Derk Pereboom recognizes in Kant’s theory of Transcendental Freedom two specific goals. First, there is the ambitious endeavor to reconcile an arguably Libertarian conception of freedom “while at the same time accepting an uncompromising scientific determinism about the natural world.” (Pereboom, p. 2) And second, there is the less audacious (though perhaps equally challenging) goal of showing that any credence given to such a conception of freedom is properly grounded in practical as opposed to theoretical reasoning.
To establish his first claim, Pereboom thinks Kant must show that the conception of the acting human agent who is at once fully determined naturalistically as a phenomenon of possible sense is not logically incompatible with the conception of the very same agent who, insofar as he is a noumena, is also transcendentally free. According to Kant, to be determined as a phenomenon is to be determined entirely by physical (i.e. spatiotemporal) conditions that obtain antecedently in time according to the empirical laws of nature. Conversely, to be autonomous or free in the transcendental sense is to have a “power to act independently of the natural causes that determine our actions.” (Ibid. p. 6) Pereboom thinks that Kant is successful in establishing not only the possibility of such a harmonious account but also considers Kant’s account of Transcendental freedom to be made more credible when it is combined with modal arguments (inspired by Luis de Molina) that show how one and the same object can be determined both phenomenally by the laws of nature and noumenally by a transcendentally free will, either human or divine. Pereboom is, however, much more critical of Kant’s second goal of showing that our belief in Transcendental freedom can be grounded in purely practical reasons. It is towards these critical remarks that I would here like to develop some responses.
Even though according to Kant we cannot establish on theoretical grounds that we are transcendentally free, he wants to argue that it is nevertheless legitimate for us to believe that we are free in this sense. The grounds for legitimacy are practical – we have reasons that derive from morality in particular for believing that we are transcendentally free. (Ibid. p. 32)
There are, broadly speaking, two reasons that entitle us to believe on practical grounds that we are transcendentally free. The first is that ‘ought judgments’ (whether moral or even prudential) imply ‘can’ or the ability to do otherwise and, as such, cannot, logically speaking, refer to deterministically caused phenomena which are by definition unable to do (or be) otherwise than they in fact do (or are). Hence, if normativity in general and morality in particular are to have a proper subject matter, we must posit ourselves as transcendentally free agents in order to cognize ourselves as bound by and subject to the dictates of the moral law if not also to the dictates of rationality in general. The second reason we have for believing in transcendental freedom is that it offers us a way to ground our intuitions and judgments about moral responsibility and blameworthiness when evaluating the actions of ourselves and others as we move through the world. Stated briefly, the point is that it is only on the supposition that a person could have acted otherwise that we are willing to judge him as morally responsible for his actions. But since determinism rules out any such alternative possibilities, it is only by affixing such judgments to the agent taken as a noumenon that we retain the right to blame or praise him for his actions.
Pereboom offers objections to each of these practical justifications for transcendental freedom. Respectively, he insists that what is required for ‘ought judgments’ and moral principles to be true of us or to actually hold for us is not merely the belief that we are transcendentally free but rather “our actually being transcendentally free” (Ibid. p. 35) and that there are, similarly, alternative conceptions of moral responsibility, rightness and wrongness, that do not require that we have the ability to actually do otherwise. In other words, Pereboom thinks that we can secure a place for ought judgments and moral responsibility without grounding either of these conceptions in transcendental freedom. Further, he argues, we are not justified in punishing someone whose moral responsibility we can only posit for practical purposes, even if those purposes enshrine our entire jurisprudential system of justice and social cohesion. As a contrast, Pereboom considers the practice of inductive reasoning and argues that we are justified to continue to reason according to such laws even though Hume proved such reasoning to be valid only in accord with our psychological predilection but not with things in themselves. This is informative for it reveals to us the principle implicit in Pereboom’s idea of rational warrant.
Essentially, Pereboom implies that we are rationally justified in believing some proposition or set of propositions for which sensuous experience gives us no evidence if and only if it is psychologically impossible for us to dispense with such beliefs. (Ibid. p. 43) Pereboom thinks induction passes this test while transcendental freedom (or even merely the belief that an agent could have done otherwise) does not. Presumably then, Pereboom thinks we can make sense of normative ‘ought judgments’ without believing them to refer to agents understood as transcendentally free or as able to do otherwise. Such an interpretation judges normative and moral concepts as functional from a merely “action-guiding” perspective (Ibid. pp. 36-7) whilst being, from a metaphysical perspective, impotent.
Suppose that causal determinism is true, and that hence no agent could ever have done otherwise. Frequently, it is significantly probable that expressing a moral ‘ought’ judgment will causally influence the selection of options for action, and thus there is a good moral reason to do so – even if it turns out that because causal determinism is true the agent could not have complied with the judgment. (Ibid. p. 36)
But it seems clear that the coherence of the purely action-guiding conception of normative and moral judgments would require that the causal force with which such judgments motivate an agent to action can be understood in exclusively empirical terms. And this, I think, cannot be done. For a judgment is not an object of sensuous experience. That is to say, if practical reasons are action-guiding (i.e. action causing) in any way then there is no difference between my believing that they are valid and their actual validity. This is so because it is only through my believing that I ought to do A over not-A that the judgment ‘I ought to do A over not-A’ can move me to action.