Kant on Common Sense and Empirical Concepts

Abstract Kant’s notion of common sense (Gemeinsinn) is crucial not only for his account of judgements of beauty, but also for the link he draws between the necessary conditions of such judgements and cognition in general. Contrary to existing interpretations which connect common sense to pleasure, I argue that it should be understood as the capacity to sense the harmony of the cognitive faculties through a sui generis sensation distinct from pleasure. This sensed harmony of the faculties is not only the ground of judgements of beauty and the basis of pleasure in the beautiful, but is also essential, I argue, for the reflecting judgements through which we acquire empirical concepts.

project of the third Critique, which seeks to discover the a priori principles that govern judgement as such.
In section 1, I set the stage for my account of common sense by responding to some preliminary worries about Kant's discussion. In section 2, I show that there are a number of reasons not to identify what is sensed through common sense with the feeling of pleasure, as many interpreters do. Instead, I argue that the bulk of the textual evidence supports an alternative account, according to which common sense is the capacity to sense the harmony of the cognitive faculties through a sui generis feeling distinct from pleasure, a feeling which, on my reading, is the basis of pleasure in the beautiful and the ground of judgements of beauty. Armed with this understanding, in section 3 I examine Kant's claim that common sense is necessary for the universal communicability of cognition, a claim that is sometimes overlooked because of the tendency to identify common sense with pleasure in the beautiful. I argue that, for Kant, common sense is necessary not only for judgements of beauty but also for the reflecting judgements by which we acquire empirical concepts. As such, it provides a key link between these two seemingly disparate types of judgement that Kant focuses on in the Critique of Judgement.
1 Kant introduces the idea of a common sense in the Fourth Moment of his exposition of judgements of beauty. In §20, he first claims that judgements of beauty can be made only under the presupposition of a common sense. Next, he appears to answer in the affirmative the question that makes up the title of §21: 'Whether one has good reason to presuppose a common sense'. Setting aside the details for now, we can sketch his argument as follows:  1. Cognition requires that the cognitive faculties (i.e. the imagination and the understanding) be brought into the optimal 'proportion (Proportion)' or 'inner relationship (innere Verhältniß)'. 3 2. The optimal proportion of the cognitive faculties cannot be determined through concepts. 3. So, the optimal proportion of the cognitive faculties must be determined through feeling. 4. If the optimal proportion of the cognitive faculties and the feeling accompanying it were merely private, then cognitions would be merely private. 5. Cognitions are not merely private, i.e. they are universally communicable. 6. So, the optimal proportion of the cognitive faculties and the feeling accompanying it are universally communicable. 7. A universally communicable feeling presupposes a shared capacity to sense it. 8. So, there is good reason to presuppose a common sense.
I postpone further discussion of this argument until I present my own reading of common sense. First, I want to consider a question concerning the aim of §21 that has divided interpreters: does Kant intend the Optimal Proportion Argument to be a 'deduction' of judgements of beauty? That is, is Kant's claim that the optimal relationship of the faculties is determined through a universally communicable feeling in cognition meant to play a role in justifying the claim to universal validity made by judgements of beauty? Given the question in the title of §21, as well as the broad similarity between Kant's answer and the argument of the official deduction in §38, many interpreters believe that the Optimal Proportion Argument is a first attempt at a deduction. 4 A problem for this view, however, is that Kant concludes the very next section, §22, as follows: Whether there is in fact such a common sense, as a constitutive principle of the possibility of experience, or whether a yet higher principle of reason only makes it into a regulative principle for us first to produce a common sense in ourselves for higher ends : : : this we would not and cannot yet investigate here : : : (5: 240) Here, Kant indicates that he does not take himself to have investigated, much less settled, two related questions: whether there is a common sense, and whether the normative claim on the feelings of others that is made by a judgement of taste (i.e. beauty) is justified in itself or only expresses an independent interest in the possibility of intersubjective agreement. This suggests that Kant does not take §21 to contain a deduction of judgements of beauty that appeals to the role played by common sense in cognition. For if he did, how could he continue to regard these questions as unresolved in §22?
This is a primary reason that some interpreters deny that the Optimal Proportion Argument is a deduction of judgements of beauty and, indeed, that it is even concerned with such judgements. According to Allison and Matherne, for example, the common sense discussed in §21 is not the one said to be presupposed by judgements of beauty in §20, but a distinct species of common sense that Kant believes is required for cognition. 5 Kant brings up this 'epistemic common sense', on this reading, because its existence provides indirect support for the presupposition that there is also an 'aesthetic common sense', although it does not directly confirm it. And this is said to be why Kant continues to take the question about aesthetic common sense to be unresolved in §22, since he does not provide direct proof that there is an aesthetic common sense until the official deduction in § §38-40.
However, the suggestion that Kant posits two distinct species of common sense across § §20-2 strikes me as implausible for a few reasons. First, both §20 and §22 are explicitly about the role common sense plays in judgements of beauty, and so this interpretation must read Kant as switching to an entirely different common sense in §21 and back again in §22, without giving any indication that he is doing so. Second, the claim that the discussion of the conditions of cognition in §21 is not directly related to judgements of beauty ignores the many occasionsincluding in the official deduction itself (5: 290)on which Kant explicitly identifies the conditions of making judgements of beauty with the conditions of cognition in general. 6 Here is just one example: [P]leasure [in beauty] must necessarily rest on the same conditions in everyone, since they are subjective conditions of the possibility of a cognition in general, and the proportion of these cognitive faculties that is required for taste is also requisite for the common and healthy understanding that one may presuppose in everyone. For this very reason, one who judges with taste : : : may also require : : : his satisfaction in the object, of everyone else : : : (5: 292-3, my  emphasis) 7 In this passage, Kant says that the 'proportion' of the cognitive faculties that is required for judgements of beauty is also required for cognition in general. This strongly suggests that he does not intend his discussion of the optimal 'proportion' in §21 to be limited to cognitive judgements, but to describe what is required for both judgements of beauty and cognition. 8 Third, a closer look at the question Kant says is left open in §22 undercuts the very motivation for distinguishing two species of common sense: Th[e] indeterminate norm of a common sense is really presupposed by us: our presumption in making judgments of taste proves that. Whether there is in fact such a common sense, as a constitutive principle of the possibility of experience : : : this we would not and cannot yet investigate here : : : (5: 239-40, my emphasis) Two points are relevant here. First, recall that one of the main reasons to introduce a distinctively epistemic common sense is to explain how Kant can continue to claim in the above passage that the question of whether there is an aesthetic common sense is unresolved. But the question that Kant says is left open is not whether there is an aesthetic common sense in particular, but whether there is a common sense that is constitutive of the possibility of experiencein other words, precisely what would be considered the epistemic common sense. But this then rules out the suggestion for reconciling § §21-2, according to which Kant confirms the existence of an epistemic common sense in order to provide indirect support for an aesthetic common sense. Second, the passage clearly identifies the common sense that is 'presupposed by us' when we make judgements of taste as itself 'a constitutive principle of the possibility of experience'. As I see it, this tells decisively against the proposal that there are distinct aesthetic and epistemic species of common sense discussed in these sections. 9 If there is only one species of common sense under discussion across § §20-2, however, we are again faced with the question of why Kant claims in quick succession both that there is good reason to presuppose a common sense and that he has not yet investigated whether there is one. We can resolve this puzzle, in my view, by attending to the difference between the goals of the Analytic of the Beautiful and the Deduction in §38. In the Analytic, Kant seeks to provide an 'exposition' (5: 279) of judgements of beautythat is, a description of what is claimed by particular subjects that make such judgements. In the Deduction, in contrast, Kant considers whether subjects are justified in these claims. Now, Kant's discussion in § §20-2 occurs in the Fourth Moment, within the Analytic of the Beautiful. And he makes clear that his point there is simply that particular subjects in fact presuppose a common sense when they make judgements of beauty. As he puts it, a common sense 'is really presupposed by us: our presumption in making judgments of taste proves that ' (5: 239-40). Whether subjects are justified in this presupposition, he continues, will not be investigated in the midst of his exposition: 'for now', he says, 'we have only to resolve the faculty of taste into its elements' (5: 240).
The puzzle, of course, is how §21with its title question, 'whether there is reason to suppose a common sense'fits into the expository goals of the Analytic. Since the question appears to be whether the presupposition of a common sense is justified, many interpreters take §21 to contain a first attemptalbeit a poorly located oneat a deduction. I think this is incorrect. We can distinguish two questions: first, whether particular subjects who make judgements of beauty have available to them, from their point of view, any grounds for believing in something like a common sense; second, whether there are philosophical and, indeed, transcendental grounds for doing so. As I see it, it is the former question that Kant asks in §21 and the latter that he puts off until the Deduction. 10 To further support this reading, note that the former question arises quite naturally at the point that Kant raises it. He has just explained that subjects who make judgements of beauty expect everyone to feel the pleasure they do, and that they make this surprising demand on others' feelings by presupposing a common sense.
Here it seems reasonable to ask why subjects would presuppose this: do they have any reasons available to them for doing so? In response, Kant answers that they do, since it seems to them that they also rely on something like a common sense in cognition. I put off the question of why this is the case until section 3. For now, I wish to argue only that Kant's claim that particular subjects may take themselves to have reason to suppose a common sense is compatible with his denial that he has himself philosophically resolved the matter.
On my view, then, it is one and the same common sense that is presupposed by judgements of beauty and related to the conditions on cognition; Kant introduces it as part of his exposition of judgements of beauty in the Analytic, and goes on to confirm its existence in §40. Next, I present my interpretation of what this common sense amounts to.

2
To begin to understand what Kant means by common sense, it will be helpful to recall the conditions under which judgements of beauty are made. According to Kant, a judgement of beauty and the corresponding pleasure in the beautiful are grounded in the harmonious free play of the cognitive faculties of imagination and understanding. The harmony of the faculties also obtains in the case of cognition, since on Kant's view cognition requires that the imagination synthesize the manifold of intuition in accord with concepts of the understanding. Thus, both faculties must harmoniously cooperate to generate a cognition with both intuitive and conceptual content. In the case of beauty, however, the harmony obtains while the faculties are in free play: although the imagination is 'productive and self-active' (5: 240), its synthesis is nevertheless found to be in accord with the understanding; as Kant puts it, 'the former in its freedom is in harmony with the latter in its lawfulness ' (5: 287). This occurrence of harmony while the faculties are in free play results in a feeling of pleasure that is judged to be universally valid. Now, despite their other differences, interpreters assume that the 'sense' in Kant's common sense refers to the feeling of pleasure in the harmony of the faculties. 11 In other words, they tie common sense essentially to the feeling of pleasure in the beautiful. On my view, there are a number of reasons not to do so. First, this identification is difficult to square with the role common sense actually plays in judgements of beauty: Kant describes it in §20 as a 'principle : : : which determines only through feeling and not through concepts, but yet with universal validity, what pleases or displeases' (5: 238, my emphases). 12 Note that common sense is said to determine 'through feeling : : : what pleases' (I set aside the case of displeasure). If we identify the former feeling with pleasure, however, this sentence becomes difficult to interpret: it amounts to the seemingly trivial claim that common sense determines through pleasure what pleases. 13 Moreover, Kant's phrasing here is not accidental. In §40, in which he describes taste itself as a common sense, he defines it as 'the faculty for judging a priori the communicability of the feelings that are combined with a given representation (without the mediation of a concept)'. Here too, then, common sense is tied not directly to pleasure itself, but to a capacity that enables us to judgewithout appeal to concepts, and thus through feeling (5: 239)whether the pleasure that accompanies a representation is universally communicable. This suggestsas I will argue belowthat what is directly sensed by common sense is not the feeling of pleasure, but rather another feeling that grounds the claim to the universal validity of pleasure.
A second reason to resist tying common sense to pleasure is that, as we have seen, Kant claims that the optimal 'proportion' or harmony of the faculties sensed by common sense obtains not only in judgements of beauty, but in cognition in general (5: 292-3). 14 But if this harmony were sensed through pleasure, this would entail that all cognition is pleasurable, a seemingly implausible thesis that most interpreters agree Kant is not committed to. 15 Third, Kant argues both that the feeling sensed through common sense is the ground of judgements of beauty, and that pleasure cannot be the ground of a judgement of beauty. The following passage states the former point clearly: In all judgments by which we declare something to be beautiful, we allow no one to be of a different opinion, without, however, grounding our judgment on concepts, but only on our feeling, which we therefore make our ground not as a private feeling, but as a common one. Now this common sense cannot be grounded on experience : : : (5: 239) In contrast, as has been well-discussed, Kant raises a question in §9 that he claims is 'key to the critique of taste'namely, 'whether in the judgment of taste the feeling of pleasure precedes the judging of the object or the latter precedes the former' (5: 216)and argues there that pleasure cannot be the ground of a judgement of beauty, but rather must be its consequence (5: 217). As for the 'determining ground of the judgment' of beauty, he identifies it as the 'feeling of the free play of the powers of representation : : : for a cognition in general'. Thus, Kant maintains both that (i) the ground of a judgement of beauty is a common feeling of the relation between the cognitive faculties; and (ii) the feeling of pleasure is not the ground but a consequence of a judgement of beauty. If we take these passages at face value, it follows straightforwardly that the feeling sensed through common sense is not identical to pleasure. 16 Admittedly, Kant does appear to link common sense to pleasure on one occasion. While rejecting the colloquial description of certain intellectual judgements as 'common sense', Kant points out that concepts like truth or justice utilized by such judgements cannot be expressed through a sense (5: 293). Instead, he notes that 'if one would use the word "sense" of an effect of mere reflection', it is aesthetic judgements that are better referred to as common sense, since 'there one means by "sense" the feeling of pleasure' (5: 295). As the arguments above demonstrate, however, the bulk of Kant's text resists the identification of common sense with pleasure. Thus, I suggest we read the above sentence as making the narrower point that the colloquial tendency to describe certain judgements as common-sense judgements is better applied to aesthetic judgements than intellectual judgements, since only the former are expressed through a sensationnamely, pleasure.
To recapitulate, although Kant clearly connects common sense to a feeling, there are a number of reasons to conclude that this feeling is not identical to pleasure. First, the feeling in question is itself meant to play the role of determining what pleases or displeases; second, the feeling and the state it accompanies are said to be conditions on cognition in general, and cognition in general is not pleasurable; third, the feeling is said to be the ground of judgements of beauty, and Kant insists that pleasure is not the ground of judgements of beauty but rather its consequence.
In fact, there is direct textual evidence that two distinct feelings play an essential role in Kant's account of judgements of beauty: the first, a sui generis feeling of the harmony of the faculties that is not itself pleasurable; the second, the feeling of pleasure itself. The following passage clearly distinguishes the two: [A]n aesthetic judgment is that whose determining ground lies in a sensation (einer Empfindung) that is immediately connected with the feeling of pleasure and displeasure (mit dem Gefühle der Lust und Unlust unmittelbar verbunden ist) : : : [I]n the aesthetic judgment of reflection [i.e. judgements of beauty] : : : it is that sensation which the harmonious play of the two faculties of cognition in the power of judgment, imagination and understanding, produces in the subject : : : which relation in such a case produces through this mere form a sensation that is the determining ground of a judgment which for that reason is called aesthetic and : : : is combined with the feeling of pleasure (mit dem Gefühle der Lust verbunden ist). (20: 224) Here, Kant makes clear that the determining ground of judgements of beauty is the sensation of the harmonious play of the faculties, and twice repeats that the feeling of pleasure is not identical to this sensation but connected or combined with it. And the same distinction is repeated elsewhere. 17 On the basis of this textual evidence, as well as the above considerations concerning the role played by common sense, I argue that judgements of beauty have the following structure on Kant's account. Their ground is a sui generis sensation or feeling 18 of the relation between the cognitive faculties as they represent the beautiful object. Through this sensation, the subject becomes aware of the relation of her faculties and judges that they are in harmony, as is required for cognition in general. The result is a feeling of pleasure whichsince it is based in a state that is a universally valid condition on cognitionis itself universally valid. 19 This structure is clearly reflected, for example, in the following passage: The subjective universal communicability of the kind of representation in a judgment of taste : : : can be nothing other than the state of mind in the free play of the imagination and the understanding : : : for we are conscious that this subjective relation suited to cognition in general must be valid for everyone. : : : Now, this merely subjective (aesthetic) judging of the object : : : precedes the pleasure in it, and is the ground of this pleasure : : : but on that universality of the subjective conditions of the judging of objects alone is this universal subjective validity of satisfaction : : : grounded. (5: 217-18, my emphases) I have offered a full defence of this interpretation elsewhere. 20 Here, I want to argue that these passages support a reading of common sense that illuminates the role it plays not only in judgements of beauty but in cognition in general. So far, I have denied that what is sensed through common sense is the feeling of pleasure. I am now in a position to offer a positive characterization: what is given through common sense, I argue, is the sui generis sensation of the relationship between one's cognitive faculties, and in particular, in the case of beauty, the sensation of the harmony of one's faculties which is the ground of judgements of beauty.
We can now explain why Kant says in § §20-23 that anyone who makes a judgement of beauty presupposes the existence of a common sense. Recall that on Kant's view, a subject who judges an object to be beautiful claims that everyone ought to feel the pleasure in it that she does. As we have seen, her pleasure is based in the feeling of the harmony of her faculties. Now, in order to claim that everyone ought to share her pleasure, she must assume that the feeling of harmony in which she takes pleasure is not merely private but is available to all. She must assume, that is, that anyone can sense the harmony of their faculties in the same circumstances. As an analogy: if I claim you ought to share the pleasure I feel in a particular wine, I have to assume you can taste the wine and it tastes the same to you as it does to me. 21 Similarly, if I claim you ought to share the pleasure I feel in the state of mind a beautiful object puts me into, I have to assume you can sense your state of mind and it feels the same to you as it does to me. In short, as Kant puts it, I have to assume we share a common sense that allows us to sense the harmony of our cognitive faculties.
Moreover, we can see why, when Kant posits a common sense, he treats it as a substantive presupposition triggered in particular by judgements of beauty. 22 After all, if common sense were just the capacity to feel pleasure, the assumption that others share this capacity would arise not only for judgements of beauty, but also at least for judgements of the good, which Kant also believes are accompanied by a pleasure claimed to be universally valid (5: 213). On my account, in contrast, Kant's goal is to introduce a novel sense, namely, a capacity to sense the relationship of our cognitive faculties as they engage in representation. Kant repeatedly emphasizes the surprising claim that this relationship can be sensed: [I]n the power of judgment understanding and imagination are considered in relation to each other, and this can, to be sure, first be considered objectively, as belonging to cognition : : : but one can also consider this relation of two faculties of cognition merely subjectively, insofar as one helps or hinders the other in the very same representation and thereby affects the state of mind, and [is] therefore a relation which is sensitive (empfindbar) : : : (20: 223, my emphases) 23 Let me consider an objection here. When Kant first introduces common sense as a condition of judgements of beauty in §20, he describes it as 'the effect of the free play of our cognitive powers' (5: 238, my emphasis). This appears to conflict with my claim that it is specifically the harmony of the faculties that is sensed through common sense. Moreover, since the cognitive faculties are not free in the case of cognition, it also makes it unclear how common sense so described could play any cognitive role. 24 In response, I would note that Kant often uses 'free play' as shorthand for referring to the state the cognitive faculties are in in the case of beautywhich, as we know, is more fully described as a state of harmonious free play. A particularly relevant example of this is at 5: 218, where Kant refers to 'the state of mind in the free play of the imagination and the understanding' as a 'subjective relation suited to cognition in general', clarifying however that this is 'so far as [these faculties] agree with each other as is requisite for cognition in general' (my emphases). On my reading, it is this agreement or harmony of the facultieswhich happens to obtain in free play in the case of beautythat is suited to cognition in general. 25 And this is correspondingly the strategy I suggest we adopt in reading §20 as well. Doing so is supported by the fact that Kant's other discussions of common senseboth in the case of beauty and cognitionconnect it explicitly to the feeling of an optimal relation between the faculties, rather than to their being free or in play.

3
Let us now return to §21 and the Optimal Proportion Argument, where, as we have seen, Kant appeals to the role common sense plays in cognition to argue that particular subjects have reason to presuppose a common sense. My goal in this section is to identify this role, and in so doing, defend my claim that common sense provides the essential link between beauty and cognition for Kant.
I have argued that common sense enables us to sense the harmony of the faculties, and that this harmony is a necessary condition both of judgements of beauty and cognitive judgements. The latter point suffices to explain Kant's initial discussion of common sense in §21. He begins by claiming that cognitions must 'be able to be universally communicated (allgemein mittheilen lassen)', and continues: [I]f cognitions are to be able to be communicated, then the mental state, i.e., the disposition (Stimmung) of the cognitive powers for a cognition in general, and indeed that proportion which is suitable for making cognition out of a representation : : : must also be capable of being universally communicated; for without this, as the subjective condition of cognizing, the cognition, as an effect, could not arise. (5: 238) Cognition, on Kant's view, is essentially universally valid and universally communicable. When a subject makes a cognitive judgement about an object, he argues, she in effect claims that her judgement is universally valid: that is, that it does not express a merely private mental state that she happens to be in, but rather that any subject in her cognitive circumstances ought to make the same judgement. 26 In doing so, she also in effect claims that her judgement is universally communicable: that is, that the necessary conditions for making the judgement are possible for any subject. 27 Now, as we have seen, it is Kant's view that the harmony of the faculties is a necessary condition on cognition in general. It follows that any cognitive subject must be able to have the harmony of the faculties. In Kant's words, the 'proportion' of the faculties that is 'suitable for making cognition out of a representation' must itself be universally communicable.
But Kant does not stop with the claim that all cognitive subjects must be able to have the harmony of the faculties. Rather, as we saw in the Optimal Proportion Argument, he adds that every subject must be able to feel this harmony, and thus, that there must be a common sense: [T]here must be one [disposition of the cognitive faculties] in which this inner relationship is optimal for the animation of both powers of the mind : : : with respect to cognition : : : in general; and this disposition cannot be determined except through the feeling (not by concepts). Now since this disposition itself must be capable of being universally communicated, hence also the feeling of it : : : but since the universal communicability of a feeling presupposes a common sense, the latter must be able to be assumed with good reason, and indeed without appeal to psychological observations, but rather as the necessary condition of the universal communicability of our cognition : : : (5: 238-9, my emphases) The question, of course, is why exactly Kant claims all cognitive subjects must be able to feel the harmony of their faculties. One explanation, offered by Guyer, is that he is relying on a 'same cause, same effect' principle: if cognition requires that every subject have the harmony of the faculties, and this harmony causes a feeling, then we can predict that every cognitive subject feels the harmony. 28 However, this explanation seems to rely on the very kind of 'psychological observation' Kant rules out as a basis for assuming a common sense, since it appeals to the fact that the harmony of the faculties happens to cause a feeling in subjects like us. In contrast, Kant argues in the passage above that we can know a priori that cognitive subjects must be able to feel the harmony of their faculties, since cognition depends on subjects being able to determine through feeling when their faculties are in the 'optimal : : : disposition' or harmony. It is for this reason that he concludes that common sense is a condition on the universal communicability of cognition itself.
But why should we think cognition depends on a feeling and so, requires a common sense? Here, some interpreters appeal to Kant's claim in the Critique of Pure Reason that judging whether a particular falls under a concept requires a 'special talent' (A133/B172). According to Allison, for example, this suggests that judgers must have a capacity for 'immediately seeing (without appeal to rules and therefore "feeling") whether : : : a given intuited manifold accords with a particular concept'. 29 On this view, it is this capacity to 'feel' whether a concept one possesses applies to a particular that calls for a common sense. However, it should be noted, first, that Kant never associates the talent in question with a feeling; and second, that he makes clear it is by no means common. In fact, he expressly allows that even a learned physician, judge or statesman might 'stumble in their application' of theoretical rules because they lack the talent for judgement (A134/B173). As I see it, this makes it a poor candidate for a shared common sense that is meant to be a condition on cognition as such. 30 In order to understand the role the feeling of harmony plays in cognition, I argue we should look not to the Critique of Pure Reason, but to the Critique of Judgement itself, and in particular, to what could be considered its unifying topicthe type of judgement that Kant calls 'reflecting (reflectirende)' judgement. When a subject subsumes a given particular under a concept she already possesses, she makes a 'determining (bestimmende)' judgement; in contrast, when she subsumes a given particular under a new concept that she thereby acquires, she does so through a reflecting judgement (5: 179). As Kant notes in the introductions to the Critique of Judgement, although he did not discuss reflecting judgements in the first Critique, they are nevertheless essential for empirical cognition, since it is through them that we come to have the empirical concepts necessary for experience (20: 212-13; 5: 180-2). I will now argue that common sense plays a necessary role in these reflecting judgements through which we acquire empirical concepts, and it is this role Kant has in mind when he claims common sense is necessary for the universal communicability of cognition. 31 Reflecting judgements through which we acquire empirical concepts share an important feature with judgements of beauty (indeed, both are species of reflecting judgement). In both cases, the subject is presented with a particular for which she does not have an adequate concept and, as a result, the imagination synthesizes the manifold in a manner that is not determined to accord with the understanding. After this, the two cases diverge in important ways. In the case of beauty, the freely synthesized form of the object is itself felt to give rise to the harmony of the faculties, and the result is the distinctive pleasure in the beautiful. In the other case, in contrast, the subject compares the given representation with others she judges to fall into the same empirical class as it, a class for which she thereby acquires a concept. A wellknown example of the latter is found in the Jäsche Logic, where Kant describes the process of acquiring the concept 'tree' by comparing representations of particular trees: I see, e.g., a spruce, a willow, and a linden. By first comparing these objects with one another I note that they are different from one another in regard to the trunk, the branches, the leaves, etc.; but next I reflect on that which they have in common among themselves, trunk, branches, and leaves themselves, and I abstract from the quantity, the figure, etc., of these; thus I acquire a concept of a tree. (9: 94-5) An obvious question about Kant's example is what makes it the case that I select the spruce, willow and linden for comparison in the first place. The answer cannot be that I recognize they are all trees, or even that I recognize they are similar, because that recognition is only supposed to arise as a result of my comparison and, as such, cannot antecedently inform what I choose to compare. 32 To respond to this question, I will draw in part on Ginsborg's reading of Kant's account of empirical concept acquisition, which appeals to the role played by the reproductive imagination (Ginsborg 2015: 159-62). For Kant, the reproductive imagination is the capacity to recall to mind representations previously had by a subject. It is governed by laws of association, which make it the case that representations that have often accompanied each other in the past come to mind together (A100). In his lectures on anthropology, Kant specifies the bases of association, namely 'accompaniment, contiguity, and : : : relation : : : [of] similarity and derivation'. Successive or simultaneous events bear the relation of accompaniment, leading to the type of association in Kant's example: 'if we see smoke, then the representation of fire immediately appears'. Things at or near the same location bear the relation of contiguity: 'if one thinks about the place where one enjoyed oneself, the people who were present there come to mind'. Things which 'come from one basis' bear the relation of derivation: 'if it rains and the sun shines, one immediately looks around [to see] if there is not a rainbow'. Finally, things that belong 'to certain classes' bear the relation of similarity, such that 'if we think of one thing, the other comes to mind' (Lectures on Anthropology, 25: 512-13). Now, one or another of these relations could very plausibly explain why a subject finds herself thinking together of a spruce, willow and linden. Perhaps these representations are associated because she encountered these trees in the same forest. Or perhaps their very similarity makes it the case that that seeing one automatically calls the others to mind. Either way, we can explain why these trees come to a subject's mind by appealing to associations due to her reproductive imagination, without needing to attribute to her the antecedent capacity to recognize that they are all trees or even that they are similar in specific respects.
Indeed, there is evidence that Kant takes the imagination and its associations to play an important role in the acquisition of concepts. In his notes on logic, amidst an extended discussion of empirical concept formation, he says it is through the comparison of imaginatively combined representations that we clarify the concept that applies to them (Refl. 2878, 16: 556-7). In another note, he adds that we do so by 'compar [ing] only what is common in the rule of our apprehension (Regel unserer Auffasung)', and offers the following examples of the kind of rule he has in mind: 'one sees a bush, so one can represent a tree; an elongated rectangle gives rise to a square' (Refl. 2880, 16: 557). The similarity between these examples and the ones from Kant's lectures above strongly suggests that what he means by a 'rule of : : : apprehension' here is a particular associative habit that exists for a subject. This would be in line with Kant's claim in the first Critique that apprehension essentially involves reproduction (A102), as well as his description of associations as subjective rules (A100, A121). 33 Thus, these passages support the claim that, on Kant's view, we acquire empirical concepts through a comparison of what is common among representations that have been reproduced according to an associative rule.
In order to count as acquiring a new concept, however, it is not enough to have a set of representations come to mind and recognize similarities in their content. Rather, the similarities must be taken as a basis for judging that the represented objects belong to the same empirical class. For Ginsborg, this further step consists in the awareness that a particular association is 'appropriate' (Ginsborg 2015: 160), that is, that it expresses 'how I (and everyone else) ought to associate representations' (p. 163). She claims this awareness is 'primitive': there are no antecedent criteria associations are judged to satisfy in virtue of which they are appropriate (pp. 163-8).
One problem, however, is that most associations will not be regarded as appropriate, at least not in the relevant sense of presenting objects that belong to the same empirical class. We only need to look back at Kant's own examples of associations to see this: fire with smoke, a sunny rainfall with a rainbow, etc. Thus, an important question that Ginsborg does not answer is why we only regard some associated representationssuch as those of the spruce, willow and lindenas belonging to one empirical class and, in so doing, acquire a new concept. 34 What distinguishes these representations from those we do not classify together?
The answer may at first seem obvious: unlike the other cases, comparing representations of the spruce, willow and linden reveals that they have a number of features in common. In general, we might respond, it is only if compared representations are sufficiently similar that we classify them together. However, it is hard to more precisely specify what it means for representations to be sufficiently similar in the relevant respect. For one, almost all objects will appear similar in some respects. Moreover, objects may appear similar in many respects without being judged to fall into the same classconsider, for example, a leafless tree and a wooden pole. Conversely, objects may be judged to fall into the same class despite having few apparent similaritiestwo chairs with very different forms and materials, say, or a leafless tree and a flowering bush. In short, it is not clear that there are determinate criteria that can be said to systematically guide the comparison of associated representations and distinguish those judged to form the basis of an empirical concept from those that are not.
Indeed, Kant argues that reflecting judgements by which we acquire empirical concepts 'requir[e] a principle' that governs them just as determining judgements do (20: 211). Whereas the latter are governed by the particular concept(s) one applies in them, Kant points out that, in the case of a reflecting judgement, 'one cannot refer it : : : to already known empirical concepts' (20: 213). Instead, he argues, reflecting judgements must be governed by an a priori principle, which he states as the principle that 'for all things in nature empirically determinate concepts can be found' (20: 211). 35 As Kant understands it, this amounts to the assumption that, despite the 'immeasurable multiplicity of things' in nature, we can find 'sufficient kinship among them' to subsume them under a system of empirical concepts governed by empirical laws (20: 215). In short, Kant argues that reflecting judgements must assume that nature is 'purposive' for our faculty of judgement (20: 202-4; 5: 184), that is, that it is suitable to be judged by us. Now, by itself, Kant's principle of reflecting judgement appears much too general to offer a response to our question about why some associated representations form the basis of empirical concepts whereas others do not. Indeed, he describes it as an 'indeterminate (unbestimmtes) principle', one that cannot be 'explain[ed]' or 'determine[d] : : : more precisely' (20: 214). How, then, is this principle applied in particular reflecting judgements? Here is what Kant says: The reflecting power of judgment : : : proceeds with given appearances, in order to bring them under empirical concepts : : : not schematically, but technically (technisch), not as it were merely mechanically, like an instrument, but artistically (künstlich), in accordance with the general but at the same time indeterminate principle of a purposive arrangement of nature in a system, as it were for the benefit of our power of judgment. [W]e shall : : : use the expression 'technique' where objects of nature are sometimes merely judged as if their possibility were grounded in art, in which cases : : : nature itself is judged : : : in subjective relation to our cognitive faculty, not in objective relation to the objects. (20: 201) Kant's point in these passages is that reflecting judgements by which we acquire empirical concepts are not determined by objective criteria that we antecedently possess. Instead, they treat nature as if it were 'technical' or 'artistic' in relation to our faculty of judgementthat is, as if nature had designed its products and their relation to each other with a purpose in mind, namely, the purpose of being judgeable by beings with cognitive faculties like ours. Kant describes this as a subjectively necessary principle (20: 209): although we do not have objective grounds for its truth, it is necessarily presupposed by our empirical investigation into nature (5: 184). 36 In our search for empirical concepts, then, we assume that nature is purposively designed in relation to us and set out, so to speak, to uncover that design. For Kant, as we have seen, acquiring new empirical concepts requires comparing representations that have been called to mind by the reproductive imagination. I have argued that a question we face in such cases is whether particular reproduced representations are sufficiently similar to count as presenting objects that belong to the same empirical class. Kant's claim that the procedure we follow in acquiring empirical concepts is 'technical' suggests that this question should be understood in relation to our cognitive faculties as follows: does it seem as if these objects were designed in order to be graspable by us under one empirical concept? That is, does this set of representations that has been reproduced by the imagination conform with the general requirements of the understanding as a faculty for concepts? 37 How can the question so restated be answered? Since the purpose of the assumed design is for objects in nature to be suitable to our cognitive faculties, I suggest we can discover instances of it by attending to the effect particular representations have on our cognitive faculties. Indeed, Kant makes clear that when we reflect for the sake of an empirical concept, we can have what he calls an 'inner perception of a purposiveness of representations' (20: 220, my emphasis): How can the technique of nature : : : be perceived? : : : In our power of judgment we perceive purposiveness insofar as it merely reflects upon a given object : : : in order to bring the empirical intuition of that object under some concept : : : Thus the power of judgment is properly technical; nature is represented technically only insofar as it conforms to that procedure of the power of judgment : : : (20: 219-20, my emphasis) What this 'inner perception' enables awareness of, he adds, is whether or not the imagination and the understanding are in harmony, as required for judgement in general (20: 220). If having a particular set of representations in mind puts our faculties into harmony, then since the harmony of faculties is a necessary condition for cognition, the represented objects will appear as if designed to be grasped together in one cognition by us. 38 It is in such cases, I suggest, that we classify the objects together and acquire a new concept for them. On my view, this is why Kant claims that the same 'procedure of the power of judgment' is employed both 'for the sake of an empirical objective concept' and in 'aesthetic judging', namely a procedure that relies on sensing the harmony of the faculties (5: 292).
If this is right, we are finally in a position to understand Kant's claim that common sensewhich I have argued is the capacity to sense the relation of our facultiesis a necessary condition on cognition. Cognition requires the application of empirical concepts, and as we have just seen, the reflecting judgements by which we acquire empirical concepts depend on our ability to sense when our faculties are in harmony. This is the role for common sense I believe Kant has in mind when he claims in the Optimal Proportion Argument that cognition depends on our ability to determine the optimal relationship between our faculties 'through the feeling (not by concepts) ' (5: 239). 39 Additionally, it should be noted that a judgement that classifies a set of objects under an empirical concept essentially claims universal validity. When a subject classifies the spruce, linden and willow as trees, she does not claim that this is merely how she privately sorts them; rather, she effectively claims that it is correct to, and so that everyone ought to so classify them. But since, as we have seen, making this judgement partly depends on feeling the harmony of the faculties, this claim presupposes that everyone can feel the harmony in the same circumstances. This explains why Kant says specifically in the Optimal Proportion Argument that the universal communicability of cognition presupposes the existence of a common sense. If the capacity to feel the harmony of the faculties were not shared, judgements involving empirical concepts would not be universally communicable and could only claim private validity. Since cognition is essentially universally communicable, Kant concludes, there must be a common sense.
One last point is in order here. In section 1, I suggested that the question Kant is asking in §21 is whether particular subjects who make judgements of beauty have any reasons available to them for presupposing a common sense. As we have seen, he answers affirmatively, on the ground that common sense plays a necessary role in cognition. In this section, I have sketched this role by arguing that common sense is necessary for the reflecting judgements through which we classify objects. But does Kant really think this role is accessible to particular subjects? One indication he does is his claim that, although we may not ordinarily notice the purposiveness displayed by nature in 'the unity of its division into genera and species, by means of which alone empirical concepts are possible', this is only because 'the most common experience would not be possible without it ' (5: 187). He adds that we can indeed become 'attentive to' this purposiveness and that doing so even gives rise to a feeling of pleasure in us (5: 188). 40 What we can attend to, on my view, is the feeling of harmony that underlies judgements by which disparate objects are unified under a system of concepts. Since these judgements are essential for cognition, claim universal communicability and, as I have argued, are partly determined by the feeling of harmony, particular subjects indeed have good reason available to them to regard the capacity to feel the harmony of the faculties as a shared capacity. In short, they have good reason to presuppose a common sense.
Is the claim that a feeling of harmony plays a necessary role in the acquisition of empirical concepts plausible? I cannot offer a full interpretation and defence of Kant's account of empirical concepts here, but let me offset a possible worry. In assigning this role to common sense, I do not intend to deny that other factorssuch as the relevant scientific knowledge or our practical interestsplay a crucial role in determining how we classify objects. My suggestion is that in most cases these factors will leave open several possibilities and, as such, will underdetermine the correct way to proceed. Kant's claim, as I understand it, is that in such cases we must proceed in a way that feels cognitively right to us, even if we later revise this based on further information. As he puts it, the fact that 'there is such a manifold of forms in nature' means that in 'ascending from the particular : : : to the universal', the reflecting power of judgement must 'regar [d] as necessary' what 'may seem to be contingent' to our understanding (5: 179-80). The choice of one classificatory scheme over another can seem contingent because it is underdetermined by the available objective criteria. In the face of this indeterminacy, Kant argues, we determine how to proceed partly on the basis of subjective criteria having to do with our own cognitive faculties. We proceed, that is, on the assumption that nature has placed its joints where they are most comprehensible to us and best accord with our cognitive needsincluding, importantly, the need that our faculties be in harmony. 41

Conclusion
The identification of common sense with pleasure has led many interpreters to minimize the central role it plays not only in grounding judgements of beauty but in cognition itself. In this article, I have argued that common sense is the capacity to sense the relation between our cognitive faculties through a sui generis feeling distinct from pleasure. On my view, this feeling of harmony serves as a ground for and unifies two disparate species of reflecting judgement Kant focuses on in the Critique of Judgement: judgements of beauty and judgements by which we acquire empirical concepts. In sum, common sense enables us to become aware of occasions on which the products of nature appear as if designed to bring our cognitive faculties into harmonyan awareness which, on Kant's view, underlies and links our experiences of beauty and our ability to empirically cognize nature. 42 Notes 1 Translations of Kant's text are from the Cambridge Edition of the Works of Immanuel Kant, ed. Paul Guyer and Allen W. Wood (Cambridge: Cambridge University Press, 1992-) and are given using the volume and page number in the Akademie edition. References to The Critique of Judgement and the First Introduction only mention volume (5 and 20, respectively) and page numbers. References to the Critique of Pure Reason use the standard A/B pagination. 2 Guyer, for example, claims that common sense is a 'needless complexity' on which 'nothing crucial : : : depends' (1997: 274). 3 'There must be one [proportion] in which [the] inner relationship is optimal for the animation of both powers of the mind : : : with respect to cognition (of given objects) in general' (5: 238-9). 4 Guyer 1997: 248-73;Ameriks 2003: 286;Kalar 2006: 133;Rind 2002: 2, 17;Stoner 2019: 93. 5 Matherne explicitly distinguishes between 'an aesthetic and a cognitive species of common sense' (2019: 3; see also 4-8). Allison likewise claims that §21 argues for 'an explicitly cognitive version of common sense' only in order to 'give some support to the idea that a pure judgment of taste may likewise be thought to be grounded in a common sense, albeit not a cognitive one ' (2001: 153, my emphasis). He also argues that reading Kant as claiming that 'the aesthetic common sense or taste must itself be presupposed as a condition of cognition' is not only 'implausible' but 'completely impossible' (ibid.). Longuenesse also considers the possibility that there may be 'merely a kinship, not a generic identity, between the sensus communis that grounds judgments of taste : : : and the sensus communis that grounds judgements in empirical cognition ' (2006: 216). 6 For further discussion of my reading of Kant's argument in the Deduction, see Sethi 2019: §3. 7 See 5: 219, 238-9, 292-3. 8 Here, it should be noted that Kant also claims in §21 that the 'disposition of the cognitive powers has a different proportion depending on the difference of the objects that are given ' (5: 238). This seems to suggest that the required proportion of the cognitive faculties varies for different objects. As I discuss in notes 14-15, some interpreters opt for this reading, in part to block the implausible implication that all cognition is pleasurable. However, Kant immediately continues as follows: 'Nevertheless, there must be one in which this inner relationship is optimal for the animation of both powers of the mind : : : with respect to cognition (of given objects) in general' (5: 238-9, my emphases). Of course, this sentence is ambiguous. It could be read as claiming that each (type of) object has a particular optimal proportion that is required to cognize it, or that there is one optimal proportion that is required to cognize any object. The latter does fit better with the conclusion of the sentence, since Kant says that the proportion is optimal for 'cognition : : : in general'. It is also preferred, I argue, when read alongside the passage at 5: 292 cited above, which indicates that the same proportion of the faculties is required both for taste and for cognition in general. Thus, I suggest that Kant's claim concerning different proportions be read as follows: different objects trigger different responses in the imagination and the understanding, resulting in them having different 'proportions' to each other. But in order for cognition to result, they must be 'harmonized' or brought into the optimal proportion that is required for cognition in general. This is admittedly somewhat metaphorical, but that is the case for Kant's talk of 'proportion' and 'harmony' between the faculties in general. I am grateful to an anonymous referee for urging me to consider this worry. 9 Here, it might be objected that Kant is not questioning the existence of the so-called epistemic common sense, but only whether it plays a constitutive or merely regulative role in experience. Even if this is granted, however, it still seems to pose a problem for the type of interpretation I am rejecting, since the latter does not read Kant as raising any questions about the epistemic common sense in §21-2, but rather, as relying on the necessary role it plays in experience to argue that there may also be an aesthetic common sense. Moreover, after asking whether there is a common sense that is 'constitutive : : : of experience', or if it is a regulative principle that we 'produce a common sense in ourselves', Kant immediately restates this as identical to the question 'whether taste is an original and natural faculty : : : or : : : acquired and artificial' (5: 240). Here again, then, he takes a question about the role common sense plays in experience to be at the same time a question about taste, indicating that he does not draw a distinction between an epistemic and an aesthetic common sense. I am grateful to an anonymous referee for raising this worry. 10 Dobe (2010: 47) makes the similar claim that Kant's presentation consists in a 'process of discovery' culminating in the Deduction. Rather than taking him to initially present incomplete and potentially misleading information, however, my reading appeals to the different goals of the Analytic and the Deduction. 11 Allison 2001: 148;Kalar 2006: 138;Dobe 2010: 48-50;Hughes 2007: 178;Rind 2002: 18;Stoner 2019: 88;Matherne 2019: 9-10;Wang 2018: 261-5;Zinkin 2006: 157;Zuckert 2007: 336, n. 26. 12 (Princip : : : welches nur durch Gefühl und nicht durch Begriffe, doch aber allgemeingültig bestimme, was gefalle oder mißfalle.) 13 The sentence could be read as claiming that feeling pleasure allows us to determine which objects are pleasing. But this still seems a fairly trivial claim: it is not clear why one would need a principle in order to judge that an object that causes pleasure is pleasing. The other passages I go on to discuss also tell against this more trivial reading.
14 As I have discussed, Kant asserts that the same proportion of the faculties is required for beauty and cognition in general. Thus, I disagree with interpreters who argue there is a different proportion unique to beauty. See Ameriks 2003: 290, 336;Kalar 2006: 139-41;Fricke 1990: 166;Hughes 2007: 175, 181;Rind 2002: 19;Makkreel 1990: 62. A primary reason for holding this view is to block the implication that all cognition is pleasurable, which would follow if the proportion of the faculties that is the ground of judgements of beauty is sensed through pleasure and this proportion also obtains in cognition in general. As we will see below, this problem does not arise for my interpretation.
15 Indeed, this claim has been thought to give rise to a well-known dilemma for Kant's deduction of judgements of beauty. See Guyer 1997: 263;Ginsborg 2015: 54;Allison 2001: 184;Longuenesse 2006: 216;Fricke 1990: 167;Ameriks 2003: 285, Rind 2002. I discuss my response to this dilemma in Sethi 2019: §3. Cohen (2020 has recently argued that the harmony of the faculties in cognition does give rise to a pleasure, albeit one that is not normally noticed. As we will see in section 3, I agree with Cohen that feeling plays an essential role in cognition, but I disagree that this feeling is pleasure. Now, Kant does claim that we can take pleasure in finding and systematizing empirical concepts and laws (5: 187-8, 242). On my view, this pleasure does not accompany cognition in general, but only the particular reflecting judgements by which we acquire the system of concepts necessary for cognition. Moreover, I do not believe this pleasure is due to the harmony of the faculties: in both passages, Kant describes it as a satisfaction we feel as a result of achieving our 'aim' of systematizing nature. See n. 40 for further discussion. 16 Kant's claims in §9 may seem to be incompatible with his claim in §12 that 'consciousness of the merely formal purposiveness in the play of the cognitive powers' just 'is the pleasure itself' (5: 222). He also makes the similar claim elsewhere that the 'representation ' (20: 228, 249) or 'concept' (20: 230) of the 'formal : : : purposiveness of the object' (20: 230) is identical to pleasure in the beautiful. As I understand it, however, being conscious of formal purposiveness is not simply feeling the harmony of the faculties, but rather representing their relationand, correspondingly, the object that causes itas purposive. On my view, this is why Kant lists formal purposiveness as one of the moments of a judgement of beauty: it is part of what is claimed by the judgement itself, rather than its ground. Thus, when he says that the representation of formal purposiveness is identical to pleasure, I take him to mean that the subject's state of mind (and the beautiful object) is represented as purposive through the feeling of pleasure. This is in line with his claim that pleasure serves as a predicate in judgements of beauty (5: 191, cf. 288-9)that is, as I understand it, that it is part of the content of the judgement itself. This is compatible with the ground of judgements of beauty being an independent feeling of the relation between the faculties, which is the reading I will go on to defend. See Sethi 2019: §2.3 for a more extended discussion of this worry. 17 For example, 20: 226. 18 Kant defines a feeling as a type of sensation through which a subject is aware of a merely subjective state of herself rather than of a property that can be ascribed to an object (5: 206). He uses both 'sensation' and 'feeling' to describe our awareness of the harmony of the faculties, as well as 'pleasure'. Here, it might be objected that Kant does not allow for any feelings other than pleasure and displeasure. One piece of evidence against this, however, is that he does refer to other subjective sensations that are not instances of pleasure or displeasure as feelings: colour, warmth and sweetness, for example (A28/B44; Prolegomena, 4: 299n.), and even a subject's feeling of her own existence (4: 334n.). A second consideration is that Kant's definitions of pleasure and displeasure are narrower than his definition of feeling as a merely subjective sensation. He defines pleasure as 'the consciousness of the causality of a representation with respect to the state of the subject, for maintaining it in that state' and displeasure as a 'representation that contains the ground for determining the state of the representations to their own opposite' (5: 220). See Sethi 2019: §2.4 for further discussion. 19 Thus, on my reading, there are two key senses in which a judgement of beauty is aesthetic. First, in Kant's words, 'its determining ground is not a concept but [a] feeling' (5: 228)namely, the feeling of harmony between the faculties. Second, the judgement is itself made through a feeling, namely the feeling of pleasure. As Kant says, in a judgement of beauty, the feeling of pleasure is 'connected with' the representation of the beautiful object 'just as if it were a predicate ' (5: 191). 20 See Sethi 2019. 21 Since there is no a priori guarantee of this for sensory taste, Kant denies such judgements can claim universal validity (5: 224). 22 Here, I agree with Stoner 2019: 91. 23 See also 20: 223-4; 5: 219, 287, 291. I take these passages to bear on another question that has been discussed in the secondary literature: whether common sense is a natural or cultivable faculty (Guyer 1997: 270-2;Allison 2001: 219-20;Dobe 2010: 57-8). As we have seen, Kant raises these two possibilities at the end of §22, but puts off settling the question there. Although he does not return to it explicitly, I take his remarks such as the above that the relation between the cognitive faculties is a 'sensitive (empfindbar)' one (20: 233) to indicate that he takes us to have a natural capacity to sense the harmony of the faculties. This also fits with his claim that subjects have good reason to presuppose a common sense, since it is 'a necessary condition of the universal communicability of our cognition' (5: 239). It might be suggested that Kant's later remarks that there is a moral interest in judging with universal communicability (5: 296, 353) imply instead that we have a duty to acquire a common sense. This is not how I read those remarks. As I see it, we could not be entitled to so much as claim universal communicability for our reflecting judgements if we did not already have the faculty of common sense; what we can cultivate is our skill at grounding such judgements in this common sense rather than on merely private grounds such as 'charm and emotion ' (5: 294). 24 In addition, it might be objected that Kant's description of common sense as an 'effect' conflicts with my description of it as a capacity. But as several commentators have noted, Kant variously describes common sense as a principle, a faculty and a feeling. On my view, its role as a faculty or capacity is primary; it is treated as a principle when a judgement assumes the existence of this faculty, and since it is a common faculty for a particular sensation or feeling, the sensation itself can also be described as common. See Guyer 1997: 249-50;Allison 2001: 156-7, Dobe 2010 An alternative reading, defended, for example, in Wang 2018: 260, is to take Kant's mentions of 'cognition in general' to refer to a 'peculiar' technical notion that is 'logically distinct' from any conceptually determinate cognition, such that the conditions of the former are satisfied by a free rather than conceptual harmony. This has theto my mind, undesirableimplication that determinate cognitions do not count as instances of cognition in general. Moreover, the passage above tells against this reading, since after referring to the free play as 'suited to cognition in general', Kant goes on to say that 'any determinate cognition : : : still always rests on that relation as its subjective condition' (5: 238, my emphasis). 26 See Prolegomena, 4: 298. 27 Interpreters typically treat universal validity and universal communicability as equivalent. An important difference, I think, is that the universal communicability of a state refers to the necessary possibility that other subjects come to be in that state, whereas universal validity does not (although it entails it, 5: 218). 28 Guyer 1997: 255. 29 Allison 2001see also Fricke 1990: 170-1. 30 One could object that, although the talent may come in degrees, cognition is not possible if it is entirely lacking. But Kant's discussion tells against this: he says that 'he who lacks the natural talent for judgement' is wholly dependent on practising with concrete examples to learn the application of abstract rules (A134/B174-5). 31 Kant notes in the First Introduction that, although the necessary conditions on cognition identified in the Critique of Pure Reason specify the transcendental laws that constrain experience in general, they do not suffice to explain how experience of nature is possible, since the latter also requires bringing particular appearances under a system of empirical concepts and laws (20: 208-9). This is in line with his claim in the Transcendental Deduction that the particular laws required to empirically determine appearances 'cannot be completely derived from the categories, although they all stand under them' (B165). As I understand it, then, the type of reflecting judgement Kant discusses in the third Critique is necessary for a complete account of how empirical cognition of nature is possible. I am grateful to an anonymous referee for asking me to consider the relation between my discussion here and the Transcendental Deduction in the first Critique. 32 See Ginsborg 2015: 70;Zuckert 2007: 54-5. 33 Here, I differ from Longuenesse, who takes the rules of apprehension here to be schemata (1998: 116-18). Kant's examples seem to clearly suggest a natural movement of the mind between similar representations, rather than a general procedure for providing a concept with an imagewhich is how Kant describes schemata. See also Schlösser 2013: 191, n. 26. 34 Ginsborg recognizes such examples (2015; 163) but it is unclear how her view addresses them. She takes the limits on the empirical concepts we can acquire to be set by the limits on our associative dispositions (p. 164), but we clearly do not take many of our associative dispositions to relate objects that fall into one empirical class. 35 For other statements of the principle, see 20: 214-16. 36 I do not have space to fully discuss this claim. See Guyer 1997: 136-47;Allison 2001: 35-42;Ginsborg 2017;Zuckert 2007: 23-63. 37 A full explanation and defence of this claim would require a fuller consideration of Kant's account of empirical concepts than I have space for here and would likely take us well beyond the third Critique. One possible explanation might appeal to the role schemata play in mediating concepts and intuitions. Kant defines a schema as the 'representation of a general procedure of the imagination for providing a concept with its image' (A140/B179-80). Schemata are connected to concepts in virtue of their generality and to sensible intuitions in virtue of their role in image-formation (see Matherne 2015: 762-7). We might conjecture that a set of reproduced representations conforms to the requirements of the understanding to the extent that they can be jointly captured by a schema that can be attached to a concept we thereby acquire. Relevant here is Kant's claim that the imagination can reproduce 'an immense number of objects : : : of one and the same kind', and by 'superimpos[ing] one image on another' can 'arrive at a mean that can serve them all as a common measure' (5: 234). See Longuenesse 1998: 116-18. 38 In a newly published paper, Wang similarly suggests that the formation of empirical concepts requires reflecting on the 'lawfulness without law' exhibited in the mutual agreement of the imagination and the understanding (2021: 213-15). There are, however, a number of differences between our proposals. First, Wang does not connect this reflection to common sense and the necessary role Kant claims it plays in cognition. Second, Wang takes such reflecting judgements to proceed on the basis of pleasure. Although he admits that the involvement of pleasure in such judgements 'seem[s] rather counter-intuitive' (p. 214), he points to Kant's claim that the achievement of our cognitive aim in systematizing nature can be the source of a pleasure (5: 187). As I discuss further in n. 40 below, however, Kant claims that this pleasure arises as a consequence of succeeding in our aim of systematizing nature, not as the ground of the judgements through which we do so. Finally, Wang does not understand the role of the imagination in the formation of empirical concepts to be governed by laws of association as I have argued it is. Rather, he suggests that 'the imagination synthesizes according to the understanding's requirement' (p. 208) in apprehending the manifold of intuition even before it can be grasped under a concept. On my view, in contrast, it is because the imagination proceeds according to its own laws that the discovered harmony that enables empirical concept formation counts as free, and thus analogous to the free harmony in beauty. 39 Here, it might be objected that Kant never mentions common sense in the context of empirical concept formation. Nevertheless, as I have discussed, he does make clear in §21 that it is a condition on cognition in general, without referring the reader to any other discussion of why this is so. Thus, interpreters find themselves faced with the task of identifying a cognitive role for common sense to play: as we have seen, Allison finds this role in the 'special talent' for judgement mentioned in the first Critique (A133/B172), even though there is no indication there that that talent is common or involves a sense. In contrast, as we have just seen, Kant does claim explicitly that empirical concept formation involves a perception of purposiveness (20: 220); this not only implies that we have a sense by which we perceive purposiveness, but also that it must be commonly shared if the judgements involving these empirical concepts are to be universally communicable. Moreover, as noted above, Kant says at 5: 292 that the same 'procedure of : : : judgment' is employed both for the sake of objective empirical concepts and in aesthetic judgement, and we know that the procedure in the latter case essentially involves common sense. 40 Kant explains that 'the attainment of every aim is combined with the feeling of pleasure', and describes the relevant aim as that of systematizing nature according to the principle of reflecting judgement (5: 187). This pleasure clearly does not serve as a ground for such judgements; rather, Kant says it arises once we have 'succeed[ed]' (5: 188) in unifying particulars under universals. 41 Although I cannot defend this here, I think there is a connection between this view and contemporary discussions of 'natural' vs. 'unnatural' properties. According to Lewis (1983: 346-7), for example, although any two particulars share infinitely many properties, they resemble each other only to the extent that they share 'natural' properties, for only the latter correspond to universals and ground genuine causal powers. It remains unclear, however, how we are meant to identify so-called natural properties. Kant would argue, I suggest, that the 'naturalness' of a property is partly determined by our own cognitive