Notes on Geirsson's "Conceivability and Defeasible Modal Justification": Classic Papers in Modal Epistemology Series

I promise to stop geeking out on early debates in modal epistemology soon and get into contemporary debates, but a current paper I'm working on requires me to do some deep digging into the past. Hence this series:

Notes on Geirsson’s “Conceivability and Defeasible Modal Justification”, Phil. Studies, 2005.

Geirsson criticizes three of Van Inwagen’s four main arguments for modal skepticism from his paper, "Modal Epistemology": what he labels as The Analogical Argument, The Different Sources Argument, and The Completeness Argument. 

According to The Analogical Argument, our source of modal knowledge is analogous to our source of perceptual knowledge, in the following way: just as our source of perceptual information is a reliable judge of ordinary distances, but an unreliable one when the objects are more remote, so our source of modal information is reliable when its modal judgments are of ordinary matters, but unreliable when they are of matters remote to ordinary life.  Geirsson thinks this argument fails to take into account the fact that in both cases, the reliability of the source is relative to the person or group, and the subject matter, one is referring to.  The average person won’t be able to discern errors in a symphony, but a trained musician can.  Similarly, the average person won’t be able to discern the epistemic status of modal propositions remote from ordinary life, but a philosopher, who is trained in the relevant ways, can make such discernments.

According to the Different Sources Argument, the ordinary modal judgments of the average person are grounded in, say, rational intuition, or imagining, etc.  But the abstruse modal judgments of philosophers are grounded in a different and dubious source, viz., what their peers let them get away with asserting.  Thus, significant modal claims are suspect.  Geirsson thinks this argument begs the question that significant modal claims are the product of a different source.  In fact, we have good reason to think that they come from the same legitimate source as that of ordinary folk.  Van Inwagen also fails again to appreciate that philosophers have special training, and that, in general, specialists form reliable judgments regarding the subject matter of their specialty. , 

According to The Completeness Argument, a modal statement isn’t justified for one unless one has fleshed out all of the relevant details of the conceived scenario, including the microstructural facts.  Geirsson thinks this argument proves too much.  For it entails that even ordinary modal judgments (such as the “JFK could’ve died of natural causes” case and the “LPs could play without any staticky sound” case, that even Van Inwagen wants to accept, aren’t justified.  In addition, Van Inwagen’s argument fails to appreciate that justification admits of degrees.  Thus, modal justification isn’t an all-or-nothing matter, depending on whether or not your imagined scenario is fully fleshed out in detail.  Rather, most consistent scenarios confer at least some justification on modal statements, and the justification for these statements can increase or decrease with the presence or absence various sorts of epistemic factors.

After his critique of Van Inwagen’s modal skeptical arguments, he goes on to develop an account of conceivability.  It’s a variation on Yablo’s imaginability account, modified in two main ways: (i) it modifies his objectual imagination account to allow for propositional imagination (which amounts to imagining a consistent set of sentences), and (ii) it allows for degrees of justification, where justification can be increased or decreased by further filling out the corresponding scenario.

In the final section, he tests his account by seeing how it fares in light of Van Inwagen’s cases of supposed modal ignorance.

Below is an analytical outline of the section of the paper where he sketches and defends his positive account of conceivability:

1. First Preliminary: Clearing Objections to Conceivability as a Guide to Possibility

1.1. The mere absence of a logical contradiction isn’t sufficient. Some metaphysical impossibilities don’t entail a contradiction (e.g., “Gold has the atomic number 81”, “Hesperus is not Phosphorus”, etc.)

1.2. By contrast, there is a long tradition in philosophy of taking conceivability, broadly construed, as a guide to possibility (e.g., Descartes, Adams, Stalnaker, Swinburne, Yablo)

1.3. There have of course been many criticisms of conceivability accounts.

1.3.1. Mill: our ability or lack of ability to conceive of something has little to do with whether a thing is possible or impossible, and is largely a contingent matter, having to do with our history and habits of thought.

1.3.2. Reid: Understandability isn’t a guide to possibility since we can understand necessary falsehoods and reductios in mathematics.

1.3.3. Kripke and Putnam: we can imagine experiences that would convince us that water isn’t H20, and yet this is metaphysically impossible (given that water is H20 and the necessity of identity).

1.4. However, these criticisms aren’t sufficient to reject conceivability as a guide to possibility

1.4.1. Reply to Mill: 

1.4.1.1. The history and mental habits of a philosopher enable her to modalize much more competently than the average layperson.

1.4.1.2. Mill helpfully reminds us that competence in modalizing is a discipline-relative affair (e.g., the history and mental habits of physicists render them more competent and reliable in modalizing about physics than the philosopher)

1.4.2. Reply to Reid: 

1.4.2.1. He’s right that understandability isn’t a guide to possibility.

1.4.2.2. But that leaves open other kinds of conceivability as guides to possibility.

1.4.3. Reply to Kripke and Putnam:

1.4.3.1. Their counterexamples show that conceivability doesn’t entail possibility.

1.4.3.2. However, that’s compatible with conceivability being a defeasible guide to possibility.

1.4.3.3. In the absence of evidence to the contrary, we should take it to be such a guide.

2. Second Preliminary: Should We Accept an Epistemic or a Non-Epistemic Account of Conceivability?

2.1. Accounts of conceivability divide into two main kinds: Epistemic and non-epistemic

2.2. Epistemic accounts emphasize what resources the conceiver has available to her to think about the situation

2.2.1. Yablo: P is conceivable for S if S can objectually imagine a world she takes to verify P.

2.2.2. Van Cleve: P is conceivable for S if S intuits that P is possible.

2.3. Non-epistemic accounts don’t emphasize the resources the conceiver has available to her to think about the situation

2.3.1. Menzies:

2.3.2. Chalmers: 

2.3.2.1. The ideally conceivable is coextensive with the metaphysically possible, and in this sense, conceivability entails possibility.

2.3.2.2. Conceivability is spelled out in terms of what an ideal conceiver can coherently conceive.

2.4. Geirrson’s take and approach: 

2.4.1. It seems that our judgements about what is conceivable, and thus what we can justifiably take to be possible, are relativized to actual individual conceivers.

2.4.2. He will therefore sketch an epistemic account of conceivability

3. Toward a Positive Account of Conceivability

3.1. The big picture

3.1.1. Similar to Yablo’s in that it involves imagining a scenario

3.1.2. Improves upon it in that it:

3.1.2.1. allows modal justification to admit of degrees

3.1.2.2. gives an account of increasing and decreasing justification in terms of filling out a scenario in terms of determinacy and detail

3.1.2.3. allows both propositional and objectual imagining in proper conceiving

3.1.2.4. asserts that justification is a matter of coherence (clarified in a more recent paper as coherence in Bonjour’s sense of objective coherence – not explanatory coherence)

3.1.2.4.1. more than mere lack of contradictions

3.1.2.4.2. includes how well the parts fit together. In particular:

3.1.2.4.2.1. the extent and strength of the inferential relations among the parts of the scenario

3.1.2.4.2.2. the number of unexplained anomalies in the scenario

3.1.2.4.3. since coherence is a matter of degree, and since justification reduces to coherence, justification is a matter of degree)

3.2. What’s wrong with Yablo’s account?

3.2.1. Can’t account for degrees of justification, and thus leads to standoffs (e.g., if I claim to be able to verify a world of which naturally purple cows exist, and you do not, and if we fail to uncover a defeater on either side, we both retain the full extent of our justification in our opposed modal beliefs about the (im)possibility of purple cows). But this fails to recognize the justification admits of degrees, and this allows for the possibility of moving beyond standoffs even without finding a defeater for one of the two beliefs.)

3.2.2. The requirement of objectual imagining is too strict

3.2.2.1. We can’t pictorially imagine many modal claims (Sosa’s point)

3.2.2.1.1 Negative existentials

3.2.2.1.2. Claims about the distant past

3.2.2.1.3. Simple mathematical facts

3.2.2.1.4. Contradictions

3.2.2.1.5. Complex shapes (e.g., Descartes’s chiliagon)

3.2.2.2. Strong evidence from cognitive science indicates that while some people regularly use objectual imagining in their daily lives, others don’t seem capable of it. 

4. Geirsson’s Positive Account of Conceiving

4.1. To handle the problems with Yablo’s account, Geirsson’s account of conceivability allows for propositional imagining, objectual imagining, and combinations of both.

4.2. Objectual imagining is of Yablo's sort

4.3. Propositional imagining requires a coherent set of sentences

4.3.1.There is no minimum to size of such a set, although it’s doubtful that a single simple sentence is sufficient.

4.3.2 Assessing a sentence’s truth-value at least typically requires a set of background assumptions which must cohere with the sentence up for evaluation.

4.3.3 Other propositional imaginings are quite large. They might involve

4.3.3.1. stretches of time

4.3.3.2. large regions of space and time

4.3.3.3. lots of detail

4.4. Hybrid imagining has both propositional and objectual elements present

4.5. Generally, imaginings are indeterminate, although details can be filled in on further reflection

5. Geirrson’s Positive Account of Modal Justification

5.1. A first pass: 

5.1.1. J: If S conceives that P, then S is justified in believing that P is possible.

5.1.2. C: S conceives that P iff S understands P, and S imagines a scenario in which S is true.

5.2. Problem: the first pass account doesn’t allow a person to be justified in believing a modal claim if the claim is false.

5.2.1. The Hesperus/Phosphorus case: Suppose the ancients thought it was possible for Hesperus to be brighter than Phosphorus, since thought they weren’t identical. 

5.2.2. Then on the present account, they failed to imagine a scenario that Hesperus is brighter than Phosphorus, and thus lacked modal justification for their belief.

5.2.3. But this seems implausible: surely they were justified in such a belief.

5.3. The final account:

5.3.1. J*: If it appears to S that she conceives that P, then S is justified in believing that S is possible.

5.4. Plausibly, the background assumptions of experts in a given field are details they are privy to, and novices are not, that renders certain conceivings better than those of novices in that field. Therefore, it’s plausible to think that explicitly stating and sharing those assumptions when further fleshing out their scenario will help settle disagreements about modal claims.

5.5. Two points of special emphasis re: Geirsson’s positive account of modal justification:

5.5.1. Justification is not all-or-nothing, but admits of degrees

5.5.2. Justification is defeasible

5.6. Evidential defeaters: undermining vs. contradicting

5.6.1. Undermining defeaters: acquired evidence that destroy one’s evidence for believing that P is possible.

5.6.2. Contradicting defeaters: acquired evidence that justifies one in believing that P is impossible.

5.7. Evidential defeaters: a priori vs. empirical

5.7.1. An a priori defeater might be a discovery of a contradiction in a scenario

5.7.2. An empirical defeater might be scientific evidence that helps further flesh out a scenario

6. Illustrating Geirsson’s Account

6.1. Case 1: Seddon’s Carnivorous Rabbits

6.1.1. Stage 1: Suppose I propositionally imagine a minimal scenario in which it appears to me that P is true. That is enough for the scenario to confer at least some justification on my belief that carnivorous rabbits is possible.

6.1.2. Stage 2: However, suppose I fill out the scenario, and in decreasing its indeterminacy I realize that what I’m imagining aren’t rabbits at all, but rabbit look-alikes.

6.1.3. In that case, my new evidence constitutes an undermining defeater for my belief that carnivorous rabbits are possible.

6.2. Case 2: Hesperus could be brighter than Phosphorus?

6.2.1. Stage 1: Suppose I don’t know that Hesperus=Phosphorus, and I imagine a scenario in which it appears to me that Hesperus could be brighter than Phosphorus. Then I have some justification for the proposition

6.2.2. Stage 2: However, suppose I later acquire empirical evidence that Hesperus=Phosphorus, and I also acquire evidence from recent semantics that if Hesperus=Phosophorus, then it is necessarily true that Hesperus=Phosphorus. Then my new empirical and a priori evidence constitute a contradicting defeater for my belief that Hesperus could be brighter than Phosphorus.

6.3. Case 3: Putnam, water, and H20: Similar to case 2.

6.4. Case 4: Yablo’s Goldbach’s Conjecture (GC) Case

6.4.1. Yablo says such a case is undecidable – i.e., we can’t imagine a world that verifies it, and we can’t imagine a world that falsifies it.

6.4.2. However, this is because the scenarios he deploys involve objectual imagining, which are faulty and irrelevant.

6.4.3. Rather, a relevant scenario is one in which one propositionally imagines that it is true that there is some very large even number that is not the sum of two primes.

6.4.4. However, that doesn’t mean that we are justified in believing GC is false. For we have strong evidence that GC from testing up to astronomically high numbers that it is not disconfirmed. 

7. How Geirsson’s Account Can Explain Disagreement About Modal Claims

7.1. On the current account, disagreement about a modal claim can be explained in two main ways:

7.1.1. One interlocutor’s scenario is filled out much more extensively than the other’s and has either found a defeater or found further justification for their belief.

7.1.2. One interlocutor’s scenario is filled out differently than the other’s with (perhaps) different background assumptions they bring to it (e.g., believers and atheists)

7.1.3. Disagreement can often be resolved by:

7.1.3.1. Settling relevant factual issues (e.g., scientific evidence)

7.1.3.2. Bringing to light an a priori issue (e.g., uncovering an inconsistency or contradiction on one side)

7.1.3.3. When factual and a priori matters don’t settle a disagreement, then one might appeal to the theoretical virtues (e.g., simplicity, predictability, etc.)

8. Applying Geirsson’s Account to Van Inwagen’s Examples: Parity (Sometimes) Broken

8.1. The Naturally Purple Cow Case

8.1.1. Van Inwagen says we can’t have knowledge or justified belief one way or the other about whether naturally purple cows are possible 

8.1.2. But on Geirsson’s account, one can be justified in believing in the possibility of naturally purple cows

8.1.2.1. One can at least entertain, understand, and propositionally imagine the simple proposition that naturally purple cows are possible 

8.1.2.2. One can also imagine a scenario involving naturally purple cows

8.1.2.3. This confers at least some justification on the belief in their possibility

8.1.2.4. Furthermore, the case only requires minimal changes to the cow, unlike the carnivorous rabbits case – a change that is similar to the change in the natural color of flowers from breeders

8.1.2.5. The belief, though justified, is of course defeasible

8.2. The Anselmian Being Case

8.2.1. Van Inwagen says we can’t have knowledge or justified belief one way or the other about whether an Anselmian being is possible

8.2.2. Van Inwagen doesn’t say why in the article, but perhaps because such a being can’t be objectually imagined

8.2.3. But on the present account, one can propositionally imagine such a being, and with lots of detail filled in as well. This enables us to entertain and understand what would be involved in the truth of such a claim.

8.2.4. Furthermore, religious believers interpret the world in light of such a being’s existence, and thus imagine an extensive scenario according which such a claim is true.

8.2.5. Therefore, for religious believers at least, belief in the possibility of an Anselmian being can be justified (although, again, such a belief is defeasible).

8.3. The Gratuitous Evil Case

8.3.1. Van Inwagen says we can’t have knowledge or justified belief one way or the other about whether gratuitous unjustified evil is possible.

8.3.2. If by gratuitous evil PvI means uncaused evil, then it’s unclear whether belief in the possibility of gratuitous evil can be conceived, and thus unclear whether it can be justified.

8.3.3. But we know that by gratuitous evil PvI means evil that is not necessary for a greater good. And on that understanding, gratuitous evil is robustly conceivable in all of Geirsson’s senses, and thus justified for the conceiver.

8.3.4. Furthermore, then the conceivability of gratuitous evil can be used as a contradictory defeater for the previous claim about the possibility of Anselmian beings (though, again, this defeater can itself be defeated in principle – in this case, by a reason for why such evil is justifiable after all).

8.4. The Disembodied Soul Case

8.4.1. Van Inwagen says we can’t have knowledge or justified belief one way or the other about whether it’s possible to exist without anything material existing.

8.4.2. However, on Geirsson’s account, one can at least propositionally imagine a relevant scenario, and thereby, someone could have at least some justification for its possibility.

8.4.3. On the other hand, there seems to be lots of relevant empirical information about the mind’s dependence on the body, and this seems to constitute an undermining defeater for the original belief.