One of the virtues of Paul W. Franks book All or Nothing is that it explores the metaphysical and epistemological aspects of the Agrippan trilemma at the same time.
To escape the latter for a series X – supposing we accept the need to escape – we must identify the ground of X, which allows the series to terminate. This leads to what Franks calls the heterogeneity requirement, which Kant wanted to satisfy. According to this requirement, the ground of X – let’s call it Y – must be heterogenous to what it grounds, lest it fail to escape the “law of the series” (p. 102).
Thus for example, if X is the series of contingently existing things, then the ground of contingent existence, Y, cannot itself be a contingently existing thing. Again, the series is of contingent things, so the ground cannot be a member of this series, or else it too will demand – without being able to provide – a ground. That is what heterogeneity means in this context.
Heterogeneity also connotes difference in kind and thus the nebulous threat of dualism. In Kant’s case, Franks tells us, we end up with “two worlds under distinct legislations”, and the question becomes whether this is both sufficiently heterogenous and capable of warding off the threat of dualism. Is the search for grounds inherently (and paradoxically) dualistic? (The threat of dualism can be fleshed out a bit more by introducing the distinction between immanence and transcendence, which I take up in a moment.)
There is an argument derived by Jacobi from Spinoza which attempts to show that although Y must be heterogenous to X, Y must also be immanent to the series X, that is, it must be a member of that series. The difficulty in calibrating these two requirements – heterogeneity and immanence – can be seen as a shared concern of Spinoza and Kant. The charge Franks levels at Kant is then twofold: that (1) he thinks his solution to this difficulty is the only one available, whereas the Spinozist alternative remains viable (p. 103); and (2) his solution is problematic for reasons having to do with his blindness to (1).
To begin unpacking all of this, suppose that our protagonist – let’s call him Hume 2.0 – does not acknowledge any need to avoid the metaphysical aspect of the Agrippan trilemma, since it is driven by the PSR, which he rejects. That is, since X exists, Hume 2.0 acknowledges that there is something sufficient for the existence of X, since that much is just analytic. But he is prepared to accept that whatever is sufficient for X is not a sufficient reason in the sense of an ultimate ground or explanation of X (whether discernible to me or only to God).
Note: This may seem puzzling since it means saying that X ultimately rests on nothing. We could modify one of Spinoza’s propositions (1p16) to express this idea: all and only those things compatible with the necessary possibility of nothing are necessarily possible. Whether there is a deep problem or confusion lurking here is presently moot.
How might I frame or motivate this rejection of the Agrippan trilemma? One way to do it is to say that Hume 2.0 is a partisan of immanence, and accepts the following equation:
heterogeneity = transcendence
More accurately, he holds that the sort or degree of heterogeneity required to ground a series X requires transcendence, i.e. requires that the Y that grounds X not itself be a member of X. Now we can see the fork in the road where Kant and Spinoza part ways: for Kant, the equation of heterogeneity-sufficient-for-grounding and transcendence is basically correct, i.e. the ground of the series X must transcend X (p. 103), since that is the only way to satisfy the heterogeneity requirement, and there must be a ground of X. Spinoza, by contrast, agrees that there must be a ground of X, but holds that this ground can be immanent to the series X whilst satisfying the heterogeneity requirement. That is, being a member of X is not incompatible with terminating the series that X expresses.
Finally, Hume 2.0 agrees with Kant that grounding requires transcendence, but agrees with Spinoza that heterogeneity is compatible with immanence. Importantly, the latter is true not because there is something special about Y that exempts it from the law of X, but because there is no law of X (i.e. no demand for a ground qua sufficient reason that must be satisfied). This position adheres to an ungrounded heterogeneity. Again, since he agrees that satisfying the trilemma means satisfying the heterogeneity requirement, Hume 2.0 rejects this requirement of satisfaction, because he holds that satisfying it and the heterogeneity requirement at the same time requires transcendence, which he rejects. Different reasons for rejecting transcendence can be cooked up here: perhaps Hume 2.0 will try to appropriate Spinoza’s own arguments, or the ones Jacobi finds in Spinoza. Or perhaps he’ll appeal to the unknowability of transcendence as a reason for rejecting it.
In any case, my main goal in this post is not to explore the rationale for Humeanism 2.0 but rather to explore Franks’ contention that Kant doesn’t pay sufficient attention to the possibility that there is a Spinozist alternative to his solution to the Agrippan trilemma; and that this weakens his own response to Agrippa.
The disagreement between Spinoza and Kant centres on this question: does satisfying the heterogeneity requirement require transcendence? Franks provides a useful corrective to the belief that Kant’s response to Agrippa is more good than bad, whereas Spinoza’s is more bad than good. I do however have a couple of quibbles with his Jacobian arguments against Kant. The basic idea behind these arguments is that by making Y, the ground of X, transcendent to X, we invariably violate the PSR. This is broken down into two arguments, against temporal and modal transcendence. The former says that Y cannot preexist X, the latter that Y cannot exist without X. This excludes transcendence to just the extent that the latter requires either preexistence in time or the real distinction (in the Humean sense) of the transcendent.
My first quibble is really only a half-quibble and it concerns the second argument. Note that it posits contingency: indeed the argument is basically just that the PSR is incompatible with contingency (p. 101). In other words (here starts the quibble), if we are happy to give up contingency then, for all this argument shows, we can help ourselves to a transcendent indulgence as well. This is at best dialectically effective against Kant (who according to Franks wants to use transcendence to avoid the Agrippan trilemma, but does not want to give up contingency). At worst, Kant can respond that Y can transcend X without modally transcending X. Of course, it may be that this response slides necessarily from dualism or pluralism back to monism. I won’t try to determine if this is so now. Instead, I’ll assume that Jacobi-Franks has successfully argued for the incompatibility of modal transcendence and the PSR such that Kant must either fall back on temporal transcendence or admit that his response to Agrippa is no better – and perhaps worse – than Spinoza’s.
Given this assumption, everything comes down to the first argument, concerning temporal transcendence. The argument against temporal transcendence is this: suppose X begins at time T1, and Y is the absolute ground responsible for this beginning. Now Franks argues (p. 101-1):
Then there must have been a time prior to the absolute beginning, when the absolutely unconditioned was not engaged in such generation. But now there are only two possibilities. Either there is a sufficient reason why the absolutely unconditioned passes from inactivity to generation at the relevant time, a reason that was lacking before that time. Or there is no sufficient reason for the change… [The second case obviously violates the PSR.] In the first case, the sufficient reason cannot arise from the absolutely unconditioned itself, or else it could not have been lacking earlier. Therefore the sufficient reason must arise from elsewhere. But then what we have been calling the absolutely unconditioned is in fact conditioned…
This shows the reasoning behind the idea that positing Y as transcendent to X either leads to Y being dragged back into X or else violating the PSR. The quibble I have with this argument is that it seems to overlook the following possibility: the sufficient reason for the beginning of X does not arise at T1: rather it is true at all times that necessarily X arises at T1. Sufficient reasons are temporally indexed necessary truths, thus there is no change in the absolutely unconditioned at T1.
I say “overlooks”, but this is perhaps unfair, since Franks does have something to say about this. Specifically, he seems to think that this move commits us to modal transcendence, so that it is subject to his argument against that (p. 102):
For if talk of eternal production is to mean anything, it must mean at least that it is possible for the producer to exist without the produced.
Although convenient if correct, this is not obvious to me. (At this point I want to just record my suspicion that the concepts of immanence and transcendence are more trouble than they are worth in this context.)
If such a collapse does not follow from the move to eternal production, what other repair might we make to Franks’ argument? Perhaps we can repair the argument by asking: what is the sufficient reason of time itself? Here is the thesis I want to try and argue for: there is a series N such that necessarily any ground Y and grounded X are both immanent to – that is members of – N. If that is right then maybe it suffices to establish that the PSR is incompatible with transcendence. Hence in concrete terms, if N is time itself, and X is a subset of N (proper or improper, I don’t think it matters), then the idea is that Y must be part of N if it grounds X. The alternative is to ground time in something atemporal. But what is unacceptable about this? I can only give a very sketchy answer to this question here: if we are trying to ground temporally indexed propositions in terms of atemporal ones, then prima face the grounding would itself have to be atemporal. But then if the temporal were atemporally entailed by the atemporal, it would not be temporal. So it seems that either we violate the PSR or the absolute must be essentially temporal (as e.g. Hartshorne holds).
Supposing we accept Franks’ arguments, we must conclude that there is something problematic about Kant’s response to Agrippa. And since each of these arguments encourages us to see the absolutely unconditioned as immanent to that which it grounds, we are simultaneously given reason to believe that Spinoza’s response to Agrippa is more viable than Kant’s lack of interest in it suggests.
- Paul W. Franks, All or Nothing: Systematicity, Transcendental Arguments, and Skepticism in German Idealism, (Harvard University Press: 2005).