Nathan Wildman is an Assistant Professor at Tilburg University, and a member of the Tilburg Center for Logic, Ethics, and Philosophy of Science (TiLPs). His primary research focuses on issues in metaphysics, philosophy of language, logic, and aesthetics.
This week at The Junkyard, we’re hosting a symposium on Epistemic Uses of Imagination, a recently published volume edited by Chris Badura and Amy Kind (Routledge 2021). See here for an introduction from Chris and Amy. Commentaries will appear Tuesday through Friday.
* * *
Consider the following claim:
(A) Possibly, Aristotle is a poached egg
Notably, the modal operator here is meant to express metaphysical possibility.
Many, if not most, will feel a strong pull towards taking (A) as false. However, extreme haecceitists contend that, because the only non-trivial necessary property Aristotle has is being Aristotle, (A) is in fact true – i.e., it is metaphysically possible for Aristotle to be a poached egg.[1]
So, is Aristotle possibly a poached egg? I don’t really know. But thankfully there’s something that can help me figure it out! Modal epistemology promises to provide a story about how it is that we come to have justification for the truth/falsity of various modal claims. In this way, it looks like a way for us to finally settle the matter of Eggistotle’s (im)possibility.
There are a number of different camps within modal epistemology. But one particularly fruitful and plausible one involves imagination. The general idea here is that through imagination, we can derive evidence for modal claims. Typically, this is done via an imagination-possibility link, which says that P’s imaginability provides evidence for P’s possibility. Frequently (though not always) this is bolstered by a related unimaginable-impossibility link, according to which P’s unimaginability provides evidence for P’s impossibility.
A number of the details of this imagination-based approach need to be hammered out. And the chapters from Kung, Lam, Hanrahan, and Omoge are all broadly part of this task. To that end, I’d like to talk about them all. However, due to size constraints, I’ll here only focus on Lam’s discussion. But before doing so, I’d like to briefly say something about the other three.
First, I found the objections Kung raises to Chalmers and Jackson (2001) and Ichikawa and Jarvis (2012, 2013) convincing. However, I wasn’t sold on his positive position, which involves appealing to imagistic imagination, given that, as he admits, it isn’t clear how to extend it from ‘toy cases’ to the ‘rich, real-life cases that can do more philosophical work’ (p. 40). Second, in her excellent chapter, Hanrahan offers a imagination driven account of the epistemology of real or objective possibilities – i.e., possibilities that concern our current environment.[2] Along the way, she raises objections to Strohminger’s (2015) perception-based account, as well as an externalist variant. Finally, Omoge discusses the psychology of (metaphysical) modalizing. He is interested in understanding how imagination helps us make judgements about metaphysical modality (on the assumption that it in fact does so) – in particular, what are the undergirding psychological processes? The chapter details a naturalistically acceptable way of making sense of these processes in terms of Schank & Abelson’s (1977) notion of scripts, and is a fascinating and illuminating discussion.
§1. Lam’s etiological defeat asymmetry & its consequences
Lam’s chapter, by far the most ambitious of the four, begins by noting that, when we think about imagining or failing to imagine, there seems to be an etiological defeat asymmetry. If I can imagine (in the appropriate sense) Aristotle being a poached egg, my imagining seems to support its possibility regardless of the causal process that brought my imagining about (e.g., my taking some magic mushrooms, my having an extremely developed sense of imagination, or whatever). In contrast, the cause of my failure to imagine Aristotle being a poached egg matters. If I failed because of lingering neurological damage due to a motorcycle accident years ago, then any purported justification for the impossibility seems undercut. Thus, according to Lam (p. 48), we should accept both:
EI justification for possibilities based on imagination is not etiologically defeasible
EV justification for impossibilities based on inability to imagine is etiologically defeasible
Of course, we’d like a story that vindicates this asymmetry. With that in mind, Lam offers the following argument. Suppose that we have some evidence, via imagination, for ◇P. Assuming that we are rational and given Lewis’ Principle Principle,[3] it follows that we should therefore believe that ◇P has a non-zero chance of obtaining. Then, because ‘the fact that a state of affairs p is possible just is the fact that there is a non-zero chance that p’ (p. 50), this means we should in turn accept ◇◇P. Further, given the S4 axiom, we know that ◇◇P →◇P. Hence, ◇P is true. So, the mere evidence we had from imagination we started out with guarantees the truth of the relevant possibility claim – imaginative evidence for possibility is conclusive.
Moreover, since entailment is monotonic – i.e., no further information can block or undercut an entailment – given our initial imaginative evidence, nothing can block this derivation of ◇P. In other words, EI is not only true, but explained: the etiology of imaginative evidence for ◇P is irrelevant – once we’ve the evidence, ◇P must follow.
But consider a parallel argument: suppose that we have evidence from a failure to imagine that ¬◇P. Per standard modal logic, this is equivalent to evidence for ◻¬P – i.e, for the necessity of P’s negation. As before, given this evidence and assuming both that we are rational and Lewis’ Principle Principle, it follows that we should believe that ◻¬P has a non-zero chance. But this is ‘just the fact that [◻¬P] is possible’ (p. 55). So, we should therefore accept that ◇◻¬P. But, given the S5 axiom, we know that ◇◻¬P→◻¬P. Hence it follows that ◻¬P – in other words, P is impossible. So, if there’s any evidence for something’s being impossible, then it is conclusive. Yet this conflicts with EV, since some causal origins undermine inconceivability-based modal justifications. Lam suggests that, in light of the above, the best move is to reject the S5 axiom.
The upshot is a view according to which (i) imaginative evidence for possibility is conclusive, while similar evidence for impossibility is not, and (ii) metaphysical modality is governed by the S4, but not the S5 axiom.
§2. On errors and identities
One objection Lam considers is that sometimes, we seem to imagine (in the appropriate sense) impossibilities; this calls into question the upshot of the argument for EI. In reply, Lam adopts the Error Theory about imagination, according to which we only seem to be imagining impossibilities. Key here is that our ‘ability to identify relevant imaginings is fallible’ (p. 52); while we think that we’re imagining (say) Aristotle as a poached egg, what we are really doing is imagining watching a film of a poached egg sitting on a lectern accompanied by an ancient Greek voice over (or something along those lines).
The trouble is that, once we start questioning our ability to identify our imaginings, it is very hard to stop. We can easily and readily run similar arguments against most imaginings, undermining apparent evidence for a variety of possibility claims. In this way, we can very quickly end up being forced into a kind of hyperessentialist position, where everything is extremely modally fragile. This is one reason why we might be better off following Kung (2016: 347) in only appealing to our theory of imagination to motivate any imagining ascription/claim revisions.
A second, more substantive objection concerns the relation between chance and possibility. In discussing the Principle Principle, Lewis (1980: 265) notes that some evidence is inadmissible, in the sense that, when presented, it break the chance-credence connection. For example, my friend telling me that Partick Thistle won their last match should impact my credence for Thistle’s having won, despite the fact this testimony doesn’t change the actual chance of Thistle’s being victorious. Exactly how to discern when a piece of evidence is inadmissible is a difficult and contentious matter, but it seems clear that information about the chance of something is admissible). And when it comes to the Principle Principle, credence should track all and only admissible evidence.
This is relevant because one could claim that whatever we get from imagination is inadmissible evidence. Lam attempts to block this objection by arguing that such evidence must be admissible. This is because ‘the fact that a state of affairs p is possible just is the fact that there is a non-zero chance that p’ (p. 51). Given this apparent fact identity, it follows that evidence about possibility is evidence about chance. And since all evidence about chance is admissible, imagination evidence must be admissible after all.
The problem is that the identity claim is false. There are a number of counter-examples. Here are two. First, consider an infinite series of coin flips, the results of which are heads, tails, heads, tails, … This series is metaphysically possible, but by definition it has a chance of zero. Second, take any state of affairs featuring alien universals (i.e., universals that lack actual instances). Many of these will be metaphysically possible, though all have zero chance of occurring.
An immediate consequence is that imaginative evidence might be inadmissible after all. And a knock-on consequence is even worse: the identity claim is used in the main argument too (specifically, to go from having evidence that ◇P has a non-zero chance to having evidence for ◇◇P). So, given the failure of the identity, the main argument also looks in jeopardy.
I think that these objections are worrisome for Lam. However, I suspect that there are likely to be clever ways around them. Regardless, let me stress that I think this is an ingenious and clever chapter, and I enjoyed how it (seems to) derive something so substantive out of rather humble beginnings.
[1] For more on extreme haecceitism, see Lewis (1986: 240ff) and Mackie (2006: 160ff).
[2] In this way, Hanrahan isn’t interested in (A) but rather in: (A*) Possibly, Aristotle poaches an egg.
[3] Roughly, the Principle Principle says that one’s credences should be equivalent to (what we believe to be) the relevant objective chances. So if you believe that there’s a chance of .5 that Thistle will win the Championship, then your credence should also be at .5.