In a recent paper Desmond Hogan (Reference Hogan2009b) presents an innovative and thought-provoking solution to one of the oldest and most challenging problems in Kant's philosophy, the problem of noumenal affection. First, I will briefly explain what the problem is. Then I will explore Hogan's proposed solution, and explain why it does not succeed.

Abundant textual evidence supports attributing to Kant these two claims:

On the face of it, these claims are in tension. How can Kant claim simultaneously that we are ignorant of things in themselves, and that they causally affect us? Hogan calls this the ‘Consistency’ problem.

As James Van Cleve has pointed out,Footnote ² there is no logical inconsistency in claiming that p and I cannot know whether p. Thus, noumenal affection is not logically inconsistent with noumenal ignorance, if noumenal affection expresses merely a personal conviction on Kant's part. However, Kant makes another problematic claim about noumenal affection:

According to this claim, the reason we experience objects as having the material features we do is the manner in which we are affected by things in themselves.Footnote ³ Given that Kant regards noumenal affection as indispensable for experience, it seems implausible that it merely expresses a personal conviction on his part. What justifies Kant in claiming that noumenal affection is indispensable to experience in this way? Hogan calls this the ‘Indispensability’ problem.

Hogan reviews several other strategies for resolving this conflict and rejects them. In their place he offers the ‘Indispensability’ argument. This is a logically valid argument whose conclusion is noumenal affection and which shows that noumenal affection is a necessary condition on some empirical knowledge (knowledge justified by experience) possessed by rational human agents. He solves the consistency problem by arguing that noumenal ignorance expresses Kant's reasons for holding a key premise in the indispensability argument. Far from being inconsistent with noumenal affection, noumenal ignorance in fact constitutes a key reason Kant accepted that doctrine. Similarly, noumenal ignorance is not inconsistent with the indispensability thesis, because it supports a key premise in the argument that noumenal affection is indispensable for experience.

Hogan's indispensability argument for noumenal affection is:

1. (1) The actions of rational human agents are the product of a libertarian power incompatible with necessitation through determining causes (Freedom), and

2. (2) Rational human agents have some empirical knowledge of each others’ actions and the empirical representations through which they have this knowledge reliably track those actions (Knowledge), and

3. (3) The reliable tracking presupposed by Knowledge must rest on (a) causal affection of agents by other agents’ actions (Interaction), or (b) pre-established agreement of representational states of causally isolated subjects (Preestablished Harmony), or (c) suitable higher interventions on the occasion of relevant actions (Occasionalism), or (d) cosmic coincidence.

4. (4) The tracking required by Knowledge cannot be held to rest on (b) because the acts in question lack determining causes; nor on (c) assuming occasionalism is rejected; nor on (d), which doesn't justify the reliability claim.

   Therefore: The tracking presupposed by Knowledge must be held to rest on causal affection of agents by other agent's actions (Interaction).Footnote ⁴

Intuitively, the idea behind this argument is that if subjects have empirical knowledge of one another's actions, their empirical representations of one another's actions must reliably track those actions, and there must be an explanation of the source of that reliable tracking. One possible such explanation would be occasionalism: God reliably causes representations that accurately represent the actions of other agents. Another would be pre-established harmony: God creates the souls and bodies of agents so they develop according to inner, deterministic laws which produce a reliable match between the actions of agents and the representations of those actions in other agents’ minds, even though no causal interaction occurs. The simplest explanation, of course, would be that agents causally influence one another and this influence grounds the reliable match between agents’ actions and other agent's representations of those actions. We can reject occasionalism, because if occasionalism is true, in some obvious sense the agents in question lack the causal power to act. And we can reject pre-established harmony because agents have libertarian freedom, and thus their free acts cannot be the unfolding of a deterministic law. Since there must be an explanation of the reliable tracking – and this excludes the possibility that it is simply a coincidence – we can conclude that agents’ representations reliably track one another's actions because those actions causally affect those representations.

Two questions about this argument need to be distinguished: (1) did Kant accept its premises? (2) does it express Kant's reasons for accepting the conclusion? Hogan convincingly argues that the pre-Critical Kant did accept the premises of this argument once he converted to a libertarian theory of freedom in the 1760s. Hogan presents no direct evidence, though, that Kant ever formulated this argument, or that it expresses his reasons for adopting noumenal affection. In fact, before Kant has made the Critical distinction between phenomena and noumena, it is unclear whether the conclusion of this argument is equivalent to noumenal affection. Without that distinction, this is an argument for the less surprising conclusion that we are causally affected by other agents. Nonetheless, Hogan shows that considerations like the indispensability argument were heavily discussed in German philosophy of Kant's day, and that Kant would have been aware of them.

The main problem with Hogan's interpretation arises when he applies the indispensability argument to Kant's Critical philosophy. As Hogan himself acknowledges, this is an extension of his earlier attribution of the argument to the pre-Critical (late 1760s) Kant, and, as such, requires additional support. The problem is that in Kant's Critical philosophy the crucial Knowledge and Freedom premises are ambiguous. Depending upon how we disambiguate the term ‘action’, the argument is either invalid, or remains valid, but has an anodyne conclusion.

Recall the original Knowledge premise:

On one way of reading this premise, ‘action’ refers to individual spatiotemporally localized events in the empirical world, e.g. bodily movements. The actions of other agents of which I have empirical knowledge are actions that occur at determinate times, in determinate locations and last for determinate periods of time. They might be called ‘empirical actions’ because they are exercises of the empirical causality of agents. If we interpret the Knowledge premise this way, it should be reformulated:

But once this premise is reformulated in this fashion, it becomes clear that if the argument is to remain valid its conclusion must be:

This is not the controversial doctrine of noumenal affection; it is the relatively anodyne doctrine of empirical affection! It is merely the claim that agents’ knowledge of other agents’ empirical actions (e.g. their bodily movements) depends upon those actions causally affecting the sense organs of the agents in question (the knowers). If we take ‘action’ to mean ‘empirical action’ in the Knowledge premise, the argument becomes irrelevant to noumenal affection.

Surely something has gone wrong here since Kant holds that empirical actions of agents have fully determining empirical causes. Let us go back to the original Freedom premise:

This premise is ambiguous because the term ‘action’ is ambiguous. If ‘action’ here refers to ‘empirical action’ the premise remains true (for Kant) but irrelevant to the question of noumenal affection. It is true because, Kant holds, all of our empirical actions are caused by our ‘intelligible character’, a libertarian power of agency we possess as things in themselves (‘noumena’) outside of space and time.Footnote ⁵ Our (exercise of our) intelligible character causes our empirical character, a disposition we possess as empirical beings to act according to certain laws.Footnote ⁶ Since intelligible character causes empirical character,Footnote ⁷ and empirical character produces individual empirical actions, intelligible character causes empirical actions.Footnote ⁸ Although these empirical actions are produced by this ‘noumenal causality’ they also have fully determining empirical causes and they are causally efficacious members of the empirical world. So no problem arises about how we can have empirically based beliefs that reliably track them: those empirical actions affect our sense organs.

However, if ‘action’ refers to the exercise of this intelligible (or ‘noumenal’) causality itself, the premise remains acceptable to Kant but again does not help us establish the truth of noumenal affection. I will refer to the exercise of intelligible causality as a ‘noumenal action’. For if we reformulate the Freedom premise as:

the argument only remains valid if we reformulate the Knowledge premise of the argument to read:

But this is a premise Kant does not accept. Kant adamantly and repeatedly claims that I have no empirical knowledge of the specific content of what I am calling my ‘noumenal actions’, the exercise of my noumenal (or intelligible) causality. Specifically, I have no empirical knowledge of whether I exercise my noumenal causality in a way that is morally good, or morally evil.Footnote ⁹ If I cannot have such knowledge, surely no other finite agent has such knowledge. They cannot have it by testimony, as Hogan suggests (2009b: 519), because other agents have no empirical evidence that my beliefs about my noumenal character reliably track the truth. And they cannot have it through empirical sense perception because my noumenal character is not a possible object of sense perception.

Hogan presents no textual evidence that Kant allows that I can have knowledge (either theoretical or practical) of the specific content of my noumenal character, or that of any other agent. The closest he comes is presenting textual evidence (2009b: n. 69) that we possess practical knowledge (e.g. of our noumenal freedom) that is unavailable to us through theoretical means; in none of the texts Hogan cites does Kant claim we can have knowledge of the specific content of the free exercise of our noumenal wills or that of any other agent. But this is the knowledge relevant to the argument because it is the knowledge for which premise (3) might plausibly be thought to be true; I do not need to be causally affected by other agents to have knowledge of my own noumenal freedom, and Hogan has given no reason to think otherwise.Footnote ¹⁰

The indispensability argument, as reconstructed by Hogan, fails because the term ‘action’ is ambiguous. If it refers to empirical actions, empirical affection explains our empirical knowledge, and the argument does not show why we need to posit noumenal affection. If it refers to ‘noumenal’ actions, there is no empirical knowledge to be explained in the first place.

However, the failure of the indispensability argument ramifies through the rest of Hogan's project. If the indispensability argument fails, so fails Hogan's attempt to solve the indispensability problem: why does Kant regard noumenal affection as indispensable for experience?Footnote ¹¹ Secondly, even if noumenal ignorance expresses the motivation for the Freedom premise in the indispensability argument, this does nothing to show that noumenal ignorance is part of the motivation for noumenal affection, because the indispensability argument fails. Finally, Hogan has done nothing to show that noumenal affection is consistent with noumenal ignorance. If the content of noumenal ignorance were exhausted by the Freedom premise of the indispensability argument, then there would be no prima facie conflict between them. Hogan has not argued for this, but only for the weaker claim that noumenal ignorance ‘incorporates’ the claim that some features of reality (most importantly, free acts of finite rational agents) lack determining grounds (see Hogan Reference Hogan2009a). But noumenal ignorance seems to extend much more widely; it seems to be the claim that we cannot know anything positive and substantive about things in themselves on theoretical grounds.Footnote ¹² (I am not assuming this is true, but merely pointing out that Hogan has not ruled it out.) So there is reason to think that noumenal ignorance is not exhausted by the Freedom premise and, hence, that Hogan has done nothing to resolve the mystery of why noumenal ignorance is compatible with noumenal affection. Consequently, both the compatibility and the indispensability problem remain unsolved.