Kant on Mind-Dependence: Possible or Actual Experience?

Abstract

In Kant’s idealism, all spatiotemporal objects depend on the human mind in a certain way. A central issue here is whether the existence of spatiotemporal things requires that these things are, at least at some point, objects of some actual experience or of a merely possible experience. In this essay, I argue (on textual and philosophical grounds) for the latter view: spatiotemporal things exist (or spatiotemporal events occur) if they are objects of a (suitably qualified) possible experience.

In Kant’s idealism, all spatiotemporal objects depend on the human mind in a certain way. What this ‘certain way’ amounts to has been strongly debated in Kant scholarship. One crucial facet of this issue is whether the existence of spatiotemporal things requires that these things are objects of some actual experience (I shall call this view Actualism) or of a merely possible experience (Possibilism). In this article, I examine this question in light of important recent scholarship. I argue that the overall textual evidence and independent philosophical considerations jointly lend stronger support to Possibilism: on Kant’s view, spatiotemporal things exist (or spatiotemporal events occur) if they are objects of a (suitably qualified) possible experience.

The essay proceeds as follows. In section 1, I clarify certain key terms and explain why some (otherwise important) questions about Kant’s idealism are tangential to the topic that is the focus of my discussion. In section 2, I examine the debate between Actualism and Possibilism on textual grounds; I argue that the overall textual evidence favours a Possibilist reading. In section 3, I consider the philosophical considerations that each side can adduce on its behalf. I argue that Possibilism is not only, contrary to Actualist arguments, fully consistent with the main tenets of Kant’s doctrine; it also yields a more attractive version of transcendental idealism than Actualism.

1. Initial clarifications

I begin by clarifying some key aspects of Kant’s relevant notion of ‘experience’. ¹ Experience, for Kant, is a special type of representation, namely, an objective cognition of the spatiotemporal world that comprehends given sensible data (intuitions) under conceptual rules. Any representation that is to qualify as an experience must accord with two kinds of experience-forms. First, it must conform to the forms of human sensibility, space and time, which are (roughly) ways in which the human mind orders and structures the empirical input (i.e. the sensations) that it receives from without. Second, it must accord with the conceptual or intellectual forms of experience that are provided by our faculty of understanding. These include the categories such as substance or causality, general categorial laws such as ‘Every event has a determining cause’ and the highest (most general) law of understanding that Kant calls the principle of the unity of apperception.

The notion of unity signals a crucial theme in Kant’s conception of experience: a representation never yields an experience of objects considered all by itself, in isolation from other (at least possible) representations. Rather, a representation such as (paradigmatically) an empirical intuition or perception provides objective experience only if it is integrated into a single, unified, complex whole of perceptions that jointly make up a cohesive stream of consciousness, an entire course of experience whose component parts hang together in a non-arbitrary fashion and thereby jointly represent a single, unified, complex world (cf. A110). ² Experience thus requires ‘the representation of a necessary connection of perceptions’ (B218). ³ Hence, the question of whether things depend for their existence on being objects of an actual or possible experience amounts to the question of whether things depend for their existence on being objects of an actual or a possible unified, cohesive course of experience.

There are two important issues that I will set aside in this article. First, I shall presuppose that, in Kant’s idealism, spatiotemporal objects owe their existence (or at least their spatiotemporal way of existing, their spatiotemporal character) to our representational faculties. This presupposition is controversial insofar as it leads into debates about whether Kant’s idealism is to be understood as a metaphysically weighty doctrine that asserts the mind-dependence of spatiotemporal things or properties, or rather as a deflationary doctrine that abstains from such ontological pronouncements. Proponents of a deflationary interpretation would presumably want to reject or at least strongly qualify the idea that according to transcendental idealism spatiotemporal objects do not exist apart from being objects of our actual or possible representations. If one thinks that the very notion that things or properties existentially depend on our representations signals a dubiously metaphysical reading, then one might think that the question I seek to address – whether Kant accepts Actualism or Possibilism – is moot or confused. I cannot fruitfully engage with this kind of interpretation in the context of this article. I only want to note that there is strong textual evidence for ascribing to Kant a weighty notion of mind-dependence. ⁴ Moreover, some proponents of deflationary readings also rely on the notion of mind-dependence when they expound Kant’s views. ⁵

Second, leaving radically deflationary readings aside, my argument does not (as far as I can see) hinge on any specific account of what it means to say that spatiotemporal things ontologically depend on being objects of actual or possible human experience. For proponents of two-object readings, this means that spatiotemporal objects like chairs or trees do not exist apart from being objects of human representations, while the things in themselves that somehow ground these spatiotemporal appearances form an ontologically distinct class of entities. ⁶ For proponents of non-deflationary one-object or two-aspect readings, the dependence at issue concerns not the very existence but rather the spatiotemporal properties, character or constitution of objects. ⁷ Proponents of this view might still want to say that spatiotemporal objects depend for their existence as spatiotemporal objects – as things exhibiting spatiotemporal features – on being objects of human representations. But they will stress that the things which appear to us in a spatiotemporal form exist apart from the human mind. On this view, it is (in some sense) the same entity which has both a non-spatiotemporal constitution in and by itself and a spatiotemporal constitution insofar as it relates or ‘conforms’ (Bxvii) to our representational faculties. ⁸

As a matter of fact, two-object readings seem to favour Actualism whereas one-object readings gravitate towards Possibilism. ⁹ But this is not a necessary link: neither of these two views is inherently committed to either Actualism or Possibilism. Thus, my argument in what follows should be compatible both with a two-object and a (non-deflationary) one-object view.

Against this, one might argue that Actualism requires a two-object view since it identifies spatiotemporal appearances with actual representations and things in themselves with extra-mental entities. ¹⁰ However, a phenomenalist equation of appearances and representations need not be the only way to develop a two-object view. Moreover, even proponents of a phenomenalist interpretation might adopt a version of Possibilism. They might hold that appearances are to be understood as representations in counterfactual terms, e.g. as representations that a subject would have if it properly synthesized certain sense-data or as representations that could be constructed in a suitable manner from the subject’s actual representations. ¹¹

In what follows, I defend Possibilism in dialogue with two quite different Actualist readings that have been developed by James Van Cleve and Anja Jauernig. According to Van Cleve, ‘we should construe’ Kant ‘as saying that appearances are representeds that have no being apart from the representing of them’ or ‘from being represented’ (Van Cleve 1999: 7, 6). To say that a certain spatiotemporal object ‘exists is shorthand for saying that a certain kind of representations’ or sequence of representations ‘occurs’ (p. 9). Even objects in a ‘weighty’ sense, i.e. material things like ‘houses, ships and all the furniture of the earth’, are ‘logical constructions’ out of ‘non-weighty’ objects such as ‘patches of color and stretches of sound’ (pp. 93, 91). ¹² Since these non-weighty objects ‘are themselves constructions out of representations (in the ‘ing’ sense, i.e., representings)’, and since ‘the relation “is a construction out of” is transitive’ (p. 93), it follows that outer appearances like ships and houses also depend for their existence on the occurrent representings out of which they are constructed: matter ‘has no being apart from our perceptions’ (p. 71). ¹³ Thus, in Van Cleve’s Actualist reading Kant endorses Berkeley’s dictum esse est percipi (pp. 9–10). ¹⁴

Jauernig also claims that ‘appearances would not exist if they were not presented as existing’, which means that ‘their existence is thus closely tied to perception’ (Jauernig 2021: 74). But her account of the way in which appearances depend on actual representations differs from Van Cleve’s in important ways. For one, Jauernig treats appearances as real rather than merely virtual existents. ¹⁵ Moreover, Jauernig stresses the difference between occurrent perceptions and an entire course of shared, collective human experience: on her view, the relevant (existence-determining) notion of actual experience designates the entire ‘series of ew-accounts that the community of human minds develops throughout the course of history’ (p. 96). ‘Ew’ is shorthand for ‘empirical world’. An ew-account is a particular conception of the empirical world at a given time that involves: a list of the general (a priori, transcendental) laws of nature and of the special empirical laws that have been discovered by science thus far (including ‘commonsensical descriptions’ of these laws); ‘a record’ of all appearances that have been perceived or inferred from available empirical evidence thus far; and ‘a collection of empirical judgments and judgments of experience made so far concerning the determinations of these appearances, including their relations to each other’ (pp. 94–5).

Jauernig’s and Van Cleve’s accounts are the most sophisticated versions of Actualism that I am familiar with. My argument in what follows engages with both versions. I first examine the textual evidence. While a consideration of all relevant textual details is beyond the scope of this article, I argue that key passages from the first Critique suggest a Possibilist view.

2. Actualism versus Possibilism: textual data

I begin by examining some passages that Actualists use to support their reading. Van Cleve cites A104, where Kant says that appearances must not be regarded as objects apart from our power of representation. However, this statement suggests Possibilism rather than Actualism because the appeal to a power of representation goes significantly beyond the actual representations that result from specific exercises of that power: Kant here ties appearances to what we are capable of representing. Similarly, Kant’s remark (also cited by Van Cleve) that laws of appearances exist ‘only relatively to the subject in which appearances inhere insofar as it has senses’ (B164) is at least consistent with the Possibilist reading that appearances and their laws do not exist apart from their relation to the general representational (including sensible) capacities of cognitive subjects.

Van Cleve’s strongest textual case for Actualism rests on Kant’s remarks in the A-edition Fourth Paralogism. Here Kant says that only that ‘which is represented in’ space can be regarded as an actual spatial object; ‘nothing at all is in space except insofar as it is actually represented in space’ or insofar as it is ‘immediately given through empirical intuition’, i.e. perception (A374–5). However, these Actualist pronouncements must be taken with a grain of salt. Kant’s refutation of ‘empirical idealism’ in the A-edition Fourth Paralogism is one of the sections that he rewrote in the second edition – presumably, precisely because he recognized that this section had encouraged (as he learned from the Garve/Feder review) a Berkeleyan-Actualist understanding of his idealism that he repudiated. Actualist formulations are notably absent from the B-edition Refutation of Idealism. ¹⁶

Moreover, even in the A-edition Fourth Paralogism, at least some of Kant’s seeming Actualist remarks can be understood in a Possibilist vein. For instance, his claim that only what is actually represented in space ‘can be counted as actual’ (als wirklich gelten) (A374) need not be taken as supporting the view that only actually represented things exist. We might rather take this remark to suggest the epistemic point that only our actual representations provide us with valid grounds for judging that something actually exists. As Kant says a bit earlier (A371), with regard to sensible appearances ‘their immediate perception … is at the same time a sufficient proof of their actuality’. This is consistent with Possibilism because for Possibilists the domain of actual existents far outstretches our finite capacity for proving (on the basis of valid cognitive grounds that must involve actual sensations) what belongs to this domain.

This point is also relevant for assessing Jauernig’s claim that ‘Kant never characterizes the perceptions that establish the existence of appearances as possible but does explicitly characterize them as actual at several places’, for example, when he introduces his Postulate of Actuality (Jauernig 2021: 100). Possibilists can accept that only actual perceptions epistemically establish the existence of appearances. They can also take Kant’s initial formulation of the Postulate as a principle of cognizing actuality at face value: ‘The postulate for cognizing the actuality of things requires perception, thus sensation of which one is conscious’ (A225/B272). The fact that our ability to cognize appearances as existing depends on actual sensations does not entail that the very existence of appearance depends on actual sensations as well.

Let us now consider in more detail the textual implications of what Kant says in the Postulates and in thematically related passages from his solution to the Antinomies.

Kant’s Second Postulate states: ‘That which is connected with the material conditions of experience (sensation) is actual’ (A218/B266). This clearly establishes a link between the existence of appearances and what is given as actual through sensation. However, this link is not direct or straightforward. The first thing to note here is that Kant does not tie the actual existence of some empirical thing to the actual sensation of it. Rather, he argues that whatever exists in the spatiotemporal world must stand in a certain connection with some actual sensation: namely, in a dynamical connection according to the Analogies of Experience (A225/B272) that involves the a priori principles of substantiality, causality and community and the specific empirical laws of nature that instantiate these general principles (A231/B284). While sensation or perception ‘is the sole character of actuality’ (A225/B273), an empirical thing which we do not itself perceive or have perceived in the past can still be regarded as actual ‘if only it is connected with some perceptions in accordance with the principles of their empirical connection (the analogies). For in that case the existence of the thing is still connected with our perceptions in a possible experience, and with the guidance of the analogies we can get from our actual perceptions to the thing in the series of possible perceptions’ (A225–6/B273). Kant here ties the existence of (spatiotemporal) things to non-actual perceptions of these things that belong to a merely possible course of experience which is dynamically connected with actual (given) perceptions. This passage thus raises a problem for Jauernig’s claim that ‘Kant never characterizes the perceptions that establish the existence of appearances as possible’ (Jauernig 2021: 100). This claim passes over the distinction between the Actualist claim that spatiotemporal existence depends upon actual perceptions and the claim (which Kant makes in the above passages) that spatiotemporal existence depends upon merely possible perceptions which we can obtain from actual perceptions in accordance with proper dynamical rules of connection.

The sense in which the perceptions on which appearances existentially depend are ‘possible’ for us also requires qualification. Kant does not mean that these perceptions are really (empirically) possible for us given our empirical constitution. Rather, he means (roughly) that they are possible for us in principle, if we consider only the formal a priori conditions of our experience and abstract from the arbitrary empirical limitations of our physical sense organs ‘the crudeness of which does not affect the form of possible experience in general’ (A226/B273; cf. A522/B551). For example, if these organs ‘were finer’, then we could get from our immediate perception of a moving piece of iron to an immediate perception of magnetic matter as the causal source of this phenomenon (A226/B273). ¹⁷ This rather strong qualification of the sense in which the perceptions on which spatiotemporal existence depends are possible for us raises a problem for Jauernig’s claim that ‘sometimes Kant can be read as using “possible” and “actual,” not in a proper modal sense but in a primarily temporal sense, in which the terms mean “future” and “present” relative to the present moment’, so that ‘Possible perceptions or experiences in this sense are perceptions or experiences that, with respect to the present moment, will actually be had later on’ (Jauernig 2021: 99). ¹⁸ But Kant does not suggest that we ever will perceive magnetic matter at some future time (say, when bioengineering refines our perceptual capacities). Rather, he accounts for the existence of magnetic matter through the irreducibly modal claim that our actual perception of moving iron would lead us to an immediate perception of magnetic matter if our physical sense organs were finer, since the movement of the iron is dynamically (causally) connected with magnetic matter. This counterfactual grounds the existence of magnetic matter. Generalizing from this case, we can say: spatiotemporal appearances actually exist whenever we could, in principle (abstracting from the contingent limitations of our empirical sense organs), obtain a perception of these appearances from some actually given perception in accordance with proper dynamical laws of connection. ¹⁹ (Note that I do not intend this conditional – ‘if we could … obtain a perception of appearance x … then appearance x actually exists’ – to yield a sufficient account of idealist mind-dependence. ²⁰ )

There are further reasons for denying the Actualist suggestion that Kant conceives the relevant (existence-grounding) perceptions as possible in a non-modal, temporal sense which involves a kind of prediction that these perceptions actually will be had later on. For one, Kant never indicates that his seemingly modal phrases in these contexts really have a non-modal temporal meaning so that ‘possible perception’ means ‘perception that will be obtained at some future point’. Moreover, there is no valid basis for such predictions: for a variety of reasons, it is impossible to anticipate what we will be able to perceive at some future time based on our current perceptions. ²¹

Let us now consider some of the closely related remarks Kant makes in his solution to the Antinomy of Pure Reason. This solution hinges on the idealist denial that the spatiotemporal objects of our empirical cognition are mind-independent things in themselves. Crucial to Kant’s account here are two similar notions: an empirical regress from actually given objects qua conditioned phenomena to their empirical conditions, and an empirical progress from actually given perceptions to further perceptions that follow from them. ²² (Kant does not always use these terms consistently: he sometimes also speaks of ‘progress’ when he talks about moving backwards in the series of perceptions or conditions.) Through these notions, Kant seeks to clarify what the mind-dependent status of empirical objects amounts to in his transcendental idealism, i.e. what it means to say that spatiotemporal appearances ‘cannot at all exist outside our mind’ (A492/B520).

According to Jauernig, Kant holds that each empirical condition for some actually perceived conditioned object ‘exists only in virtue of us actually extending the relevant empirical regress to where it is located in the series’ (Jauernig 2021: 100). However, Kant specifies the cognitive course from the actual perception of the conditioned object to its existing conditions as a merely ‘possible progress in the series of appearances’ (A511/B539). To confirm that Kant grounds the existence of the conditions for a given conditioned object in a merely possible course of experience, let us consider his account of the truth-conditions for the judgement, ‘there could be living beings on the moon’.

According to Kant, this means that ‘we could encounter such beings in the possible progress of our experience’ (A492–3/B521). Notably, the explanans here invokes two different possibilities: Kant explains the possible existence of moon-inhabitants via our possible awareness of such beings (the first appeal to possibility) in a possible progress of experience (the second appeal to possibility). He does not indicate the conditions under which we indeed could become aware of (living) moon-inhabitants. But these conditions can be gleaned from his explanation of the conditions for the actual (rather than merely possible) existence of moon-inhabitants. To account for these conditions, Kant alters his analysis of the conditions of merely possible existence as follows: he replaces the first possibility (‘possible awareness of moon-inhabitants’) but retains the appeal to possible experience. Moon-inhabitants actually exist if they are actually contained in a possible experience whose content is dynamically connected with the contents of our actual perceptions: ‘for everything is actual that stands in one context with a perception in accordance with the laws of the empirical progression’. Since a dynamical connection involves a necessary link, this means that unperceived spatiotemporal things actually exist if we would necessarily encounter them in the (merely possible) case where our course of experience continues far enough: ‘To call an appearance an actual thing prior to’ (actual) ‘perception means … that in the continuation of experience we must encounter such a perception’ (A493/B521). The ‘must’ here indicates a necessary dynamical connection between our actual perceptions and the perceptions that we would have in the possible progress of our experience. ²³ Returning now to the issue of merely possible existence, we can see that the criterion for the possible existence of x (namely, the possible awareness of x in a possible experience) coincides with the criterion for the actual existence of x (namely, the actual awareness of x in a possible experience). This is because perceptions of x are possible relative to our actual state of experience just in case they are necessary relative to our actual state of experience. Our perceptual experience of x is possible in a real, more than logical sense only if this experience is dynamically, i.e. lawfully and necessarily connected with the contents of our actual perceptual experience. ²⁴

As I noted earlier, the fact that an appearance x actually exists apart from being the object of an actual perceptual experience is consistent with the further epistemic point that we can establish or cognize an appearance x as actual only if we happen to obtain actual empirical evidence that supports the belief in the existence of x. Thus, from our current epistemic vantage point we are unable to cognize whether moon-inhabitants do (or really could) exist.

To confirm that in Kant’s view a possible course of experience (proceeding from some actual experience) determines actual spatiotemporal existence, let us consider his remarks on the past existence of things or the past occurrence of events. He says: ‘The actual things of past time are … actual in past time only insofar as I represent that a regressive series of possible perceptions (whether under the guidance of history or in the footsteps of causes and effects) leads to a time-series that has elapsed as the condition of the present time’ (A495/B523). A past event actually occurred just in case the perception of it is contained within the regressive series of perceptions that we would reach if we extended the series far enough backwards (in a rule-governed manner), starting from our current state of experience. This suffices for actual past existence even if the relevant past moment in time is so far removed from our present experience that we cannot think or conceive this moment or the things that filled it: ‘all those events which have elapsed from an unthinkable past time prior to my own existence signify nothing but the possibility of prolonging the chain of experience, starting with the present perception, upward to the conditions that determine it in time’ (A495/B523). I can legitimately say that unperceived past objects ‘exist prior to all my’ (actual) ‘experience’, but this ‘means only that they are to be encountered in the part of experience to which I, starting with the’ (actual) ‘perception, must first of all progress’ (or better, regress) (A496/B524).

When Kant says that I ‘must first of all progress’ (better, regress) to past temporal moments, he does not mean that these past moments actually unfolded only once one actually reached them in the empirical regress. Rather, the ‘must’ here has a prescriptive sense. Reaching these past moments by moving backwards (in a rule-governed fashion) from my actual experience is set for me as a cognitive task-to-be-accomplished: I ‘must’ first of all reach these past moments insofar as reaching them is set ‘as a problem for the understanding, thus for the subject in initiating and continuing, in accordance with the completeness of the idea, the’ (rule-governed) ‘regress in the series of conditions for a given conditioned’ (A508/B536), for example, from currently perceived events to past events that are not yet contained in our actual course of experience. The ‘idea’ that Kant mentions here is the regulative idea of an absolute unconditioned whole of the series of conditions that would be encountered in a fully comprehensive experience, ‘the thought of a possible experience in its absolute completeness’ (A495/B524). Since this thought only represents an asymptotical end point of empirical cognition that finite cognizers can never fully reach, ‘the idea of reason will only prescribe a rule to the regressive synthesis in the series, a rule in accordance with which it proceeds from the conditioned, by means of all the conditions subordinated one to another, to the unconditioned, even though the latter will never be reached’ (A510/B538).

The Possibilist implications of these passages from the Antinomy can be summarized as follows. Suppose we would eventually encounter some past event in the rule-governed (dynamically determined) regress from our actual perceptions if we extended the regress far enough: in that case, the event actually occurred, even if no human cognizer ever extends the regress far enough. Likewise, suppose we would eventually obtain the perception of moon-inhabitants if we extended the rule-governed progress from our actual perceptions far enough: in that case, moon-inhabitants actually exist, ‘even if no human being has ever perceived them or ever will perceive them’ (A496/B524).

I suggest that Kant’s endorsement of Possibilism in the above passages is motivated by his attempt to develop a metaphysics of the spatiotemporal world that combines realist and idealist elements. Tying the existence of spatiotemporal objects to the merely possible (rule-governed) continuation of our experience goes some way towards accommodating the realist intuition that the existence of these objects does not depend on the arbitrary, contingent ways in which the actual course of our experience happens to continue, forwards or backwards. (I further expound this point in section 3.) But although empirical things may exist aside from being actually perceived, they do not exist ‘outside this’ (merely possible) ‘progress of experience’ that begins from some actual experience; thus spatiotemporal objects do not exist ‘without relation to our senses and possible experience’ (A493/B521). This is the idealist component of Kant’s view: spatiotemporal things or events do not exist or occur independently of their relation to our mind and our course of experience.

A further textual datum that supports Possibilism is Kant’s twofold definition of the natural world or (simply) nature. First, ‘Nature is the existence of things, insofar as it’ (i.e. this existence) ‘is determined according to general laws’ (Prol, 4: 294); ‘Nature is the epitome of that which exists’ in space and time (V-Met/Volckmann, 28: 364). Second, from the a priori laws or conditions of possible experience we can ‘determine nature as the entire object of all possible experience’ (Prol, 4: 297); ‘nature and possible experience are altogether the same’ (4: 320). These two definitions jointly entail that things actually exist in space and time if they belong to nature qua entire object of all possible experience.

I have argued that several important passages from Kant’s writings suggest a commitment to Possibilism. However, even if this textual argument is correct, there is still the question of whether Possibilism yields a viable philosophical option for Kant.

3. Actualism versus Possibilism: philosophical considerations

To assess whether a philosophical case can be made for Actualism, let us first consider an argument that Jauernig gives to show that Possibilism is inconsistent with Kant’s idealism. ²⁵ This argument begins by noting that we indeed can continue the progress or regress from our actual experience to its further components: ‘it is certainly possible for us to cognize the conditions of a given appearance and the conditions for the conditions and so on’. Now suppose Possibilism is correct: appearances exist as objects of a possible experience. According to Jauernig, it follows that the entire series of conditions for a given conditioned ‘would exist as a never ending series even before we actually embarked on the regress’. But this is untenable because Kant’s idealism clearly denies that the total series of conditions, the entire empirical world, is given to us as an existing thing. Thus, to ensure that the empirical world or series of conditions does not already exist as an actual object prior to our (regressive or progressive) synthesis of perceptions into a unified whole of experience, ‘the regress in which appearances and the empirical world exist must be an actual regress’. This is an important line of argument that brings out a key aspect in the debate between Actualism and Possibilism. Above, I suggested that Possibilism is motivated by Kant’s attempt to combine a robust idealism with a robust realism. Jauernig’s argument suggests that the robust Possibilist-realism comes at the expense of Kant’s idealist tenet that the (empirical) world does not exist as a complete, determinate thing in itself – a tenet which carries major systematic weight since Kant believes that it is needed to solve the Antinomies.

Possibilists can respond here as follows. Jauernig’s initial premise states that ‘it is … possible for us to cognize the conditions of a given appearance and the conditions for the conditions and so on’. But the phrase ‘and so on’ conceals an ambiguity between the claim (1) that we can, in principle, always proceed from one specific member in the series of appearances to the next, and the claim (2) that we can carry on the regress/progress in the series to completion until we have cognized all the relevant conditions or consequences. Claim (1) is true whereas claim (2) is false since the relevant regress/progress extends ad infinitum or at least ad indefinitum and thus cannot ever be completed by finite human cognizers. But Jauernig needs (2) in order to commit Possibilism to the untenable consequence that the entire, complete series of appearances or empirical world exists as an object of a merely possible experience. With (1), Possibilism does not have this consequence: since (1) does not entail that the entire series of appearances or the entire (spatiotemporal) world is a possible object of experience, the Possibilist claim that appearances exist as objects of a (merely) possible experience does not imply that the series of appearances or empirical world exists as a complete, determinate whole. Thus, Possibilism seems compatible with the core idea behind Kant’s idealism that he invokes to solve the Antinomies.

Jauernig offers a further philosophical consideration in favour of Actualism. She claims that Actualism is preferable over Possibilism because:

[I]t renders appearances less ontologically suspect. For something to exist in virtue of being represented is already somewhat peculiar, but for something to exist in virtue of being possibly represented sounds like a situation in which a magician pulls an actual rabbit out of a possible hat. Since Kant is not at risk for winning any popularity contests for his idealism anyway, we should try not to saddle him with an especially eye-brow raising version if it can be avoided. (Jauernig 2021: 99)

Jauernig rightly points out that any idealist view which asserts that spatiotemporal things depend on human representations is likely to raise suspicion because it offends deep-seated realist intuitions. She claims that a Possibilist version of this view yields an especially odd or counter-intuitive form of idealism, presumably because it tries to account for actual existence in terms of mere representational possibilities. However, it is misleading to say that Possibilism grounds existence only in possible representations. Possibilism does not completely sever actual existence from actual sensible data that are given to representing subjects when existing things affect their senses: as we saw in section 2, according to Possibilism the actual existence of spatiotemporal things depends on whether these things belong to a possible course of experience that is connected with something actually given. Moreover, this connection is not defined through a bare notion of merely possible representations that might or might not follow or precede actually given representations, but through objective laws of connectivity that dynamically, necessarily relate actually given things to all other things (incorporated and interacting within the same world) that could be given to us through our senses.

My estimation of how much philosophical appeal we can ascribe to the two opposing views is opposite to Jauernig’s: I shall now argue that Possibilism yields a more attractive, less counter-intuitive version of idealism than Actualism.

As I suggested earlier, what accounts for the intuitive plausibility of Possibilism (and what likely explains Kant’s constant use of Possibilist formulations when he clarifies the implications of his position vis-à-vis empirical, Berkeleyan idealism) is that it gives Kant’s transcendental idealism a fairly robust realist flavour. It does so by blocking the unhappy idealist consequence that the actual existence (and constitution) of appearances or the actual occurrence of events is hostage to the contingent, arbitrary limitations and imperfections that affect the actual state and progress of human experience. To stress, in (empirical) realist fashion, that things should not depend upon the contingencies that affect our actual cognitive progress is compatible with holding, in (transcendental) idealist fashion, that spatiotemporal things depend upon our cognitive capacities. As I see it, the core idea behind Possibilist-idealism is that it makes spatiotemporal appearances dependent on the cognitive powers of a certain type of finite cognizing subjects, namely, upon what these subjects can (in principle) grasp as belonging to a series of appearances based on their cognition of some actually given, existing members of that series. ²⁶ Accordingly, what I see as the central idea behind Kant’s idealism is that spatiotemporal appearances depend on, or ‘conform to’ (Bxvii), the necessary formal components of our cognitive faculties that jointly define our object-representations as a distinctive type of cognition: namely, as a discursive cognition that conceptually unifies given data received through spatiotemporal forms of sensibility into a whole of experience. This is an essentially finite type of cognition whose incomplete nature shapes the ontological character of the world it cognizes: qua conforming to our inherently piecemeal, searching, never fully complete mode of cognition, the spatiotemporal world we cognize does not exist as a given complete totality. But this essential element of imperfection and limitation that shapes both our cognitive faculties and the world of our cognition must be distinguished from the merely contingent, arbitrary imperfections and limitations which happen to affect human cognition. These include: the coarse empirical constitution of our sense-organs which prevents us from directly perceiving things like magnetic matter; the contingent interests which lead us to actually follow the chain of appearances backwards or forwards in one specific direction rather than others; our cognitive mistakes which halt the progress of our empirical knowledge at certain points; and whatever empirical events may (perhaps rather soon) annihilate us as a species of embodied cognizers and thus put a final end to the actual course of spatiotemporal experience.

Somewhat more concretely, I regard the following contrast as a crucial element of Kant’s view. On the one hand, certain general facts about the things that populate the empirical world of our cognition depend upon the general character of our cognitive capacities: these things are in space and time, exhibit extensive and intensive magnitudes, involve substances that persist over time, change their properties in accordance with natural causal laws and mutually interact as a community of natural objects. If it were not for the a priori features of our sensible and intellectual faculties of representation, then the things we cognize would not be in space and time, would not have substantial permanence and would not be governed by universal laws of nature (etc.). Moreover, since our essentially incomplete mode of cognition grasps the world only by progressing or regressing further and further in the series of actually given appearances, via an infinite or indefinite process of synthesizing given data, the empirical world of our cognition does not exist as a complete, determinate whole. But on the other hand, many specific empirical facts obtain whether or not we (collectively) happen to register them in the actual progress or regress of our experience. The event that a 2 foot branch fell off an oak tree at a certain location in the Carpathian forest at 9:34 a.m. on 22 January 1634 occurred even though this event is never actually perceived. (Perhaps a deserting soldier would have perceived this event if he had not been distracted by a trumpet alarm just a split second earlier.) A complicated physical law governs the behaviour of particles in a way that determinately affects the formation of certain cell structures under certain conditions, even if this law is never actually discovered. ²⁷ Planet Z exists far away from earth at a specific location in the galaxy even if the human species of cognizers goes extinct before its course of experience progresses to Z.

Can Actualists account for these commonsense realist truisms? It is hard to see how this could work on Van Cleve’s explicitly Berkeleyan take on Kant’s view, which seems to imply that appearances are merely virtual objects that depend on being currently perceived (since their very esse consists in their percipi). For Berkeley, things do not depend on the arbitrary limitations of human perceptions because the actual perceptions determining existence are ideas in God’s mind, which comprehensively perceives all there is without limitations. But this response is unavailable to Kant for various reasons. For one, Kant cannot appeal to God as the ontological determinant of spatiotemporal existence since he argues that the existence of God itself cannot be known or proven on theoretical grounds. Moreover, divine cognitions qua intellectual intuitions would pertain only to the supersensible or noumenal world (which God would grasp as a completely determinate whole) rather than to the sensible world of our spatiotemporal experience. ²⁸

Perhaps Jauernig’s non-Berkeleyan version of Actualism has a better chance of accommodating the above commonsense assumptions. Let us consider here what Jauernig says about the existence of the Aachen Cathedral. She stresses that her Actualist conception of Kant’s idealism does not entail that appearances ‘“pop” in and out of the empirical world depending on whether anybody currently represents them’. For example, ‘if the right kind of representations represent Aachen Cathedral as persisting in time … then it does persist in time in the empirical world even if there are periods during which nobody represents it’ (Jauernig 2021: 36–7). For Jauernig, the ‘right kind of representations’ are the ones which belong to the series of ew-accounts that human cognizers develop throughout the course of history (see section 1). This class of representations is obviously not limited to currently occurrent perceptions: it includes representations that have been incorporated, via properly rule-governed syntheses of given perceptions, into the unified course of a single, collective and public human experience. ²⁹

This Actualist interpretation is clearly more congenial to commonsense realist intuitions than Van Cleve’s Berkeleyan reading. Nonetheless, Jauernig’s Actualist view has implications that seem like a rather striking offence to common sense. Her example of Aachen Cathedral conceals these implications because this object is a human-made artefact whose creation guarantees that it is actually perceived and incorporated into the one course of human experience. But let us focus on some particular feature of this building: for instance, suppose that there is a tiny crack in one of the cathedral’s side walls. Suppose, further, that this crack escapes the notice of the entire species of human cognizers: it is never actually perceived and incorporated into the one course of human experience. On the face of it, both of these suppositions make good sense. But on Jauernig’s Actualist view, they jointly yield an ontological impossibility because on this view being actually incorporated into the one course of human experience is a necessary condition for spatiotemporal existence. To make vivid why this implies a counter-intuitive offence to common sense, let us suppose that a particular cathedral conservator would have noticed this crack if only she had not been distracted by an itch on her eyebrow which made her look in a slightly different direction. Intuitively, whether or not the cathedral has some specific crack should not hinge on arbitrary circumstances like these which affect what happens to enter into the actual course of human experience. But this dependency does obtain on Jauernig’s Actualist view. By contrast, Possibilists can say that that the existence of the crack is secured by the fact that the crack would be encountered in the possible progress or regress of human experience from some given data in accordance with the proper rules of connection.

One might reply that what sounds intuitive or commonsensical is neither here nor there when it comes to interpreting the contours of Kant’s idealism, especially since commonsense intuitions embody realist prejudices that are not the proper touchstone for appreciating Kant’s idealist ontology. However, this reply overlooks the dialectical situation: we are considering Jauernig’s claim that Possibilism fails (in part) because it renders Kant’s idealism especially odd or ‘eyebrow-raising’. The above considerations show that Actualism, even in its (arguably) most realist or non-Berkeleyan version, is more prone to elicit the incredulous stare. Moreover, on a strictly interpretative approach there is also a good reason for limiting the extent to which Kant’s idealism yields striking offences to realist common sense: although Kant recognizes that the common understanding is firmly attached to the erroneous idea that the objects of human cognition are things in themselves (see e.g. GMS, 4: 451–2), he also insists that rejecting this idea does not necessitate any serious change for the mindset that one adopts from the empirical standpoint of common human (theoretical) cognition or experience. Since the non-sensible thing in itself as the correlate of sensible appearances ‘is … never asked for in experience’ (A30/B45), one can rightly regard a public empirical object like a rose ‘in an empirical sense as a thing in itself’ (A29–30/B45); ‘with respect to all merely possible experience, everything remains just as if I had never undertaken’ this idealist ‘departure from the common opinion’ (Prol, 4: 291) that spatial objects are things in themselves. It is hard to see how these claims could be sustained in light of the Actualist implication that whatever for whatever contingent reasons escapes our actual collective notice fails to exist (or occur).

Actualism has a further (though related) problematic implication. What if the actual course of human experience involves what we would ordinarily regard as mistaken representations of the empirical world? This problem is especially acute on Van Cleve’s Berkeleyan Actualism, which measures spatiotemporal being (or virtual being) in terms of actual sensible data. Note how Berkeley struggles with this issue: he sometimes seems to want to bite the bullet and concede that actual sensible ideas always determine actual being, even when these ideas involve what realists would regard as perceptual illusions; for example, when one sees an oar as crooked in the water. ³⁰ But the arguably best Berkeleyan route here is to measure actual being in terms of God’s perceptual ideas which are always accurate. For the abovementioned reasons, Kant cannot take this route. ³¹

Does Jauernig’s non-Berkeleyan version of Actualism fare better in this regard? Recall that, on Jauernig’s view, the actual experience that determines spatiotemporal existence is the ‘series of ew-accounts that the community of human minds develops throughout the course of history’ (Jauernig 2021: 96). This includes ‘a collection of … judgments of experience made so far concerning the determinations of … appearances, including their relations to each other’ (pp. 94–5). Now suppose that one of these judgements is mistaken: some empirical world-accounts which human minds have developed throughout the course of history contain a judgement about, say, the chemical properties of some type of matter M2 which is reasonable in light of the currently available evidence and the present state of science but nevertheless false. Now suppose further that this judgement never gets corrected in the further course of human history. (Perhaps humans go extinct too early, or their cognitive interests shift to empirical investigations that are not concerned with M2, etc.) If actual experience determines spatiotemporal being (in the way Jauernig suggests), and if nothing in the actual series of ew-accounts supersedes the relevant judgement about M2, then it follows that this judgement must be correct. So, even in Jauernig’s sophisticated version of Actualism, Berkeley’s crooked-oar problem lingers since Actualism implies that an actual judgement of experience simply cannot be mistaken if it never happens to get corrected during the contingent actual course of collective human experience. By contrast, Possibilists can say that the relevant judgement is false since it would be corrected in a possible course of human experience that proceeds, under the direction of proper dynamical and other epistemic rules, from actual judgements of experience to further judgements which correct the former ones on the basis of new evidential inputs and corresponding better scientific theories.

Actualists might respond here as follows: an actual judgement of experience indeed cannot be mistaken, not because human cognizers cannot make mistakes about what actually exists but because experience is by (Kant’s) definition always factually correct. Since on sophisticated versions of Actualism it is actual experience (rather than mere perception) which determines actual spatiotemporal existence, Actualists might hold that human judgements that never get corrected in the progress of human cognition do not necessarily determine actual existence because those judgements might be insufficiently synthesized and thereby fail to qualify as factually correct judgements of experience (e.g. they might be mere judgements of perception that lack universal validity). However, this response faces two problems.

First, this response aggravates rather than mitigates the worry that on Actualist views spatiotemporal existence becomes hostage to the contingent, arbitrary limitation of human representations. If the only human representations that determine spatiotemporal existence are judgements of experience that are both actual and factually accurate, then the range of representations that are suitable for determining actual being, and thereby the domain of actual being, shrinks considerably. We can imagine an actual course of collective human representations that (for whatever contingent reasons) involves very few objectively correct, incorrigible representations or judgements of experience. According to Actualism, human cognizers in this scenario would be dealing with a very sparsely populated natural world.

Second, this response requires an account of Kant’s notion of experience that is at least controversial. While for Kant experience and judgements of experience clearly have (unlike mere perceptions or judgements of perception) objective and universal validity, it is far from clear whether objective validity must always entail (empirical) truth or incorrigible correctness. ³² On an alternative view, judgements of experience are universally and objectively valid insofar as they can lay claim to objective correctness, which requires that they be based on publicly accessible reasons for judging that can be presented to other judging subjects as topics for shared rational communication. This seems like a plausible reading of Kant’s definition that if I make ‘a judgment of experience, I … require that this connection [of sensations] be subject to a condition that makes it universally valid … [and] that I, at every time, and also everyone else, would necessarily have to conjoin the same perceptions under the same circumstances’ (Prol, 4: 299). Kant’s formulations here and elsewhere refer to the rule-governed synthesis that elevates public judgements of experience above mere representations of one’s private mental states. Based on their (at least implicit) consciousness of this synthesis, judging subjects presuppose that their judgements represent empirical facts (i.e. state that something objectively is the case: cf. B141–2). This gives their judgements objective validity and renders them candidates for empirical truth. Accordingly, the judging subject has, as Henry Allison puts it, the ‘epistemic right’ to demand agreement (or good reasons for disagreement) from other subjects (Allison 2015: 442). But a judgement of experience may be mistaken despite having universal validity and despite legitimately expressing an epistemic demand for intersubjective agreement. For example, consider the judgement that a stone which was previously cold is now warm due to the causal impact of sunrays. This judgement has objective validity since it conforms to the epistemic rule that in order to represent an objective change of states one must presuppose some causal condition that renders the temporal sequence of states necessary (see e.g. A198/B243–4), and since it involves a publicly verifiable hypothesis about what this causal condition is. But the judgement might still be mistaken if (e.g.) someone had lighted a fire next to the stone a short while ago.

I do not want to make it seem here as if an account of (judgements of) experience as equivalent to empirical truth or factual correctness is obviously at odds with Kant’s notion of experience. But it is worth noting that an Actualist view which trades on the equation of experience with empirical truth (in order to salvage the prima facie implausible idea that actual judgements of experience which never get corrected cannot be mistaken) is not just philosophically problematic but also requires a controversial conception of Kantian experience.

4. Conclusion

In this article, I have examined whether we should understand Kant’s idealist view that spatiotemporal objects are mind-dependent as asserting a dependence of these objects on actual or possible human experience. I have argued that the Possibilist interpretation is preferable over the Actualist alternative in two ways.

First, I have argued that there is rather little explicit textual support for Actualism; the most striking Actualist passages belong to a section (the A-edition Fourth Paralogism) that Kant rewrote in the second edition Critique, (likely) precisely because he was unhappy with its Berkeleyan-Actualist implications. By contrast, in some key passages Kant consistently uses Possibilist formulations and explications.

Second, although Actualists raise some important worries about Possibilism, I have argued that these worries fail to show that Possibilism is incoherent with Kant’s core idealist tenets such as the claim that the empirical world does not exist as a complete given totality. Actualists and Possibilists typically claim that, on their respective views, Kant’s idealism seems like a more appealing (or less unappealing) philosophical position. It is hard to assess this debate, in part because it is not clear what criteria we should use for determining the philosophical appeal of Kant’s view and to what extent we should defer to commonsense realist intuitions in making this determination. But it seems fair to point out that, even on sophisticated Actualist readings, Kant’s idealism has a strikingly counter-intuitive implication: the existence of spatiotemporal objects or properties and the occurrence of events is hostage to the arbitrary, contingent ways in which the actual course of human representations happens to unfold or stop unfolding. By contrast, on a Possibilist reading Kant’s idealism allows for a fairly robust (empirical) realism that secures the independence of spatiotemporal objects or events from the contingent, fortuitous (as opposed to the necessary, essential) limitations of human cognition.

In sum, my overall argument implies that we should construe Kantian mind-dependence as the view that spatiotemporal appearances depend for their actuality on being objects of a possible human experience: namely, on what human cognizers would experience in a possible progress or regress that proceeds from actual states of experience to further, non-actual perceptions in accordance with the right (epistemic and dynamical) rules. As I mentioned earlier, this conclusion concerns only one important aspect of Kant’s idealism: his view that the spatiotemporal world depends on the human mind raises many further questions that are deliberately left open by my argument in this article. My Possibilist take on Kant’s idealism should be compatible with a wide range of further views about appearances, things in themselves and their relation(s).