The Grounds of a Critique of Pure Reason

ABSTRACT

For the realist metaphysician, certain notions in metaphysics are objectively theory-guiding. But what makes them so? Echoing others, Dasgupta (2018) suggests that the realist metaphysician faces trenchant difficulties here—resulting in the problem of missing value. I first propose that Kant’s project of a critique of pure reason faces this problem: he supposes that the notion of ground is objectively theory-guiding in metaphysics. This investigation reconstructs his response. I argue that, for Kant, a notion is objectively theory-guiding in metaphysics if (and because) it is required for the successful exercise of reason. By analysing the requirements on undertaking a critique of pure reason, we will see why his notion of ground would be objectively theory-guiding in this sense. I conclude that his response generalizes to yield a potentially promising response to the problem of missing value.

KEYWORDS:

• Kant
• Metaphysics
• Critique
• Reason
• Ground
• Normativity

Acknowledgments

Thanks to Rosalind Chaplin, Damian Melamedoff, Clinton Tolley, Bas Tönissen, Eric Watkins, several anonymous referees, and audiences at Florida State University, the University of Pennsylvania, and UC San Diego for invaluable feedback.

Disclosure Statement

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

Notes

1 Recall that x is grue ≝ (x is examined before time t and green) ∨ (x is not so examined and blue). All citations from the Critique of Pure Reason follow the standard A/B-edition pagination. All translations of Kant’s published works are taken from the Cambridge Editions of Kant's works (including Kant 1996, 1997, and 1998), unless noted otherwise. All translations of Kant’s unpublished works (found in Kant 1900–) and of other figures are my own, unless noted otherwise.

2 Dasgupta (2018) traces the problem back to earlier discussions in metaethics, e.g., in Enoch (2006) and Dreier (2015).

3 Kant sometimes characterizes metaphysics as concerned with real grounds, rather than logical grounds. I will bracket this distinction here. For discussion of Kant’s extensive use of grounds (and of this distinction), see Langton 1998, Longuenesse 2005, Watkins 2005, Hogan 2009, Smit 2009, Proops 2010, Anderson 2015, Stang 2019, and Watkins 2019.

4 Willaschek (2018: 82), Stang (2019: 81), and Watkins (2019) advance the former view; Langton (1998, 198) advances the latter view.

5 This definition is also found at AK 17:28, AK 18:118, AK 28:401, AK 28:408, and AK 29:818.

6 Stang (2019) claims that the notion of positing itself presupposes the notion of ground, and thus cannot be used to define the latter. Yet Kant's metaphysics lectures indicate otherwise: ‘But there are cases where something is posited, and another thing is posited after, yet where the one is not a ground of the other. E.g., when the stork comes, good weather follows.’ (AK 28:549 [∼1790/1]). Although the arrival of the stork does not ground the arrival of good weather, there is a positing relation between them. This suggests that the positing relation is rather weak (plausibly no stronger than a material conditional). It is therefore not circular to use positing as part of a definition of ground.

7 Cf. Woodward 2021. A deductive-nomological notion of ground has garnered recent attention. Cf. Kment 2014, Wilsch 2015, Rosen 2017, and Schaffer 2017.

8 For discussion of the idea that causation is a specific kind of grounding (a position ubiquitous among Kant and his German rationalist predecessors), see Watkins 2005 and Stang 2019.

9 I shall treat Kant’s two core conditions on ground (the positing condition and the universal rule condition) as definitionally necessary conditions on ground below. Yet to capture the idea that the consequence obtains in virtue of its ground, additional conditions (or at least a further specification of the universal rule condition) are evidently needed. To use the above example, a figure’s trilaterality grounds its triangularity. Yet triangularity also seems to satisfy the two core conditions of the nomological notion of ground, since the converse rule also holds (any triangular figure is trilateral). I surmise that Kant appreciates this need for additional conditions, but I leave them for discussion elsewhere. Cf. AK 11:36, AK 28:399, AK 28:489, AK 28:629, and AK 29:809. This issue is parallel to the problem of asymmetry plaguing the deductive-nomological notion of explanation.

10 As Wolff (1730) puts it, ‘a proposition that articulates a determination in conformity with a ground is called a rule.’ (Ontologia §475). Cf. Wolff (1730, §56) and Baumgarten (1730, §14 and §80–3). Kant’s unpublished notes and lectures sometimes leave it difficult to tell where his summary of Wolff and Baumgarten ends and where his own position begins. But since their positions in this case oppose his above definition of ground, he cannot simply be summarizing their positions here. One might still harbour worries about relying on unpublished works to ascribe the nomological notion of ground to Kant. Yet our investigation will attest that this ascription is not only consistent with his published critical works, but also explicable in terms of their core doctrines.

11 Cf. Cf. Axi-Axii, A11/B24–5, A13/B27, A751/B779, A758/B786, A761/B789, A836–7/B864–5, and AK 5:167.

12 Cf. A126–7, A302/B359, A307/B363–4, A797–8/B825–6, AK 5:119–20, AK 5:184, and AK 5:187. For recent discussion, see Willaschek 2018, Schafer 2019, and Tolley 2020.

13 Cf. A303–9/B359–65. Kant attributes to the faculty of understanding a limited power of immediate inference (A303/B360). Such inferences involve only a single premise (e.g., from all $X$s are $Y$s to no $X$s are not $Y$s). I will bracket consideration of immediate inferences below.

14 Cf. A302/B358–9 and A330/B387. More on this in §4.

15 Cf. A841–42/B869–70 and A849–51/B877–9.

16 (Obj) is a sufficient condition on objective theory-guidingness, not a necessary one. On the one hand, it does not specify what makes notions that are tied to other faculties objectively theory-guiding in metaphysics. On the other hand, it leaves open that notions can be objectively theory-guiding in metaphysics because they stand in some other relation to the faculty of reason (e.g. because they serve the interests and aims of the faculty of reason, rather than merely a condition of its successful exercise). For recent discussion of the latter point, see Willaschek 2018 and Schafer 2019.

17 Cf. Bxxiv-Bxxix and A850–1/B878–9.

18 One remaining question: does accepting a contradiction amount to an unsuccessful use of reason, or does it fail to amount to a use of reason at all? For discussion, see MacFarlane 2000 and Tolley 2006.

19 Admittedly, the self-consistency constraint is rejected by some contemporary Kantians. Korsgaard (1989) argues that reason should hold onto its practical commitment to freedom, even if it contradicts reason’s theoretical commitments. Cf. Rawls (1975) and O’Neill (1989). Nonetheless, I surmise that Kant endorses the self-consistency constraint. He explicitly claims that reason’s practical commitment to freedom is tenable only if theoretical reason is not forced to concede the impossibility of freedom (Bxxix-Bxxx). Cf. Rauscher (1998).

20 Cf. Axi-Axii, Bxii-Bxvi, A11/B24–5, A13/B27, A396, and A836–7/B864–5.

21 One source of this neglect is the failure to adequately distinguish boundaries and limits. Many systematic reconstructions of the Critique—e.g., Strawson (1966), Grier (2001), Allison (2004), and Allais (2015)—do not distinguish them. But this neglect extends to those who do distinguish them (e.g., Callanan 2021 and Howard 2022).

22 Cf. A395, A744/B772, A758–9/B786–7, A761–2/B789–90, A842–3/B870–1, AK 4:352, AK 4:360–1, and AK 5:188.

23 Howard (2022) likewise identifies necessity as a distinguishing feature of boundaries. Callanan (2021) and Howard (2022) both identify other features of boundaries. Notably, both insist that the boundary of a domain belongs to that domain without being a part of it (AK 4:360–1). So precisely as Kant suggests (A841/B869), an investigation of the boundaries of reason’s cognitive powers in metaphysics belongs to metaphysics without being a part of it.

24 Cf. A423–4/B451–2, A507/B535, and A761–9/B789–97.

25 Cf. A758–9/B786–7, A761–3/B789–91, A768–9/B796–7, and A842–3/B870–1.

26 Kant speaks of ‘a priori grounds’ above. For discussion of the connection between a priori cognition and cognition from grounds, see Adams 1994, Hogan 2009, and Smit 2009. One might accordingly propose that Kant’s rationale instead hinges on a connection between a prioricity and necessity: whatever is cognizable a priori is necessary, and thus in so far as reason’s boundaries are cognizable a priori, they must be necessary. Although I am sympathetic to this proposal, defending it would require elucidating the vexed connection between a prioricity and necessity—which (on my view) would give way to the assumptions about grounds discussed below anyhow.

27 For Kant’s articulation of this assumption for grounds in general, see AK 17:28, AK 18:118, AK 28:401, AK 28:408, and AK 29:808–9.

28 On Hogan’s (2009) reading, Kant denies that all grounds necessitate their consequences—free actions are not necessitated by their grounds. Even if Hogan’s reading is correct, it would not undermine my argument here. For I am merely claiming that the grounds required for a critique of pure reason to determine reason’s boundaries are necessitating grounds. This is compatible with the possibility of non-necessitating grounds. With that said, if the notion of ground that is required for a critique of pure reason were the only possible notion of ground, that would indeed exclude the possibility of non-necessitating grounds. I leave open whether the antecedent of this conditional should be denied, or if its consequent ought to be affirmed.

29 Kant glosses absolute necessity earlier as follows: ‘That whose opposite is internally impossible, that whose opposite is clearly also impossible in all respects, is therefore itself absolutely necessary.’ (A325/B381–2). For discussion of this and other notions of absolute necessity in Kant, see Stang 2016.

30 Cf. A321–3/B378–80, A333/B390, A642–3/B670–1, A763/B791, and A834/B862.

31 The successful exercise of reason’s cognitive powers may require the operation of other cognitive faculties, including the understanding and sensibility (A306/B363). It is therefore unsurprising that the Critique investigates those faculties.

32 Cf. Axi, B421, A735/B763, and A850–1/B878–9.

33 A complete reconstruction would require showing how reason’s cognitive powers, in turn, satisfy Kant’s general constraints on cognition. Fortunately, our comparatively narrow aim does not require this. Though for recent discussion of these constraints, see Watkins and Willaschek 2017 and Schafer 2023.

34 Cf. A300–2/B357–9, A306–7/B363, and A330/B387 (quoted below).

35 Cf. A330/B386, A333–5/B390–2, AK 4:459, AK 5:412, AK 9:65, AK 16:343–4, AK 16:95, AK 18:417–8, AK 24:50, AK 24:539, and AK 24:730–1.

36 After inferring a connection among particulars via a rule, reason could inferentially cognize further particulars. For instance, after inferring $\mathrm{Ga}$ from cognition of (i) $\mathrm{\forall }x(\mathrm{Fx}\to \mathrm{Gx})$ and (ii) $\mathrm{Fa}$, reason could inferentially cognize further particulars from $\mathrm{Ga}$. This does not undermine the above point, however, that reason’s cognition of particulars must ultimately proceed from (discursive) rules.

37 Cf. B4, A91–2/B124, A306–7/B363–4, A646–7/B674–5, A713–4/B741–2, and A837/B865. Kant himself distinguishes multiple senses of the term ‘principle’ (A299–300/B356–7). This terminological issue is tied to a substantive one about what exactly distinguishes principles from other (strictly) universal rules. Fortunately, nothing hinges on this issue here.

38 Cf. AK 17:28, AK 18:118, AK 28:401, AK 28:408, and AK 29:808–9.

39 This pair of points applies, mutatis mutandis, to those who treat Kant’s notion of ground as primitive.

40 Cf. A11/B24–5, A13/B27, A761/B789, A836–7/B864–5, and AK 5:167.

41 Addressing (i) would require a systematic reconstruction of the first Critique. Addressing (ii) would require showing that the rest of metaphysics must be concerned with rational cognition (rather than some weaker attitude) of connections of grounding (rather than some other kind of connection). I take it that (ii) is closely tied to reason’s essential aim of ascertaining unconditioned grounds (Bxix-Bxx). For discussion, see Grier 2001, Willaschek 2018, Schafer 2019, and Watkins 2019.

42 For one response, see Sider (2022).

43 One might even expand the relevant notion of phenomenal acquaintability to include what is phenomenally acquaintable to non-humans.

44 One might take issue with Kant’s claim that cognitive capacities belong to underlying faculties (sensibility, understanding, reason, etc.). Hence my framing here in terms of capacities, rather than faculties. Though see Schafer 2019.

45 Notions that are phenomenal acquaintable might well satisfy the above scheme (though this would require a separate argument).

46 So construed, would objectively-theory guiding notions still include those used from God’s point of view (à la traditional realism)? That depends on how God’s intellect is construed. For his part, Kant holds that God would have a fundamentally different kind of intellect—intuitive (rather than discursive). Cf. AK 5:401–10, AK 28:996, AK 28:1017, and AK 28:1051–3. Since God’s intellect would grasp all truths intuitively (and thus immediately), God would not even need to undertake metaphysics (taken as an inferential enterprise) to grasp the sorts of truths pursued by metaphysics.

47 Cf. Priest, Tanaka, and Weber (2018). Likewise, one might worry that the cognitive capacities necessary for undertaking metaphysics can only be ascertained by means of undertaking metaphysics (rather than via an antecedent specification of its conditions and aims à la Kant). This would threaten the idea that objectively theory-guiding notions can guide metaphysics from the outset. For discussion of this Hegelian worry, see Watkins 2014.

48 And, of course, one might ask why objective theory-guidingness in metaphysics (rather than in schmetaphysics) is what matters. Although this question is not unfair, to ask it is to ask a different question than the Kantian realist approach (as construed here) purports to answer. Though for relevant discussion of this issue in Kant, see Willaschek 2018 and Schafer 2023.