeutung des Seienden nach Aristoteles. Freiburg im Brisgau: Herder’sche Verlagshandlung.) reconstruction of Aristotle’s categories (which claims to have addressed them), develops three objections to this reconstruction, and recommends (Trendelenburg, A. 1846. Geschichte der Kategorienlehre. Berlin: Verlag von G. Bethge.) as a better – albeit still flawed – Aristotelian reply to Kant.

KEYWORDS:

• Franz Brentano
• categories
• being
• Aristotle
• Kant
• Trendelenburg

Acknowledgement

I would like to thank the anonymous reviewers of this paper for their generous feedback, which led to considerable improvements, and the audiences at the University of Vienna, the Complutense University of Madrid, the University of Edinburgh, and Maria Curie-Skłodowska University, where I presented versions of this paper, for stimulating discussions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 KrV: A81/B107.

2 Cf. Brentano 1981, 123.

3 Cf. Met.: III, 3, 998b21–27.

4 In Section 4.2 below.

5 Cf. Ward 2008, 168–171. In plainer terms: ‘a crane’ does not tell us much as an answer to the question ‘what is it?’ asked of a common crane (grus grus) unless we know whether this stands for a bird or for a hoisting machine.

6 Cf. Czerkawski 2022.

7 Cf. Brentano 1981, 98.

8 KrV: A81/B106-107.

9 For Brentano’s own account of the genealogy of his interpretation of the origin of the categorial scheme, see Brentano 1981, 118–123.

10 Cf. Studtmann 2008.

11 I suspect that the problems I will generate below for Brentano carry over to other names on Studtmann’s list – as well as to Studtmann’s own reconstruction of Aristotelian categories indebted to them – but the question of whether they do so indeed will have to be left until another occasion.

12 Cf. Cat.: 2, 1a20.

13 KrV: B159.

14 KrV: A64/B89.

15 ‘[T]here may perhaps be concepts which relate a priori to objects, not as pure or sensible intuitions, but solely as acts of pure thought […].’ KrV: A57/B81.

16 ‘These concepts we shall, with Aristotle, call categories, for our primary purpose is the same as his, although widely diverging from it in manner of execution.’ KrV: A79-80/B105.

17 KrV: A68/B93.

18 I am simplifying Kant’s argument considerably. What he says exactly is this: ‘[t]he same function which gives unity to various representations in a judgment also gives unity to the mere synthesis of various representations in an intuition; and this unity, in its most general expression, we entitle the pure concept of the understanding.’ KrV: A79/B104-5. For a discussion of some difficulties with this sentence, see Nussbaum 1990. In her influential interpretation of the metaphysical deduction, Longuenesse (2006, 142) summarises Kant’s reasoning as follows: ‘[i]f the understanding as a whole is [thus] nothing but a Vermögen zu urteilen [the capacity to judge], then identifying the totality of functions […] of the understanding amounts to nothing more and nothing less than identifying the totality of functions present in judging, which in turn are manifest by linguistically explicit forms of judgments.’ Cf. Longuenesse 2001.

19 Cf. KrV: A70/B95.

20 Cf. KrV: A80/B106.

21 KrV: A67/B92.

22 Note that some of the abovementioned concepts are not the Aristotelian categories proper but pertain to distinctions Aristotle wants to make within some of them. Cf. Studtmann 2012, 65–66.

23 For another illustration, any of Aristotle’s four causal relations – formal, material, efficient, and final – might arguably be thought to be instances of intelligibility as it is conceived by Kant, for they all seem to be variations on the theme of the category of cause and effect. And yet, they do not figure in Kant’s Table of Categories. Cf. KrV: A80/B106

24 KrV: A70-1/B95-6.

25 See especially Lotze 1884.

26 Cf. Longuenesse 2006, 154–157.

27 Cf. Longuenesse 2006, 157–161.

28 Cf. Met.: V, 7, 1017a7-1017b9; VI, 2, 1026a33 – b2.

29 Met.: III, 4, 1001a21.

30 Met.: III, 4, 1001a21.

31 I take it that this includes merely possible objects. I can conceive of some of them as well as of all actual objects. So, if the concept of being applied to actual objects only, it would be less general than the concept of things I can conceive of. But, for Aristotle, being is the most general concept. So, for Aristotle, the concept of being does not apply to actual objects only.

32 KrV: A81/B107.

33 KrV: A67/B92.

34 Thus, Categories and Topics list ten categories, whereas there are only eight in Posterior Analytics and Metaphysics, Delta and as few as four in Metaphysics, Nu. Cf. Cat.: IV, 1b25-2a4; Top.: I, 9, 103b22; APo, I, 22, 83b15; Met.: V, 7, 1017a22-30; XIV, 2, 1089b18-25.

35 Cf. Studtmann 2012, 67.

36 Ackrill 2002, 78.

37 For an earlier version of this interpretation, see Brentano 1981, 128–129 (who, in turn, adapts it from Occam).

38 KrV: A81/B106-107. Cf. KrV: A66/B91.

39 KrV: A81/B107.

40 Met.: III, 4; 1001a21.

41 Brentano 1981, 98.

42 There are actually two ways in which this may happen. See Brentano 1981, 58–66.

43 In his recent article on the same subject, Respa (2021, 196) claims that, ‘following Bonitz, Brentano interprets being as a supreme genus and on that basis provides a deduction of the categories.’ However, this claim stands on an ungenerous – and extremely implausible – reading of the remark Brentano (1981, 96) makes about Metaphysics, Gamma 2: ‘Here again [Aristotle] speaks of the kinds [Arten] of being and the corresponding kinds of the one as if being were a genus [nicht anders als ob das Seiende eine Gattung wäre]; and these so-called kinds [diese sogenannten Arten] are, of course, the categories.’ (Cf. Brentano 1862, 147, Respa 2021, 194.) As if. So-called. These remarks are part of Brentano’s discussion of the famous passage in Metaphysics, Gamma 2 where Aristotle argues that, even though being is homonymous – i.e., objects the corresponding term picks out do not belong in any single γένος – they are still sufficiently unified to be addressed by a single science. Since, for Aristotle, the subject-matter of a single science normally forms a γένος, Brentano is free to make an entirely uncontroversial point that beings in Aristotle’s categorial sense enjoy γένος-like unity, which, as we have just noted, he ultimately analyses in terms of the concept of ‘analogy’. Nothing in Brentano’s texts suggests that he misunderstands Aristotle on this basic point. I am also wondering how much sense Respa can make of the passage I have excerpted above and why, on his interpretation, Brentano’s divisions stop when they do – why does he stop at substances, for example, but not at accidents?

44 Cf. Cat.: 2, 1a20.

45 Cat.: 2, 1a20.

46 ‘[F]urther real divisions of substance […] will have to take the form of divisions of a univocal concept through added differentiae [diaphorai] in the manner of a specification in the narrower sense.’ Brentano 1981, 98. Brentano conflates – not unproblematically – univocity with the possession of a γένος. Cf. Loux 1973.

47 Brentano 1981, 99.

48 Brentano 1981, 99.

49 Cf. ‘These absolute accidents also do not all seem to be predicated of substance in the same way.’ Brentano 1981, 100. I should note that Brentano recognises other technical uses of the term ‘πάθη’ in Aristotle. Cf. Brentano 1981, 100.

50 Brentano 1981, 99.

51 Brentano 1981, 99.

52 Brentano 1981, 101–103.

53 Brentano 1981, 109–114.

54 Brentano 1981, 103–104.

55 But see Brentano 1981, 107–112.

56 Brentano 1981, 99.

57 One might think that Brentano’s procedure forces him, so to speak, to conclude his deduction with a flourish of inductive brainstorming and that therein lies an even greater flaw of his approach. It might seem that, if Brentano were to work only by identifying purely logical distinctions between F and not F (and, in one case, partly F and partly not-F), then there will be always something that does not fall under a γένος, but is simply characterised negatively as ‘not something’. If so, the real problem with Brentano’s derivation isn’t that it in fact ends on inductive note, but that it condemns itself to doing so on pain of never finishing its divisions. To this graver charge, however, Brentano has a good reply, I think. Conceptualising something as ‘not something’ surely does not preclude its positive characterisation. The sexual difference has two poles, so we can conceptualise male as not-female and female as not-male. But this does not mean the biologist can only offer a positive characterisation of one of these poles and treat the other derivatively. Both are independent objects of scientific study and thus allow for positive characterisation. So, it seems to me that Brentano’s approach survives this objection.

58 Brentano 1981, 99.

59 Cf. Met.: IV, 2, 1003a34-1003b6.

60 One might think that Brentano’s failure to offer an explicit definition of a γένος affects all steps of his proof (given his general approach) rather than, as I propose, merely the second one and the remainder. But it seems to me that nothing stops Brentano from explaining that the standard understanding of a γένος above has been implied all along, and us – from accepting this explanation.

61 ‘that can occur’

62 An anonymous reviewer of this paper proposed to me that a ‘qualitative’ interpretation of Brentano’s claim above is more plausible than the ‘quantitative’ one I have just offered. The reviewer phrases the qualitative interpretation as follows: ‘there is more of a difference between any substance and any accident than there is between any two accidents.’ I have two things to say about this (besides reiterating my own – opposed – judgment). First, I’m not sure if the alternative is genuinely qualitative. For, as long as we want to say that there is more of something here than there, we are surely invoking quantities. It is true that these need not be, as I propose, quantities of concrete objects themselves. They could also be quantities of parts of concrete objects (in a way an apple, for example, could enjoy more redness than another or the couple of Romeo and Juliet could be more in love than the former couple of Brad Pitt and Angelina Jolie). Still, they would be quantities no less. Second, I believe that, insofar as quantities are involved, a version of each of my three objections could be formulated for the reviewer’s variant, too. However, while my objections 1 and 3 transfer rather easily, the translation of the second objection would take up a lot of additional space, making this paper unnecessarily long and technical. Note that difference in the relevant sense is a relational property like being in love above (predicated of tuples rather than of individuals) rather than a simple property like redness above. So, spelling out what it might actually amount to – i.e., what kind of relational property it is – would require some formal machinery. And spelling out the new variant of the objection would involve a lot of clunkiness.

63 Brentano’s (1981) concern with what follows – beyond what Trendelenburg notes himself – is just that it’s not the whole story.

64 See especially Bonitz 1853.

65 For the place of ‘Urteil’ and ‘Satze’ in Aristotle’s thinking, see Chapter 5 of Trendelenburg 1846 (13–18).

66 Cf. Trendelenburg 1846, 25, 33, 180, 216.

67 Cat.: 4, 1b25.

68 All translations from Geschichte der Kategorienlehre are mine.

69 Trendelenburg 1846, 12. For more textual evidence for Trendelenburg’s grammatical interpretation see Trendelenburg 1846, 24–33. For a range of interesting objections, see Bonitz 1853, and, for some responses, Brentano 1981, 126–128.

70 Cf. ‘[Aristotle] thought he was defining the attributes of objects but he was really setting up linguistic entities; it is the language which, thanks to its own categories, makes them to be recognised and specified.’ Benveniste 1971, 61.

71 Cf. Trendelenburg 1846, 23.

72 One might object that any evidence for Chomsky’s theory of universal grammar is at the same time evidence against this claim. But Chomsky does not seem to deny that this universal grammar is contingent on our biology (he claims that it is innate, but ‘innate’ falls short of ‘necessary for language’ – it falls short of saying that, had our biology been a little different in the relevant respect, we would no longer be speaking a language). Nor does he seem to think that this universal grammar (hence, our shared biology) completely determines the shape of any natural language. If it did, whence the variation among the natural languages? Cf. Chomsky 1986.

73 In the words of Brentano (1981, 123), ‘a procedure which … has to rely on mere agreement with grammatical relations cannot escape being reproached for its superficiality.’

74 An interesting question to consider here, though, is whether, at the end of the day, Kant’s metaphysical deduction does not meet the same fate. Frege (1967) would certainly argue that it does. For Frege, as his view is reconstructed by Longuenesse (2006, 158), ‘Kant’s subservience to the traditional, Aristotelian model of subject-predicate logic is grounded on [a] confusion. For the subject-predicate model really takes its cue from the grammatical structure of sentences in ordinary language [which typically have a subject and a predicate]. And ordinary language is itself governed by the subjective, psychological intentions and associations of the speaker addressing a listener. But […] what matters to logic are the structures of thought that are relevant to valid inference, nothing else. Those structures, for Frege, include the logical constants of propositional calculus (negation and the conditional), the analysis of propositions into function-argument rather than subject-predicate, and quantification.’ Thus, for Frege, Kant’s own Leitfaden is hardly any less grammatical than the one Trendelenburg finds in Aristotle. It just focusses on a different aspect of the grammar of a natural language.