We could imagine that all the computation a deterministic brain does being done by an incredibly complex system of gears operated by a single turn of the crank to generate all the different intermediate computational results in different gears. Now, imagine a Newtonian world with frictionless, perfectly rigid and perfectly meshing gears, and suppose that the computations are done by that system. Perfectly rigid and perfectly meshing gears compute instantly. So, all the computation of a life can be done with a single turn of a crank. Note that the computational states will then have an explanatory order but need not have a temporal order: all the computations happen simultaneously. So:
- On a computational theory of mind, it is possible to live a conscious mental life of many years of subjective flow of consciousness without any real temporal succession.
It follows that:
- Either computational theories of mind are false, or the subjective flow of consciousness does not require any real time.
I think there is a potential problem in (1) and (2), namely a potential confusion between real time and external time. For it could be that internal time is just as real as (or more real than!) external time, and is simply constituted by the causal order of interactions within a substance. If so, then if the system of gears were to be a substance (which I think it could only be if it had a unified form), its causal order could actually constitute a temporal order.
This and other recent posts could fit into a neat research project—perhaps a paper or even a dissertation or a monograph—exploring the costs of physicalism in accounting for the temporality of our lives. As usual, I am happy to collaborate if someone wants to do the heavy hitting.
11 comments:
This is cool. I was actually working on an argument just like this several months ago against functionalism, but I set it aside to work on other things. In my argument, I was thinking of using a scenario in which a case of simultaneous causation duplicates the entire causal structure of our physical universe, but I think your example of computation with gears is a lot cleaner (it's easier to conceive of, and it's easier to motivate the claim that your case is metaphysically possible).
One way to develop an argument along these lines would be like this: Consider a day-long stream of experience, where your day begins with a sensation of red and ends with a sensation of green. The following "temporal congruence" thesis is very plausible:
(1) Necessarily, any stream of experience that is phenomenally exactly like your day-long stream of experience includes a green sensation that occurs objectively after a red sensation.
Then it can be argued that, if computationalism is true, then (using a scenario like the one you've described) it's possible for there to be a system computationally isomorphic to your brain over the course of the day, but where the computations corresponding to your green sensation don't occur objectively after the computations corresponding to your red sensation. From here, it plausibly follows that, if computationalism is true, then (1) is false, hence computationalism is false.
There's a growing literature in philosophy of mind discussing the possibility of cases where there's a mismatch between subjective temporal ordering and objective temporal ordering. But while many think there can be slight mismatches, it seems absurd to think that there could be mismatches of a kind that would falsify (1).
I might be interested in co-writing something that develops an objection to computationalism along these lines, if I can find time in the next month or two to start writing something up. (I probably won't be able to devote any significant time to it over the next couple weeks though.) I'd be happy to do "heavy hitting" in the sense of writing up an initial draft. But it's totally fine with me if someone else wants to collaborate with you instead. I agree that *someone* should do it, since there's a fair amount of good work being done on time consciousness these days, but not a lot of it makes contact with the big metaphysical debates about the nature of consciousness.
Brian,
Thanks for writing!
Please do tell me more about the mismatch literature. Here is an innocent mismatch: A occurs before B, but the perception of B occurs before the perception of A. This happens commonly with external events A and B (i.e., you see the ball move away from the kicker and then your hear its being kicked), but there is also nothing particularly problematic about a case where A and B are internal states, since the perception of an internal state need not happen when the internal state does.
Here's a different innocent mismatch: experience A is had before experience B, but instead of having the experience of A-before-B, one has the experience of B-before-A. That's innocent, because the experience of B-before-A is a third experience over and beyond the experiences of A and B, and one shouldn't expect this third experience to be infallible.
But in my recent posts, the mismatch that interests me is this: experience A is had before experience B in the experiential sequence but experience B is in the order of time before (or simultaneous with) experience A. That at least sounds like a self-contradiction. (Note that to have A before B in the experiential sequence does not require that one also have the experience of A-before-B. In cases of amnesia, it's can be true that one has A before B but one doesn't have A-before-B, since by the time B happens, one may have forgotten A.)
However, I am worried about the internal/external time issue.
Without an internal/external time distinction, (1) is falsified by time travel.
With an internal/external time distinction, (1) is still falsified by time travel unless we say that "internal time" is *objective*. Which I think we should.
But if we do that, then we have to walk a fine line to run the argument. For your or my case to run afoul of (1) on a computational theory of mind, the causal order cannot actually define an objective internal time. But it seems to me that it might.
Perhaps one way to get out of all this is to aim the argument only at A-theorists and conclude that A-theorists have to be dualists, or at least can't be computationalists. We can then take care of the time travel issue by saying that the A-theory is not compatible with time travel (there is a literature on this that I am not familiar with), and recognizes only one objective timeline.
Two more minor points:
1. Another way to run the thought experiment is to embed the causal order of our 3+1 dimensional world inside the spatial order of a 4+1 dimensional world.
2. To falsify (1), one needs an internalist computationalism about sensations. It seems to me that the best computationalist theories of mind will have a significant dollop of externalism. In particular, I doubt that on the best versions of these theories there will be any perceptions of color in worlds whose laws are significant different from ours. To get around such worries, I think it helps to think in terms of introspective experience of a fairly abstract nature (the experience of thinking about integrals, etc.).
Another thought experiment that seems to me to be potentially helpful with responding to my internal time worry.
Alice is a conscious computer running deterministically. On days 1-4 she thinks about calculus. Then her memory is backed up. On day 5, she thinks about foundations of mathematics (FOM). On day 6, she thinks about group theory (GT). Then, the backup is restored by the operator, and so on day 7, she again thinks about FOM. And then she is deactivated. So, the last part of Alice's experiential sequence goes: FOM, GT, FOM. But notice that there is no computational causal connection between GT and the last FOM portion. So the order in experiential sequences is not determined by computational causal connections. And there really an order. Suppose that Alice had a really unpleasant time thinking about FOM on day 5. Then at the end of day 6, it would be rational for her to dread the restoration of the backup, precisely because her *next* experiences will be unpleasant.
The mismatch literature I have in mind mostly focuses on cases like the second class you describe, e.g., a case where your experience of A occurs before your experience of B, but you experientially represent B occurring before (or maybe at the same time as) A. I guess I agree that cases like this are "innocent," in the sense that they really do occur. Still, not everyone agrees that they really occur, and I think even these cases are pretty weird. Example: you have an experience as of "do" occurring before "re" occurring before "me," but you actually experience the "me" first, then the "re," then the "do." Again, I think this kind of thing can happen, at least when the notes occur in quick succession, but it's still pretty weird and counterintuitive. And when we consider really extreme cases of this, it can start to seem metaphysically impossible. Consider, e.g., an extended experience phenomenally just like your experience of listening to a very slowly-sung scale, "dooooo, reeeeee, meeee, faaaaa...etc." but where the experienced temporal order of the notes is the opposite of the order in which you experience each note.)
Re: the literature: Ian Phillips has a survey paper in philosophy compass called "experience of and in time," which surveys the controversy over mismatch cases like this. (Not all the cases he discusses are exactly like this, but they are all in the same general ballpark.) Related issues are discussed in Simon Prosser's recent book, "Experiencing Time," and in Geoff Lee's cool papers, "Temporal Experience and the Temporal Structure of Experience," and "Consciousness in a space-time world" (the second of which deals with some puzzles about temporal consciousness stemming from special relativity---in the same neighborhood as those you've talked about in recent posts, actually.) Phillips defends the "inheritance" thesis, which says something like: if an experience represents a temporal feature, then the experience has that feature (so I think he'd deny that the "do re me" cases described above are real possibilities).
You say, "But in my recent posts, the mismatch that interests me is this: experience A is had before experience B in the experiential sequence but experience B is in the order of time before (or simultaneous with) experience A. That at least sounds like a self-contradiction."
Yeah, that claim sounds like a contradiction if it doesn't just mean the thing described above (the thing violated by the do-re-me case). (Or unless there are two separate timelines in the same world, where a single event can be located in two distinct timelines?)
I don't think it's easy to state a *general* "no mismatch" principle such that (i) the principle is independently plausible on reflection (unlike your first two "no mismatch" principles) and (ii) violations of the principle aren't outright contradictory (unlike, arguably, your third "no mismatch" principle). That's kind of why I like just focusing on intuitions about specific cases---like the green and red sensation case.
Anyway: I agree there are worries about internal and external time (and also with the distinction between real time and external time mentioned in the original post). Will think about these more...
Here's a reason to think your simultaneous causation version of the argument is better. One might think that it's crucial to a mental life that the mental life occur within a system such that the system's existence at later times is caused by the system's existence at earlier times. In my gear case, movement is transmitted from gear to gear, but not existence.
I think few computationalists would want to go for this version, though, because I sense that they would generally want to allow minds to move from one piece of hardware to another.
Brian:
Wouldn't your slow scale example just be a case of memory failure? You experienced things in one order and you misremember them in the other order?
I think you have in mind a case like this: You experience "do-te-la-sol-fa-me-re-do," but upon hearing the last note, you have a false memory of having just experienced "do-re-me-fa-sol-la-te-do." Certainly that's possible, but I don't think it's a case in which your total extended experience is phenomenally just like an ordinary experience of "do-re-me-fa-sol-la-te-do." I think it's an experience phenomenally just like an ordinary experience of a backwards scale, except with deviant memory phenomenology at the end. Also: I think that, when you experience the descending scale with the memory failure at the end, your extended experience can be correctly described as representing the notes as occurring in descending order (e.g., with me after do), even if your memory at the end represents them as occurring in ascending order.
But it's a bit hard to know what to say about the contents of temporally extended experiences. I guess I think it's true that there are temporally extended experiences much longer than a specious present (say, minute-long experiences, maybe day-long experiences) and that we can associate contents with these experiences that assign temporal structure to long series of perceived objects and events. But it seems that the temporal contents of these extended experiences aren't totally independent of the temporal contents of their parts. Like, maybe an extended experience of an ascending scale can be said to represent the whole scale in ascending order, but only because of the simpler temporal contents of its shorter parts, e.g., because it's made up of an experience that represents "re after do," and another that represents "me after re," and so on.
Brian:
I am sceptical of the whole idea of associating contents with experiences. I am not inclined to distinguish the content of an experience from the experience, because I am inclined to say that experiences are just a type of propositional attitude, and their qualitative content supervenes on their propositional content. (Of course, their propositional content is typically not expressible in spoken or written language.) A mere sequence of experiences then does not, then, have a content, just as a mere sequence of assertions or desires does not have a content (the person who asserts that p and then asserts that q has not thereby asserted the conjunction).
I doubt that such a machine could even compute in the required sense, let alone be conscious.
Think about the physics of computation. When, for example, a flip-flop flips, some energy must be dissipated to lock down the new state. (Otherwise it would bounce for ever.) Plausibly, this applies to any computing device, hence to any conscious computing device. (Recall Landauer’s principle linking entropy and information.) So no computing device can be ‘frictionless’ (= reversible).
Ian:
We actually don't need literal frictionlessness. All we need is that the friction be low enough that a rotation of the crank will move all the gears at once. But in any case there is no issue with bouncing given that everything is driven by a crank (perhaps turned by Superman). Of course, we need enormous force on the crank to get everything moving. But since we're talking of perfectly rigid (and indestructible, I should have added) components, that should work.
I am thinking of this as built using combinatorial logic alone, by the way.
Here's combinatorial logic with linkages: https://arxiv.org/pdf/1801.03534.pdf
That’s a fascinating paper – thank you. My mental image was wrong. The ‘gears’ in the post mislead me into thinking of something more like fancy clockwork.
It’s not so easy to avoid the external world. When I think about mathematics, I read accounts, write equations, draw diagrams, point at imaginary diagrams, mumble formulas… You get the idea. I doubt that I could do it without these aids. I speak for myself, of course. :-)
Dreams seem to be a good candidate for a purely internal experience. They usually seem to last much longer that the external physical time they take. So, pushing this a bit further, maybe a computationalist would be prepared to bite the bullet on (2).
Post a Comment