Lewis on Counterfactuals, Similarity, and Morgenbesser's Coin

There is a mistake on page 49 of Lewis's "Counterfactual dependence and time's arrow" (1979). Since the mistake seems to be repeated all the time, it might be worth pointing it out.

Page 49 is where Lewis lists similarity standards for his analysis of counterfactuals. The analysis, recall, says that "if A were the case, then C" is true iff the closest A-worlds are C-worlds (or, more precisely, iff either there are no A-worlds or some A&C-worlds are closer to the actual world than any A&~C world). Closeness is a matter of similarity, and Lewis indicates what the relevant respects of similarity might be for certain ordinary counterfactuals in section 3.3 of his 1973 book, and again in the 1979 article on counterfactual dependence. Roughly, the closest A-worlds are those that perfectly match the actual world across as much of spacetime as possible without diverse and widespread violations of the actual laws. This won't do for indeterministic worlds, where generally no laws need to be violated at all in order to ensure perfect match of futures even after earlier divergence. So Lewis restricts his standards to deterministic worlds, returning to the indeterministic case in the 1986 postscript to the 1979 paper.

Now consider the following situation, due to Sydney Morgenbesser.

You bet that a certain fair coin toss will result in tails; but the coin lands heads. If you had bet on heads, you would have won.

This is widely supposed to be a problem for Lewis's 1973/1979 account: by this account, the closest antecedent worlds are worlds where a small miracle makes you bet on heads instead of tails. Without a big and widespread miracle, all such worlds differ from the actual world throughout the future light cone of the betting event. Introducing further small miracles would therefore needlessly detract from similarity. Hence all those worlds evolve by the actual laws of nature. Since the coin is indeterministic, laws and history leave it open whether the coin lands heads or tails. So it is not true that at all the closest worlds where you bet on heads, you win. The intuitively true counterfactuals comes out false.

But this is all wrong. For one, the 1973/1979 account is silent on indeterministic cases. And even if we ignore the restriction to determinacy and apply Lewis's standards, we can't assume that the coin toss is the only indeterministic event in the history of the world. If Morgenbesser's situation takes place in a thoroughly indeterministic universe, the closest antecedent worlds (by the 1979 standards) are miracle-free worlds where an entirely lawful, indeterministic event makes you bet on heads, and where other indeterministic events make the future perfectly match the actual future. In all such worlds, the coin lands heads. Morgenbesser's counterfactual comes out true.

One of those who mistakenly took Morgenbesser's case to be relevant to Lewis's 1973/1979 account was Lewis himself. On page 49 of the 1979 paper, he adds a further condition to the similarity standards given above. The added condition (numbered "4") says that "approximate overall similarity" is of "little or no importance" when comparing worlds. Lewis notes that "it is a good question whether approximate similarities of particular fact should have little weight or none. Different cases come out differently, and I would like to know why. Tichy and Jackson give cases which appear to come out right [...] only if approximate similarities count for nothing; but Morgenbesser has given a case [...] which appears to go the other way."

Apparently Lewis thought that if approximate similarities get some weight in Morgenbesser's example, then worlds where the coin lands heads come out closer to the actual world than worlds where it lands tails. But the patch is not only unnecessary, it also doesn't work. There are many respects in which the worlds with heads are less similar to the original world, e.g. with respect to your losing the bet. Depending on the story (think: high stakes), this dissimality can easily outweigh any approximate similarity gained by keeping fixed the outcome of the coin toss.

The upshot is twofold.

First, Lewis never had any good reason for assigning positive weight to approximate overall similarity in his standards for deterministic worlds. The only reason he gives is the irrelevant Morgenbesser case. As he points out, the addition yields wrong results in various other cases, which, unlike Morgenbesser's, do not involve indeterminacy. We should therefore just ignore the mistakenly added condition 4. Sadly, many philosophers still think that approximate similarity is somehow very important for Lewis's analysis, and that examples of the Tichy-Jackson type pose a big problem for his account.

Second, we should stop repeating Lewis's mistake to think that Morgenbesser-type cases are relevant to his 1973/1979 account. Such cases are interesting, but they should be discussed at the right place: when we're dealing with similarity standards for indeterministic worlds. In this context, the 1973/1979 standards are not meant to apply.

Comments

# on 18 June 2009, 00:33

I'd thought the Morgenbesser intuitions could be got going without the presumption that the laws were indeterministic. Presumably there's some content to "fair coin" even without fundamentally chanciness (if we wanted to get fancy, maybe we could explicate this in terms of stat mechanical probabilities).

Suppose I'm offered the chance to bet before the coin lands (or is even tossed, if you like), and the coin actually lands heads. The intuition is just as good by my lights as in the fundamentally chancy case: If you'd've bet heads, you'd've won.

Now consider the nearest bet-worlds. Typically for Lewis these'll have perfectly match up to a short time before pre-bet (but before the coin lands). Then you get a small violation of the laws, leading to my betting. If we weight approximate similarity zero, is there any reason for thinking that all the most similar such worlds would be ones where the coin lands heads? Why can't the differences mean that the state of the world (post-miracle) leads (deterministically) to it landing tails? I think it's plausible enough that without weighting approximate similarity to some extent, we'll have some nearest bet-world where the coin lands tails, falsifying "if were bet, then heads", for Lewis). Am I missing something here?

Even given all this, I'm not terribly sympathetic to clause 4. I'm inclined to agree I can get myself into a state where the Morgenbesser counterfactual seems at least not clearly true. And I'd prefer to handle these cases by appealing to contextual restrictions on the whole space of worlds, rather than trying to fine-grain similarity.

(In fairness to them, the people who do want to fine-grain it, that I know of, don't tend to want to add a blanket clause "approximate similarity counts for a little", but rather, some selective version of it: "such-and-such kind of approx similarity counts for a little, but such-and-such kind counts for nothing". The difference between relevant and irrelevant approx similarity is typically cashed out by appeal to something like preservation of causal chains. So they hope to get a single account that fits with Tichy et al as well as Morganbesser. I don't know whether you intended to suggest otherwise, but in any case, that seems to me the position that competes hereabouts.)

# on 18 June 2009, 21:34

Hi Robbie,

I don't think Morgenbesser cases work well in deterministic setups. It turns on whether your betting is causally upstream of the coin toss (or landing). If it is not, then in a deterministic world, the outcome of the coin toss must be the same as it actually is. But if it is, e.g. if the way you bet influences the exact way the coin is tossed, then Morgenbesser's intuition becomes very questionable. This is reflected in the common response to Morgenbesser cases which you mention: that we should hold fixed what is causally (or probabilistically or nomically) independent of the antecedent. This would be of no help at all if there were intuitive Morgenbesser cases where the coin toss is causally downstream of the betting.

# on 19 June 2009, 00:50

Hi Wo!

Hmmm... I don't feel that the Morgenbesser intuitions slacken off at all where the betting is some weak sense causally upstream of the coin toss/landing; e.g. when I have the chance to call the bet midway through the coin toss, standing next to it (so that my breath probably changes the density of the air through which it passes in some chaotic way).

As I earlier said, I can pretty easily get readings where the Morgenbesser conditional sounds bad to endorse. (For me, this little dialogue works: if I'd bet, there'd have been a 50/50 chance of my losing; so If I'd bet, I might not have won. I'm not then inclined to go on to endorse "if I'd bet, I'd have won"). But this works equally in the indeterministic case. If primed in a different way, the Morganbesser conditionals can sound absolutely fine to me in either deterministic or indeterministic settings.

I guess it's an empirical question, how widespread and stable the Morganbesser intuitions in relevant cases are (I'd conjecture: pretty widespread; not very stable). But maybe a charitable reading of the literature citing Morganbesser cases in connection with Lewisian similarity conditions is that the folks concerned are presupposing that people are going to still get Morganbesser intuitions in the relevant cases? That's got the advantage of not interpreting Lewis et al as thinking an obviously irrelevant case is relevant to their inquiry; at worst the accusation would be that they haven't gotten the data about intuitive reactions to counterfactual straight.

Could you say some more about why you think that coin toss/landing being causally "downstream" of the betting would mean that adding causal independence constraints wouldn't help the Morgenbesserite? Just saying that event e is causally downstream of f (so that if f hadn't occured, e would have been in some way different) doesn't mean that e is causally dependent on f, surely? I totally agree that's a lot of work to be done spelling out a decent notion of "causal independence" that'll do the job the Morgenbesserite wants it to---but I'm not seeing as yet any principled reason for thinking it can't be done.

# on 22 June 2009, 18:18

I think I'm missing your point. Take a Morgenbesser case in a deterministic world where the coin actually lands heads. To get an apparent problem for Lewis's standards, the coin has to land tails in some of the closest worlds where you bet heads instead of tails. That is, among various "small miracle" ways of changing the words you uttered into words that constitute a bet on heads, some deterministically lead to tails outcomes, while others (presumably) lead to heads outcomes. But this kind of sensitivity of the outcome to the bet strongly suggests that the outcome is not causally independent of the bet. In fact, the case looks to me much like "if the coin had been tossed with the other hand, it would still have landed heads", which is generally intuited to be false.

Compare an indeterministic Morgenbesser case. Here we can stipulate that there is no lawful connection whatsoever between how you bet and how the coin lands. These are cases where I do find it plausible to say that if you had bet on heads, you would have won. FWIW, I've encountered several talks and papers where people explicitly construe indeterministic Morgenbesser cases as problems for Lewis's 1979 account, and I don't think the indeterminacy is a redundant coincidence.

I agree that there are reasons to be cautious of these judgements, though I suspect that your reasoning via "there would have been a 50% chance of tails" might also rest on an equivocation. On the relevant reading, "if A, then there would have been a 50% chance of B" can't entail the falsehood of "if A, then not-B", as witness plugging in "there is a 50% chance of B and B" for A.

# on 23 June 2009, 13:56

Hi Wo,

Just to make sure I'm with the setup. I take it that here we're talking about whether a particular tweak to the Lewis conditions---in terms of putting a "causal independence" clause into condition 4---can vindicate a putative "Morganbesser intuition" in the betting case. (We're not discussing right now whether we should be looking to vindicate that intuition by tweaking the clauses, deal with it some other way, as I like, explain it away or deny that such intuitions exist).

Here's the basic thought. There's a really strong intuitive case that when you flip with the left hand rather than the right hand, you've removed and replaced one the causes of the coin landing. Whatever individuation of events is appropriate, it seems that right-hand-flipping is a different event from left-hand-flipping. And there's a causal chain from the respective flippings to the respective landings in each world.

The thought from Morganbesserites is that if we put something like "approximate similarity in cases with same causal chain" into clause 4, then we don't get "different hand>heads" coming out true, because of the clear differences in the causal chain.

What Morganbesserites want to combine this with is that the modified clause 4 can be appealed to in the "bet>heads" case, to make the same outcome worlds come out nearer than others. So they need to say that we can have a world where we bet, which gives the same result of the coin-flipping through the same causal chain.

One way to do this would be to have the whole process from flip to outcome duplicate the actual process. But as you point out, at that point Lewis's own criterion appear to give teh same result (I'm actually not sure about this, since I think there are reasons to precisify Lewis's criteria so only exact match across a whole temporal slice weighs in---but that's non-standard, so set that aside). So take a case where we don't have exact match of the processes---in order to accommodate the betting, various particle positions are slightly different, for example.

As far as I can see, all we get from this is that if F,P1...Pn,H is the actual process, then the process in the second case, F*,P1*,...,Pn*,H* doesn't duplicate the actual one. I don't see that how we could argue, in general, that the component events are non-identical (rather than just non-duplicate). Nor do I see any general reason to think that the causal relations among the component events shouldn't be preserved.

So I don't see any argument against the Morganbesserite's idea simply from the observation that the processes can't be duplicates---or that the betting has some kind of weak causal influence on the intrinsic character of the process.

I do think, however, that the burden is on the Morganbesserite to spell out *in detail* what the condition is, what the elements of a "causal chain" are, etc. It makes a big difference whether we're working with macro-events or micro-events as constituents, for example. I'd be inclined to push them for these details. But I can't say I see the project as dead in the water.

# on 23 June 2009, 14:09

On your last paragraph... I should have been clearer about what the passage was meant to achieve. It wasn't intended as an argument that the conditional "if A, then not B" is false. Pointing to the chance and might counterfactuals was meant as contextual priming, before we consult our intuitions about the truth of the Morganbesser conditional; not as the premises for an argument that its false.

The point is that if, as I suspect, we have very low credence for the Morganbesser conditional once such priming is given, this is something that Morganbesserites are going to have to explain away, given they think the conditional is straightforwardly true. And it's not clear to me how they're going to do this.

# on 24 June 2009, 17:55

I see, agreed.

On the other point I'm afraid we're really talking past each other. I wasn't criticising the idea to embellish Lewis's standards by a causal condition. All I was criticising is the way this move is normally introduced: by pointing out that Lewis's 1979 conditions give the wrong result in indeterministic cases like Morgenbesser's coin. Different concrete examples are used e.g. by Bennett and Edgington, but they always involve indeterminacy.

In an alleged deterministic Morgenbesser case (something I've never seen in the literature), we wouldn't just have "some kind of weak causal influence on the intrinsic character of the process" which leads from flipping to heads. Otherwise the case would be no problem for Lewis's 1979 standards. To get a problem for the 1979 standards, the outcome of the coin toss -- whether it is heads or tails -- would have to be causally sensitive to the bet, so that whatever small miracle alters the bet also alters the outcome of the coin toss in some of the nearby worlds where no further miracles happen.

# on 24 June 2009, 20:07

Hi Wo,
I was presuming we'd shifted to the deterministic case, with an alleged Morganbesser intuition about it---and then the question is whether a version of clause 4 (like the one I sketched) can save the day (contrary to what you are suggesting, I take it).

On this, I don't understand why you say: "In an alleged deterministic Morgenbesser case, we wouldn't just have "some kind of weak causal influence on the intrinsic character of the process" which leads from flipping to heads. Otherwise the case would be no problem for Lewis's 1979 standards."

The point is that the region of space-time that contains the flip-to-landing process will *not* exactly match the actual world in the scenario we mention. It's intrinsic features will differ! That's consistent with the chain of causation between macro-events being held constant. If there isn't exact intrinsic match in the region, we can't appeal to the 1979 conditions 1-3. I just don't see why you'd need causal dependence of the result on the flip to get this.

# on 25 June 2009, 13:22

Ah, no, I'm not questioning whether a causal version of clause 4 can save Lewis's 1979 account in the face of deterministic Morgenbesser cases. I claim that there aren't any deterministic Morgenbesser cases, and that therefore no patch is needed to save the 1979 account.

I try to explain again why I think there is no deterministic Morgenbesser case. Maybe I'm really confused about something here.

We would need a case where a) the world is deterministic, b) the coin lands heads, c) you bet on tails, and d) in some of the closest worlds where you bet on heads, the coin lands tails -- closeness being measured by Lewis's 1979 standards. It would then follow that by the 1979 account, "if you had bet on heads, you would have won" is not true. But e) intuitively, "if you had bet on heads, you would have won" is true. I claim that any way to make (a)-(d) true makes (e) false.

Suppose (a)-(c) and consider the closest worlds (by the 1979 standards) where you bet on heads. These are worlds where a single small miracle occurs shortly before the point where you actually bet on tails, causing you to bet on heads instead, and where from then on everything follows the normal laws of nature. So far, nothing is said about whether the chain from flipping to landing in those worlds is a duplicate of the actual chain. Now (d) says that in some of those worlds the coin lands tails. That is, a local miracle which alters the way you bet deterministically alters the outcome of the coin toss. In the relevant worlds obviously the chain that leads from the flipping to the landing won't be an intrinsic duplicate of the actual chain, nor will it differ only in tiny details of micro-physics: the outcome is different! The coin lands tails instead of heads. In other words, the laws of nature are such that the outcome of the coin toss deterministically depends on how exactly you bet. And then it seems to me that (e) is just false: it is then not at all intuitive that if you had bet otherwise, the outcome of the coin toss would have been just the same.

# on 25 June 2009, 14:17

Hi Wo,

Ok---I think I see where you're coming from now. Sorry for the misunderstanding!

There's a question of what (and whose) intuitions we're trying to save. I think that saying (once we see that Germany win the penalty shoot out against Engalnd) "dammit, if I'd only put money on Germany winning, I'd have won". I guess your idea is that once we think through the situation in detail (and see that analogues of (a-d) hold for it) that we'd be prepared to deny that counterfactual.

I was thinking that there are plenty of cases and contexts where our off-the-cuff intuitions pattern like this: e.g. we do express our regret in not having bet, by saying things like "if I'd bet, I'd have more money now". Maybe I'm just wrong about even this superficial "data"---but it's where I was starting from. Granted this, one motivation for introducing the causal-independence style modifications is just to have truth-conditions for counterfactuals whereby we can save ordinary reactions like this. That way of deploying and using Morganbesser-ish intuitions seems to me ok--not a horrendous methodological or philosophical goof.

From my point of view, it sounds like what you're pointing to is that if you give people certain priming---make (a-d) salient to them, then there'd no longer have the relevant intuitions. But cases where the microdescription of the situation (and determinacy of the laws etc) are made available are rather special contexts, and it's not clear to me that the aim of the game is to design truth-conditions in those special contexts rather than the ordinary, unprimed ones.

But I don't think we're in really serious disagreement here, though---I think mostly its a matter about what off-the-cuff reactions are out there. What I haven't seen, and would like to see from the people who take this really seriously---is some indication of what theoretical use a conditional that operates so as to make-true the Morganbesser conditionals would be. What do we gain by adding a complex clause 4 rather than just setting it to zero? I can't see anything in Lewis's constructive projects of counterfactual analysis that turns on it. And so the accusations e.g. that Morganbesser cases reveal that counterfactuals are "implicitly causal" and that this vitiates e.g. counterfactual analyses of causation, seem rather overblown to me. Minimally, the "set-to-zero" counterfactual seems a perfectly good sharpening of our use of counterfactuals, and available to do the theoretical work that Lewis wanted.

Add a comment

Please leave these fields blank (spam trap):

No HTML please.
You can edit this comment until 30 minutes after posting.