Allison argues that an incompatibilist conception of freedom is a general requirement of rational agency, rather than being only required for moral agency, for the possibility of acting from duty. In support of this he claims that "it must be insisted that if the general argument from the presuppositions of rational agency fails to provide adequate support for an incompatibilist conception of freedom, then no subsequent appeals to specifically moral requirements or 'facts' could conceivably do the job. Basically, this is because any claim to the effect that our general capacity to act on the basis of principles, frame ends or incorporate incentives etc. can be adequately accounted for in a naturalistic, causal fashion could easily be extended to our presumed capacity to act from duty alone." (Allison: 1996, p. 125)
The view that rational agency presupposes freedom is criticised by Guyer and Ameriks. According to Allison, both of them share a certain metaphysical assumption which makes them suspect of his claim. They assume that the presupposition of freedom is appealed to in order to explain the 'phenomenon' of rational agency, namely that we are capable of rational behaviour, such as deliberating and choosing maxims. On the basis of this assumption they can then, according to Allison, easily dismiss the argument since the noumenal explanations of rational behaviour are supposedly vacuous or incoherent and since the argument neglects naturalistic compatibilist accounts of rational agency.
Allison thinks that this is a cogent criticism, but that it does not undermine his view since he does not share the underlying metaphysical assumption. "According to my reading of Kant, however, the insistence on the connection between rational agency and freedom is to be understood as a conceptual claim rather than a putative metaphysical explanation." (Allison: 1996, p. 126)
Allison's response is rather puzzling. Given that he accepts the hypothetical nature of the connection between freedom and morality, it is hard to understand why he does not accept the same for the connection between freedom and rationality. He says that the argument for the reciprocity thesis is "completely hypothetical and consequently does not involve any claims concerning the reality of either freedom or an unconditional practical law." (Allison: 1990, p. 203) The same can be said, mutatis mutandis, about the relation between freedom and rationality.
Though freedom is not required to explain the phenomenon of 'rational' agency, it is required to explain the possibility of rationality and morality. We do not know whether rational or moral agency ever takes place, but we do know that if there is to be rational or moral agency, then freedom must exist. Thus, the argument based on the connection between rationality and freedom should not be seen as an argument that starts from the phenomenon of rationality and then argues from there to the reality of freedom. Instead, it is an argument that if we are in fact rational or moral, then we must be free since freedom is a necessary presupposition for the possibility of rationality and morality.
However, since we do not know whether we are actually rational or moral we cannot complete this argument so as to draw any conclusions about whether or not freedom does exist. Instead, we can only make the conditional claim that if there is rationality and morality, then there must be freedom. The connection between freedom and rationality and morality is not only a conceptual connection but also a metaphysical connection.
Allison quotes a claim made by Kant in the Metaphysics Lectures: "If I say I think, I act, etc., then either the word 'I' is used falsely or I am free. Were I not free, I could not say: I do it, but must rather say: I feel in myself an impulse to do it, which something has incited in me. If, however, I say: I do it, this signifies a spontaneity in the transcendental sense." (ML, 28:269, quoted by Allison: 1996, p. 127) Similarly, in R4225 Kant says that the "question whether freedom is possible is perhaps identical with the question whether the human being is a true person and whether the I is possible in a being of external determinations." (17:464 [1769-1770], Guyer translation) It is important to note that Kant says that the word 'I' would be used "falsely". If I were not free, then using 'I' would not just be a conceptual mistake. It would be false to say that I do something, it would be false that I think. As Kant says in the Groundwork, without freedom, rationality like morality would be nothing but a 'chimerical idea', a 'phantom of the brain' (4:445).
Thus, we can make metaphysical claims about the necessary connections between freedom, on the one hand, and rationality and morality on the other. However, we cannot make claims about the reality of either freedom or morality. Our subjective consciousness does not provide us with the basis for asserting the reality of freedom since we are unable to distinguish between the reality of morality and rationality and the chimerical plays that could be produced by psychological mechanisms.
It clearly seems to us to be the case that we are rational, that we act out of reason or out of duty, that we are bound by hypothetical and categorical imperatives. However, it might be the case that these are merely appearances and that there is no underlying reality conforming to them. The actual connections between freedom and rationality and morality are noumenal connections and are as such inaccessible to us. We can know what the conceptual connections are and what metaphysical connections must obtain if something is to be the case, but what actually is the case is unknown to us.
, I came across a rather strange argument regarding the causal efficacy of reason. According to Allison, it is not intelligible how reasons could be causes. Reasons do not seem to be the right kind of things to possess causal efficacy. Allison claims that “construing free actions as literally caused by reason makes nonsense of Kant’s position.” (Allison: 1990, p. 49) This is because “it is clear that principles do not themselves literally have causal efficacy. ... reasons can be causes in the sense that beliefs or belief states can be used to explain human actions but not in the sense that the content of what is believed (the principle on which an agent acts) is itself a cause.” (Allison: 1990, pp. 49-50) This argument appears to be highly problematic since there seems to be an illicit move from talking about the faculty of reason to reasons considered as abstract rational principles. While Allison is right in saying that it would be strange to attribute causal efficacy to abstract principles, this does not imply that the faculty of reason cannot be causally efficacious.