3.a. Texts

There are six passages demonstrating Kant’s commitment to animal intuitions. I begin with three attributing intentional states to animals,Footnote ⁴ followed by three indicating that these are intuitions. The first comes from the Jäsche Logic (1800):

In normal usage, “acquaintance [kennen]” picks out an intentional mental state. When one is “acquainted with something,” the “something” is the intentional object of this mental state. Kant uses the term “acquaintance [Kenntnis, kennen]” in this sense in various works (A207/B252; A540/B568; G 4:451; KpV 5:51). Hence, when he asserts that “animals are acquainted with objects,” we have every reason to think he is crediting them with intentional states.

Several further facts confirm this: (a) Kant explicitly characterizes “acquaintance” as a way of “represent[ing]” the thing in question; and (b) the remark occurs in the course of distinguishing levels of the “objective content” of “cognition.” Kant consistently affirms that “all our cognition has […] a relation to the object” (Log 9, 33; cf. 9, 91; A320/B376),Footnote ⁶ never using “cognition” to designate nonintentional states.Footnote ⁷ By introducing “acquaintance” as a level of “cognition” with “objective content,” Kant is classifying it as an intentional representation. (c) Kant also introduces the notion of “acquaintance” in a similar graded list from the pre-“Critical” Blomberg Logic (early 1770s), characterizing it in the same way and explicitly describing it as a way of “cognizing” a thing (24: 135). Together, this makes a very strong case that the “acquaintance” Kant ascribes to animals is a representation with intentionality.Footnote ⁸

The next passage comes from the Wiener Logic (1780–1781). Once again, Kant offers a graded list of types of mental representation:

1. 1. The lowest level is to represent something. When I cognize that which relates to the object, I represent the object.

2. 2. To cognize, percipere, is to represent something in comparison with others and to have insight into its identity or diversity from them. […] For animals also cognize their master, but they are not conscious of this. (24:845–46)Footnote ⁹

What Kant here describes as “cognizing” has the same features as “acquaintance” in the previous list. It is an intentional representation which allows for the identification and discrimination of objects. As argued above, Kant’s choice of the term cognize indicates that this is an intentional state. This is again confirmed by the context: (a) earlier in this transcript, Kant is recorded as affirming that “all our cognitions” exhibit “relation to the object” (in a broad sense that explicitly covers the intentionality of intuitions, 24:805); (b) the “lowest level” on this list already requires “represent[ing] an object,” indicating that these are divisions within the domain of intentional states; (c) Kant explicitly glosses “to cognize” as a way “to represent something.” Therefore, when Kant affirms that “animals also cognize their master,” we have overwhelming reason to think this is a state with intentionality.

Why the discrepancy in terminology between the Jäsche and Wiener passages? The former states that “animals […] do not cognize [objects]” (Log 9:65), the latter that “animals […] cognize their master” (V/Lo-Wiener 24:846). Does this apparent contradiction suggest unreliable transcription? We should not draw this conclusion, because there is strong independent evidence that Kant uses the term cognition in two different senses. In the broad sense, cognition means a representation that represents an object, e.g., a concept, intuition or judgment, while cognition in the narrow sense means an objectively valid synthetic judgment (cf. Grüne Reference Grüne2009, 29; Watkins and Willaschek Reference Watkins and Willaschek2017, 84–7). Even within the Jäsche passage, Kant uses the term in these two senses: the list as a whole deals with different levels of cognition in the broad sense, while the “fourth level” isolates cognition in the narrow sense. Consequently, we can easily reconcile the two passages: in the Jäsche passage, Kant denies that animals have cognition in the narrow sense, i.e., objectively valid synthetic judgments; in the Wiener passage, he affirms that animals have cognition in the broad sense, i.e., intentional representations.Footnote ¹⁰

Is it significant for our purposes that the Wiener passage denies that animals are “conscious of” “cognizing their master”? In KrV’s “Stufenleiter,” Kant appears to categorize cognition and intuition as forms of “perception [Perzeption],” i.e., “representation with consciousness” (A320/B376–77; cf. Log 9:91). Therefore, there is a prima facie tension between Kant’s denial that animals’ cognitions are conscious and his affirmation that they are cognitions. However, there are various other passages in which Kant affirms the existence of unconscious cognitions and intuitions.Footnote ¹¹ Hence, there are good reasons for explaining away the tension, rather than reading Kant as denying the possibility of unconscious cognitions or intuitions. Here are three possibilities for reconciling unconscious cognition with the “Stufenleiter” passage: (i) the passage could be read not as defining cognition as objective perception, but as merely asserting that all objective perceptions are cognitions. This leaves open the possibility that some cognitions are not perceptions. (ii) Since Kant is interested in multiple kinds of consciousness, perhaps the kind of consciousness that animal (and other unconscious) cognitions lack is different from the kind of consciousness required for cognition (see McLear Reference McLear2011). (iii) Perhaps Kant uses the terms cognition and intuition in a nonstandard way in the “Stufenleiter,” e.g., because (general and transcendental) logic excludes unconscious representations from consideration (see V-Lo/Busolt 24:635; Log 9:33). I conclude that the apparent tension in Kant’s attribution of unconscious cognitions to animals can be diffused.

Another remark, to be found in the transcript of Kant’s anthropology lectures of 1784–1785, adds further weight: “Animals do have representations of the world but not of their I. Consequently, they are not rational beings” (V-Anth/Mron 25:1215).Footnote ¹² To have “representations of the world” is to have mental states that are intentionally directed at external things.

Together, these three passages provide robust evidence that Kant ascribes intentional states to animals. Are these intentional states intuitions, as premise (2) of the Argument from Animal Intuitions claims? Since animals possess sensibility but lack understanding, their representations of objects must be sensible representations, so it’s natural to conclude that they are intuitions. This is borne out by three more passages from lecture transcripts:

The first two passages simply attribute intuitions to animals. In the third, Kant reflects on the capacities an animal would have by imagining a creature lacking higher cognitive faculties. He states that such creatures would still have the capacity for “intuition with presence […] of the object” and “[intuition] without presence of the object,” i.e., that animals without understanding could still have intuitions.

It’s unsurprising that Kant classifies the cognitions of animals as intuitions: he holds that “besides intuition there is no other kind of cognition than through concepts” (A68/B92–93). This is reaffirmed in the numerous passages in which he presents the contrast between intuitions and conceptual representations as a disjunction, i.e., an exhaustive distinction (A320/B376–77; Log 9:91; Refl 1705 16:88; V-Lo/Wiener 24:805). Since animals lack concepts, it follows that whenever Kant credits them with cognitions he must tacitly be crediting them with intuitions.

3.b. Authenticity

I have identified six passages providing evidence that Kant credits animals with intuitions. Let me now discuss the possibility of rejecting these texts as inauthentic. All these passages come from sources of imperfect reliability. Though published in Kant’s lifetime, the Jäsche Logic was not directly written by Kant; it was compiled by a former student of his, using Kant’s marginalia and perhaps one or more lecture transcript. As a result, the consensus is that “one cannot simply assume […] that Jäsche’s manual is a reliable statement of Kant’s views” (Young Reference Young1992, xvi–xviii; cf. Boswell Reference Boswell1988). The other sources are still more questionable: they are students’ transcripts from Kant’s lectures. They have barely been edited and were certainly not checked for accuracy by Kant. Discussing the logic lectures, the editor of a recent edition writes, “[O]ne cannot look to [them], in general, for precise, carefully worded formulations of fundamental points” (Young Reference Young1992, xix). I’ve argued that the texts display a commitment to animal intuitions, but it remains possible that in doing so they misrepresent Kant’s views. Would it be legitimate for the Conceptualist to reject the textual evidence as inauthentic?

There are two reasons why this would be inadvisable. Firstly, the evidence from all these sources points in the same direction. I have identified passages from six different works asserting that animals have representations of objects, and I am unaware of any passage denying this. We would expect any transcription errors to be distributed fairly randomly, pointing in divergent directions. On the contrary, these passages all support the same conclusion thus rendering it highly implausible that they are all errata. Secondly, historians of philosophy should not be too cavalier about declaring passages inauthentic. Our task is to find the most coherent interpretation of the texts we have, so we should demand positive reason before excluding any passage from consideration. The Conceptualist might reply that these texts are shown to be inauthentic by the very fact (if it is a fact) that they cannot be reconciled with Conceptualism. However, this would simply beg the question against the Nonconceptualist, as well as conceding that the passages provide prima facie support for Nonconceptualism. At this stage, it seems likely that the Conceptualism/Nonconceptualism debate can only be settled by weighing prima facie evidence—no knock-down arguments have been identified by either side. It follows that Conceptualists should be reluctant to concede that they cannot accommodate these passages.

A further possibility is that there are more substantive reasons for denying Kant’s commitment to animal intuitions. In a recent article, Hein van den Berg (Reference van den Berg2018) argues that Kant doesn’t credit animals with “objective perceptual awareness” by comparing his views with two of his predecessors, Reimarus (1694–1786) and Buffon (1707–1788). The article highlights some interesting parallels, but its main argument seems to rely on an ambiguous use of its central term, viz the notion of a “blooming, buzzing confusion.” At best, van den Berg provides evidence that, for Kant, animal representations are “blooming, buzzing confusions” in the sense of being “confused” or “obscure” (7). But he seems to conclude on this basis alone that animals’ representations are “blooming, buzzing confusions” in the further sense of lacking intentionality (8). However, as noted above, Kant holds that some unconscious or “obscure” representations are nonetheless intentional. Therefore, it isn’t legitimate to infer that animals’ representations lack intentionality from the fact that they are obscure. Without argument to bridge this gap, van den Berg’s case is not compelling.

A different reason for denying the authenticity of the texts is that they are in tension with certain passages in KrV which seemingly state that synthesis according to the categories is a precondition for intentionality.Footnote ¹⁶ Here are two particularly compelling passages:

On the face of it, these passages say that, without the forms of unity introduced by the understanding, sensible representations would lack intentionality. One might try to reply that Kant is referring to some higher cognitive achievement than mere intuition. However, these are not statements about the objects of “cognition,” “thought” or “experience” (as with some other passages, e.g., A111, A129–30), but about “appearances,” i.e., the objects of intuition (A20/B34; A34/B51; A35/B52).Footnote ¹⁷

Do these KrV passages really conflict with the texts assembled above? I do not believe so. There are good reasons for thinking that Kant is talking here specifically about the preconditions for intentionality in human minds. He cannot be talking about the minds of all beings: that would include the divine intellect, to which Kant’s talk of “experience” would be inapplicable. The remaining options are either that Kant is considering sensible minds in general, i.e., both human and animal minds, or that he is only considering human minds. But Kant explicitly signals that he is discussing “how objects are to be given to us,” and he talks freely about apperception without signalling that some minds lack this capacity. This strongly suggests that Kant is only making assertions about human minds here. The availability of this reading means that these passages don’t straightforwardly contradict the textual evidence amassed in section 3.a. Of course, what Kant says about human minds in KrV might have ramifications for animal minds—several Conceptualist readers have thought that Kant’s account of human cognition entails that animal minds lack intentionality. But I will argue below that Conceptualists needn’t accept this. For now, suffice it to say that there is no direct conflict.

Pending other substantive objections, we ought to conclude that Kant really does credit animals with intuitions. Therefore, the Conceptualist cannot respond to the Argument from Animal Intuitions by denying either of its premises. The next section begins exploring the one remaining option: questioning the argument’s validity.

5.a. Textual motivations

There are two passages in which Kant suggests that human intuitions and animal intuitions are qualitatively different. These give us prima facie reason to take Restricted Conceptualism seriously. The first, which came up earlier, is this:

Here, Kant apparently states that animals’ intuitions are different in species from our representations. If Nonconceptualism were true, this would be surprising. On that view, there is a class of our representations, namely our (most basic)Footnote ²¹ intuitions, which float free from the cognitive differences between animals and ourselves. Hence, we would expect Kant to say that the representations of animals are different in species from our concepts, judgments, etc., but of the same species as our intuitions. But this is not what he asserts. Instead of equating animals’ intuitions with humans’ (most basic) intuitions, his phrasing indicates that their intuitions are “different in species” from any representations we possess, including our intuitions. Therefore, the passage poses a problem for Nonconceptualism.Footnote ²² In contrast, it fits perfectly with Restricted Conceptualism—it supports the idea that our intuitions, in virtue of being generated by the understanding, are qualitatively different from anything animals could possess.

Here, Kant doesn’t specify what kind of representations the cow “may” have, but given what we saw in section 3, I see no reason to doubt that he would credit them with sensible representations of objects.Footnote ²⁴ Kant is more tentative here, but he still avoids equating the cow’s representations with our own sensible representations. Instead, he raises the possibility that the cow’s representations are “completely different” from our representations. As above, the potential contrast is not between animals’ intuitions and our concepts, but between animals’ intuitions and our representations in general, including—it is implied—our intuitions. Again, the Restricted Conceptualist can readily explain this. In contrast, the Nonconceptualist, who holds that the understanding plays no role in producing our (most basic) intuitions, is faced with a puzzle. Why might the cow’s intuitions be “completely different” from all our representations, if our (most basic) intuitions are independent of the cognitive differences between ourselves and cows?

I don’t want to overstate what these passages show. The Volckmann passage positively affirms that the intuitions of animals are “different in species” from any representations humans possess. However, as noted, “one cannot look to [lecture transcripts], in general, for precise, carefully worded formulations of fundamental points” (Young Reference Young1992, xix). The Anthropology Manuscript passage is from Kant’s own hand, but it is much more tentative, raising the possibility that animal intuitions are qualitatively different without fully endorsing it. Hence, it would be unwise to make a positive case for Restricted Conceptualism based on these texts alone. What the passages do show is that the burden of proof is by no means stacked against Restricted Conceptualism. There is good prima facie reason to take seriously the hypothesis that, for Kant, the understanding is implicated in the production of human intuitions, making them qualitatively different from anything animals possess. I have not come across any evidence that Kant denies the existence of a qualitative difference between human and animal intuitions. So, if anything, these two passages make Restricted Conceptualism seem more likely than Nonconceptualism.

5.a. Restricted Conceptualism in detail

I’ll now propose a detailed form of Restricted Conceptualism and argue that it is both prima facie plausible as a reading of Kant and genuinely distinct from Nonconceptualism. Since our overall goal is to assess the dialectical efficacy of the Argument from Animal Intuitions, I won’t seek to establish whether Restricted Conceptualism is ultimately superior to Nonconceptualism. (That would require evaluating all arguments relevant to the Conceptualism/Nonconceptualism debate.) Instead, I’ll draw on an array of prima facie evidence that Nonconceptualists will likely claim is outweighed by other considerations. If I can show that Restricted Conceptualism is prima facie attractive (pending other arguments for Nonconceptualism), this will be enough to show that the Argument from Animal Intuitions is dialectically ineffective.

In short, the position I propose is: (i) humans’ intuitions are produced through synthesis of imagination for which the understanding is responsible. (ii) In animals, merely associative imagination generates intuitions. (iii) Owing to this, the intuitions of animals cannot represent the objective temporal structures (e.g., persistence, succession) that allow us to have thoughts about the objects we intuit. (iv) Animals’ intuitions can still represent objects as bundles of features located in space. They are representations of external things, not merely subjective states. (v) But, since associatively generated intuitions are intrinsically incapable of being taken up into thought, they couldn’t belong to a human mind, the boundaries of which are fixed by the unity of apperception.

(i) It is a familiar part of Conceptualist readings that human intuitions are produced through synthesis of imagination, for which the understanding is responsible. Strong prima facie evidence is provided by passages like those discussed in section 3.b, e.g., “all appearances, insofar as objects are to be given to us through them, must stand under a priori rules of their synthetic unity” (A110).Footnote ²⁵

To my mind, the most compelling evidence that human intuitions are produced through synthesis of imagination is Kant’s assertion that

This is further accentuated by his assertion that without the imagination’s synthesis of reproduction “no whole representation […], not even the purest and most fundamental representations of space and time, could ever arise” (A102). “Representations” encompasses both concepts and intuitions. Hence, that no representations of space and time can arise without synthesis of imagination entails that no intuitions of space and time can. One might think this evidence is outweighed by other considerations,Footnote ²⁶ but there is no denying that this is strong prima facie evidence that the intuitions of space and time depend on synthesis of imagination.Footnote ²⁷

Plausibly, if pure intuitions require synthesis, so do our empirical intuitions. Kant holds that we must possess pure intuitions of space and time in order to be capable of representing the spatial and temporal relations exhibited by all our empirical intuitions (see A23/B38; A30/B46; A165/B206), which suggests that “synthesis of the imagination” is required to generate the spatial and temporal features of intuitions whether they are a priori or empirical. From this vantage point, it becomes plausible that Kant’s descriptions of the imagination’s “apprehension” of sensible material describe a process through which empirical intuitions are first produced, so that their material features depend on synthesis of imagination too.Footnote ²⁸ To be clear, I don’t intend to have established once and for all that, for Kant, intuitions are produced through synthesis of the imagination; we might still conclude that, all things considered, this evidence is outweighed. Nevertheless, there are solid prima facie motivations for thinking this.

There is also good prima facie evidence that, in humans, synthesis of imagination is carried out by the understanding.Footnote ²⁹ Kant describes the imagination’s synthesis as “an effect of the understanding on sensibility” (B152) and claims that

Together, this means there is good prima facie motivation for the first part of the Restricted Conceptualism I’m proposing, viz that human intuitions depend on synthesis of imagination for which the understanding is responsible.

(ii) How can this be reconciled with the possibility of animal intuitions? First, we should note that Kant attributes imagination to animals, and that he explicitly contrasts the imaginative powers of animals and humans. Regarding capacities for “reproductive imagination,” he writes that they

By claiming that there ought to be two different names for this capacity, Kant is explicitly differentiating two species of it: reproductive imagination with and without apperception. This means that an animal’s imagination can combine sensible material through receptive and associative processes, but not according to “rules of […] synthetic unity” (A110).

Hitherto, Conceptualists have tended to think these kinds of imaginative process could not result in mental representations with intentionality.Footnote ³⁰ However, owing to the evidence discussed above, we should consider another possibility: that these associative processes are capable of generating intuitions, i.e., singular sensible representations of objects. On this reading, our intuitions depend on the understanding, whereas the intuitions of animals are produced by associative tendencies of the imagination. This interpretative route upholds the claim that human intuitions are produced by the understanding. In doing so, it accommodates the evidence that motivates traditional forms of Conceptualism. However, it also accommodates Kant’s commitment to animal intuitions. Therefore, it amounts to a promising form of Restricted Conceptualism.

This is already enough to ensure that Restricted Conceptualism is distinct from Nonconceptualism. Nonconceptualists deny that human intuitions are produced by the understanding (either by claiming that intuitions precede synthesis or that they are produced through a nonintellectual synthesis). But there is more to be said to spell out the difference between human and animal intuitions and amplify the dependence of the former on the understanding.

(iii) I’ve proposed that (there is strong prima facie evidence that) human intuitions are produced by the understanding, whereas animal intuitions are produced by associative imagination. There is good reason to think that this difference in origin leads to an intrinsic difference between animal and human intuitions. Kant holds that sensible representations produced by the understanding exhibit distinctive kinds of unity, specifically with respect to their temporal structure. Plausibly, all intuitions produced by the understanding exhibit one or more of the temporal structures which correspond to the categories, e.g., persistence (corresponding to <substance>, A144/B183), succession (corresponding to <cause>, ibid.).Footnote ³¹ It is in virtue of exemplifying these temporal structures that intuited objects can be subsumed under the categories (A138–40/B177–79). Moreover, Kant holds that our thoughts can only have relation to objects in virtue of involving the categories: “no object can be thought without them” (KpV 5:136); “by these concepts alone can [the understanding] understand something in the manifold of intuition, i.e., think an object for it” (A80/B106; cf. A51/B75). Thoughts must deploy categories in order to relate to objects, but intuited objects only fall under categories in virtue of the temporal structures they exhibit. Hence, it is in virtue of representing these temporal structures that our intuitions allow us to think about the objects they present.

In contrast, an animal’s associative power of imagination couldn’t possibly produce intuitions of these temporal structures. We can see this by considering Kant’s account of the preconditions for representing objective temporal structure: not only of producing intuitions with a certain subjective temporal form, but intuitions that represent objects as having a certain temporal form. The clearest discussion comes in the Second Analogy. Kant accepts that associative imagination can produce representations with subjective temporal form (i.e., they can produce a “subjective order of perceptions”), but argues that this is insufficient for representing temporal structure in the objects (i.e., they cannot produce representations of an “objective order”). This is because a merely associative combination of sensory material would be “entirely arbitrary” (A193/B238); for any temporal ordering the imagination happens to introduce, it could equally well have produced the reverse order (B233, A201/B246). Consequently, a temporal ordering introduced by associative processes will ultimately be attributable to contingencies about the perceiver, rather than features of the objects perceived. The result is that the temporal dimension of intuitions produced through mere association will not have the semantic significance of representing temporal structure in the objects.

In humans, where the imagination’s combination is led by the understanding’s rules, this combination can be rendered “necessary” rather than “arbitrary,” enabling our intuitions to represent objective temporal structure.Footnote ³² But in animals, nothing can remove the arbitrariness, so temporal structure in the objects can never be represented. It is not that animals intuit objects in the same way we do, and merely lack the concepts to think about them; we intuit objects as temporally structured unities, while animals cannot achieve this cognitive feat. And this difference has important ramifications: we intuit things as structured in ways that allow us to think about them; animals do not intuit things as having this structure. The fact that human intuitions are generated by the understanding makes them qualitatively different from animal intuitions, and no substitute for the understanding could produce intuitions of this kind.

(iv) I’ll now address two objections.Footnote ³³ Firstly, I’ve claimed that, since the temporal form of animals’ intuitions is introduced by associative processes, it cannot purport to represent objective temporal structure.Footnote ³⁴ Doesn’t this undermine the claim that animals’ sensible representations exhibit intentionality at all? In response, note that the argument I’ve given is quite limited in scope: it applies only to representing objective temporal structures. Admittedly, I suspect it generalizes to some other kinds of objective connection, e.g., representing the generality of properties. An animal might associatively link the representations of two objects, but without rules governing which representations belong together, this subjective linkage would not represent the two objects as objectively similar, i.e., as sharing a property. (This agrees with Kant’s claims about animals’ incapacity to form general representations; see V-Met/Mron 29:888.) But even this broader conclusion does not undermine the intentionality of animal’s mental states. Nothing I’ve argued prevents animal minds from representing particular features (though these may be presented as tropelike qualities rather than instances of universals). The same goes for spatial location: nothing in Kant’s account indicates that placing a feature in space requires representing an objective connection between the contents of different representations. Hence, Restricted Conceptualism can maintain that animals’ intuitions represent external things as spatially located bundles of features. In terms of representational content, this is importantly different from intuiting things as unified complexes bearing general properties. Nevertheless, animals have intentional representations of external things, rather than subjective states devoid of intentionality.Footnote ³⁵

(v) Now the second objection: Kant holds that humans, as well as animals, possess associative powers of imagination. If associative imagination is capable of producing intuitions within the animal mind, why couldn’t it produce “animal-style” intuitions within the human mind, i.e., associatively generated intuitions that don’t represent objective temporal structures? In response, I’ll argue that, necessarily, all human representations meet the conditions for belonging to the unity of apperception. This means that all our intuitions must exhibit the temporal structures corresponding to the categories and, therefore, must be produced by the understanding.

There is plenty of evidence that Kant’s goal in the Transcendental Deduction is to establish just this. His stated “aim” is to show that “all sensible intuitions stand under the categories” (B143), that the categories apply to “all objects of our senses” (B145), and that there is a “necessary coherency of the understanding with appearances by means of the categories” (A119) (n.b. not just all “thoughts” or “cognitions,” but all “intuitions” and “objects of our senses”). The categories wouldn’t apply to the contents of animal-style intuitions. So, by arguing that all our intuitions must “stand under the categories,” he is ipso facto arguing that our minds cannot contain animal-style intuitions.

It’s a difficult question how exactly Kant reaches this conclusion (equivalent to asking, “What is the argument of the Transcendental Deduction?”). Let me offer one possible reading, which underscores the difference between animal and human minds. Kant can reach the conclusion that all human intuitions fall under the categories via the dictum that “The ‘I think’ must be able to accompany all my representations” (B131). He holds that, strictly speaking, I am a thinking being (see Rosefeldt Reference Rosefeldt2000, 14–15), but doesn’t construe this as an immaterial substance encountered through introspection. Instead the thinking being is picked out procedurally, in a topic-neutral fashion, through the limits of possible self-ascription in first-person thought. This conception of the thinking being’s limits is nicely illustrated by Kant’s account of what it means to ascribe a mental representation to another human: to do so is to judge that it is possible for there to be a first-person thought in which that mental representation is self-ascribed (A347/B405; cf. A353, A354; Rosefeldt Reference Rosefeldt2000, 22–25). This makes it logically impossible to ascribe a representation to a thinking being while denying that he/she could become self-conscious of it (A117n, B131–32; Br 11:52).Footnote ³⁶ Applying this general dictum to sensible representations, we reach the conclusion that “all intuitions are nothing for us and have nothing at all to do with us unless they can be taken up into consciousness” (A116; cf. B132–33, B138). But the categories are the “conditions under which alone the manifold [of intuition] can come together in one consciousness”; so all our intuitions must “stand under the categories” (B143; cf. A125). To belong to a human mind, intuitions must represent the objective temporal structures corresponding to the categories, and so they must be produced by the understanding not associative imagination.Footnote ³⁷

Ascribing mental representations to animals is governed by entirely different rules. It doesn’t involve thinking it possible for the representation to be self-ascribed in a first-person thought; instead, it’s a matter of using analogical thinking to speculate about the causes of the behaviour we observe. For example, Kant claims that we ascribe representations to beavers by observing their dam building, seeing the similarity with the work of human craftsmen and positing analogous mental causes in the beaver’s mind (KU 5:464; cf. H 7:397). Hence, the criterion for a mental representation’s belonging to an animal mind is simply that it plays a certain causal role within the animal’s life.

One last clarification: if a human’s associative imagination can’t generate animal-style intuitions, what does it do? On the picture I’m defending, our associative processes go to work on a plethora of sensible representations that already stand under the categories. Rather than producing novel representations that don’t adhere to the categories, a human’s associative imagination takes categorially structured representations and combines them into novel complexes. This combining-into-complexes can be highly idiosyncratic and nonsensical, with the associations deviating significantly from how the empirical world is. Nevertheless, the components of these associations will be unified objects with properties. This account fits with Kant’s views on the cognitive errors to which we are prone. Associative “habit” leads to “prejudice” when we mistake our idiosyncratic combinations for real patterns in the empirical world; but nowhere does Kant worry that we might be swayed by unruly representations that are intrinsically inaccessible to self-consciousness, as animal-style intuitions would be. Rather, he insists that any pernicious influence by associatively generated representations can always be exposed and rectified through reflection (Log 9:75–76). In Kant’s view, then, even the most irrational products of humans’ associative imagination will meet the basic preconditions required for self-consciousness, i.e., they will exhibit objective temporal structures corresponding to one or more of the categories.

This completes my proposed version of Restricted Conceptualism, for which I’ve argued there is good prima facie evidence. Necessarily, humans’ intuitions present objects as exhibiting temporal structures which correspond to the categories. Such intuitions are produced, and could only be produced, by the understanding. Animals’ intuitions are produced by associative powers of imagination. At the very least, they present external things as spatially located bundles of features, but they cannot represent any objective temporal structure in these things. Animals not only lack the ability to think; they lack the ability to have intuitions of a kind that would allow intuited objects to be taken up into thought.