1 Bennett, Jonathan, Kant's Dialectic. Macmillan Co. of Canada, 1974. xi, 291Google Scholar pp. $6.50 (paperback).

2 Kant, Critique of Pure Reason. A297f/B354. (All translations from the Critique are my own.)

3 A239/B298.

4 A253/B309.

5 It is merely because no such concept can refer to any sensible object that empty concepts are said by Kant to be “without sense or reference (ohne Sinn und Bedeutung).” These terms should not be taken as implying that judgments in which they figure (or the sentences expressing such judgments) are “meaningless” in the positivists’ sense.

6 B xxvi.

7 B310.

8 A327/B383.

9 A338f/B396f.

10 A348.

11 A350.

12 It looks as if Bennett, after vigorously attacking the First Analogy in Kant's Analytic, has now forgotten that Kant ever wrote it. “How,” he asks, “could the notion of a substratum have shouldered its way into this part of the Critique? Kant did not accept the substratum analysis for cross-time identities generally“ (p. 106).

13 A364.

14 A363.

15 And Bennett plainly does not. “I am inclined to think that the cluster of notions conventionally associated with accountability is internally defective” (p. 204). “The ordinary notion of desert, I contend, does contain something incoherent, something that cannot be honoured by reality; and Kant's theory of freedom fails just because it so resolutely tries to rescue that doomed element in common moral thinking” (p. 210). I wish Bennett had said more about this.

16 There are, to be sure, two uncertainties here: the predictive undertainty about what I will do, and the deliberative uncertainty about what to do. Bennett is right in thinking that the second can be present only when the first is present. But he is wrong in thinking that either is required for the exercise of agency.

17 Perhaps in the back of Bennett's mind is the notion that a really “scientific“ prediction would adopt the “objective” attitude, and hence analyze me away altogether as a volitional agent. And it must be conceded possible (though not, I think, likely) that further scientific discoveries might reveal men to lack volitional agency altogether. But such a notion would not really be consistent with what Bennett is arguing here. For his contention is that acquiring the ability to make “scientific” self-predictions might deprive me of agency (that is, effect a change from a state of affairs where I am an agent to a state of affairs where I am not). But a scientific theory which analyzed me away as a volitional agent would not do this; it would rather show that I had never really been a volitional agent in the first place.

18 Let me add that while I think Bennett is wrong in denying that there is an underlying “essential problem” in the antinomies, I think he is right in holding that this “essential problem” sheds very little light on the particular antinomies themselves. The first two antinomies, it seems to me, can be made to fit the schema of the “essential problem” only be making some highly dubious assumptions. In the case of both halves of the First Antinomy, it is far from obvious that the values for R give rise to any “conditions” relation at all. In the first two antinomies, it is far from evident that every F must have some other F standing the relation R to it. (Whether this is so or not seems to be a contingent matter of fact). Even in the Third and Fourth Antinomies there seem to be difficulties; for a cause is not always a necessary condition of its effect.

19 Bennett is perplexed by Kant's use of “given” in these contexts (see pp. 141,285, 287). I think there is an ambiguity all right. “Given” can mean “given in some possible experience” or merely “existing in general.” (Surely, however, it cannot be taken to mean “actually experienced by me,” as Bennett appears to suggest on p. 141). This ambiguity plays a considerable role in Kant's resolution of the Third and Fourth Antinomies, where an “uncaused cause” is ruled out as an object of possible experience, but its possibility as a thing in itself is left open. A Leibnizian monad might also be such a non-empirical “unconditioned” in respect of the Second Antinomy, though as Bennett points out (p. 177) a monad would not really be a “part” of a composite thing.

20 A417/B444.

21 As Bennett observes (p. 116), Kant reverses these appellations in the case of the Fourth Antinomy. Here a” necessary being” is “too large” for our concept, and a world entirely composed of contingent beings is “too small” (A488f/B516f). Bennett's chapter on Infinity offers some good criticisms of Kant's general argument against the antithesis. The general argument against the thesis also strikes me as rather shaky.

22 Summa Contra Gentiles I, 28, 8.

23 A396.

24 A582f/B610f.