The Slow Switching Slowdown Showdown

Why the Sea is Boiling Hot
Originally uploaded by Pete Mandik.

Slow switching is something that has always kind of bothered me and below I pose a problem for fans of slow switching. Slow switching is this thing that happens in a bunch of thought experiments in the philosophy of mind. For example:

By “water” you mean water. But one day you go to Twin Earth, where the stuff that seems just like water and is called “water” isn’t water at all. It is twater. By the way, you were transported to Twin Earth unawares, so your initial attempts to think and utter truths about that twater stuff are doomed to fail since you are really thinking about water. However, through the miracle of SLOW SWITCHING, after a while (days? weeks? years? Nobody knows) your attempts to think and utter truths about that twater stuff now succeed. By “water” you now mean twater and all of this was achieved without any differences except those involved in the mere passage of time. Thus, slow switching happens without you noticing: it makes no subjective impact on you or anyone else.

But here’s a little puzzle that I’d like to pose for people who think this slow switching stuff is supported by intuition. Consider two Twin Earth scenarios like the one above wherein scenario A is just like the one above and scenario B differs with respect to the following demonic intervention. In scenario B, after each second of subjective time experienced on Twin Earth B, the demon freezes all activity on Twin Earth B for 99 objective seconds. The net result is that processes that subjectively take the same amount of time on A and B, take a hundred times longer objectively to complete on B.

Here’s the puzzle. Suppose that slow switching takes a year on Twinearth A. On Twinearth A, objective and subjective years are equal. However, on Twin Earth B, a subjective year takes a hundred objective years to get through. How long will slow switching take on Twin Earth B? My bet is that people that have the slow switching intuition will want to say that it takes 100 objective years. This would show, though, that the slow switching intuition requires a subjective supervenience base. But this seems in tension with the part of the intuition that says that slow switching can take place without anyone noticing, that it is the objective facts upon which slow switching supervenes (facts about e.g. H2O and XYZ).

The tension at the heart of the slow switching intuition is that slow switching both does and doesn’t supervene on objective facts.

10 Responses to “The Slow Switching Slowdown Showdown”

  1. Jason Zarri says:

    Interesting thought experiment. I’m not a big fan of twin-earth scenarios either, but it seems to me a proponent could say something like this: Slow switching doesn’t depend just on the passage of time, it depends on one’s causal interaction with the substance in question, in this case XYZ. That’s why my utterances of “water” refer to H20 and my twin’s refer to XYZ. So if I’m suddenly transported to twin earth, my thoughts and utterances gradually come to refer to XYZ the more I causally interact with it (the more I see it, drink it, swim in it, and so on) . Assuming I have no causal interaction with XYZ during the freeze periods, which seems reasonable enough to me, then on twin earth B it will objectively take 100 times longer to have as much causal interaction with XYZ as it would on twin earth A. So wouldn’t the differerences in the duration of causal interaction be enough to provide an objective supervenience base?

  2. Pete Mandik says:

    Jason, you raise an interesting point. However, I’d wonder how such a response is suppose to go together with the typiucal externalist insistence that “water” refers to the deep structure that differentiates H2O from XYZ and not the superficial causal properties that make H2O and XYZ indistinguishable to pre-scientific observers. What sort of causal interaction is it that is going to make the B supervenience base both (1) a hundred year process and (2) not limited to the superficial causal properties of XYZ? Or, to put it another way, why is it that what water is really made of matters but how much time really goes by does not?

  3. Pete Mandik says:

    Also, I wonder how much objective time “one second” refers to in each scenario.

  4. Chase Wrenn says:

    I think I’m with Jason on this one, Pete.

    All this stuff about superficial characteristics is a red herring. Here’s a picture of how natural kind terms work: They are replicators. My token of ‘water’ right now is a copy of past ‘water’ tokens. It’s reference depends on the causal histories of the tokens it copies.

    The slow switching intuition is just that the causal histories of more recent ‘water’ tokens matter more than the causal histories of less recent ones. So, eventually, my tokens of ‘water’ on TE will come to refer to twater rather than water, because they will be more proximately connected to tokens caused in the right way by twater.

    The precise amount of time that passes is not important. What’s important is that the time that passes include enough tokenings of ‘water’ that a certain counterfactual becomes true: I would not have used ‘water’ on this occasion if it weren’t for various past ‘water’ tokens that were related to twater in such and such way.

  5. Pete Mandik says:


    I remain unconvinced that the slow-down argument is unconvincing.

    The challenge the slow-down argument poses for the externalist is to give principled non-question begging reasons for why some features but not others of causal histories are content constituting. Prior to certain advances in chemistry people were incapable of detecting the properties of H2O that would distinguish it from XYZ, but nonetheless, what water is really made of matters for the content of “water” tokenings. Why doesn’t how much time really passes matter for when the content is slowly switched?

    I suppose its relevant to ask externalists how much time “one second” refers to in the A and B scenarios. My guess is that they would say that in B, “one second” refers to a hundred seconds. (Although, I actually have no idea what they would say.) If this supposition is correct, then how much time really passes does matter for some content constituting events. Why not for others?

  6. Ben Young says:

    Hi Pete,

    I like the experiment, but have the following simple concerns/questions:
    1. What is a subjective supervenience base and how is this determined?
    2. How is subjective time measured?

    I think you might have dealt with this in your reply to Chase, but if time does matter and these are subjective units I am unsure how an externalist will be willing to come on board for an experiment with such murky units.

  7. Pete Mandik says:

    Hi Ben,

    1. I suppose the “subjective” in “subjective superveneince base” is the part in question here, mostly because I’m not in the mood to spell out “supervenience base”. So, I suppose, what would make it subjective is that it is comprised of noticible properties, where noticibility is defined relative to a class of observers. Relative to pre-scientific observers, there are no noticible differences between H2O and XYZ.. The property of looking like water supervenes on properties that H20 and XYZ have in common.

    2. By looking at clocks. In B, the observers and the clocks are both slowed down, so they don’t notice that 200 seconds have gone by when it seems to them (by looking at their B-clocks) that 2 seconds have gone by.

  8. Jason Zarri says:

    “What sort of causal interaction is it that is going to make the B supervenience base both (1) a hundred year process and (2) not limited to the superficial causal properties of XYZ? Or, to put it another way, why is it that what water is really made of matters but how much time really goes by does not?”

    What sort of causal interaction is it that is going to make the B supervenience base a hundred year process? Speaking on behalf of the externalists, I would say none. It’s a *lack* of causal interaction that makes slow switching a hundred year process, specifically, a lack of any causal interaction between XYZ and the speaker during the freezes that hold for every 99 out of 100 objective seconds. The externalist should just resist the idea that the mere passage of time has any effect at all on the content of “water” tokens; instead time is essential only as a precondition for the existence of causal chains which connect XYZ to my utterances of tokens of “water”. By externalist lights, then, it would be more accurate to say that slow-switching doesn’t occur during the freezes, because nothing causes anything to happen during the freezes; the slow-switching is “frozen” along with everything else. It seems proper to say then, paradoxical though it may sound, that the slow switching is a one year process that takes a hundred years to complete. This is possible because the process is “gappy” and only occurs during 1 out of every 100 objective seconds on twin earth B. The slow-switching process is in effect “chopped up” into one-second snippets and strewn evenly over the 100 year expanse of time with 99 seconds of “gap” between each snippet. This, so the externalist should conclude, is why how much objective time passes doesn’t matter: The duration of time from the beginning of the process to the end of the process isn’t the same as the duration of the process itself. The former is 100 years and the latter is only 1 year. As to why the constitution of water *does* matter for the content of “water”, I’m not sure what an externalist should say, though Chase seems to be on the right track. What I do think is that at some point we have to take some semantic facts (or laws, if you prefer) as primitive. Could we really give an account that explained every fact about why a given state has the given content it does without begging any questions? But if we have to accept some semantic facts as primitive, I see no reason why an externalist couldn’t take the content-constitutiveness of some features of the tokens’ causal history as primitive.

  9. Pete Mandik says:

    Hi Jason,

    I’m happy to let the externalist (or anybody) take some things as primitive, that is, as things in need of no argument or explanation. But if a theory gets filled up with too many primitives, it does subtract from its appeal. Also, the primitives should be consistent with each other. How much time does “one second” refer to in the B scenario? My guess is that the externalist will want to say that it refers to a hundred seconds. Is this entailed by the primitive that answers the H2O/XYZ question? I can’t exactly see how it would, but that may be a failure of my imagination. Or, instead, does some additional primitive need to be added? If so, what would it be such that it is consistent with the H20/XYZ primitive?

  10. [...] I wonder if temporal properties are similarly split with regards to Twinearthability. I wrestled with this a bit in the puzzle raised in “The Slow Switching Slowdown Showdown” wherein I wondered out loud about how long slow switching would take on a demonically slowed Twinearth. I wonder now about which properties broadly describable as temporal would be Twinearthable and which would not. [...]