Friday, February 14, 2014

Might I Be a Cosmic Freak?

A "freak observer" or "Boltzmann brain" is a conscious being who did not arise in the normal way on a large, stable planet, but who instead congealed by freak chance out of chaos, due to a low-probability quantum or thermodynamic fluctuation -- a conscious being with rich seemingly sensory experience, rich seeming-memories, and capable of sophisticated thoughts or seeming-thoughts about itself and its position in the universe. By hypothesis, such a being is massively deluded about its past. And since random fluctuations are much likelier to create a relatively small system than a relatively large system, and since a relatively small system (such as a bare brain) amid chaos is doomed to a short existence, most freak observers will swiftly perish.

If certain cosmological theories are true, then almost all conscious systems are freak observers of this sort. Here's one such theory: There is exactly one universe which began with a unique Bang, which contains a finite number of ordinary non-freak observers, and which will eventually become thin chaos, enduring infinitely thereafter in a disorganized state. In any spacetime region there is a miniscule but finite chance of the spontaneous freak formation of any finite organized system, with smaller and less organized systems vastly more likely than larger and more organized systems. Given infinite time, the number of spontaneously formed freak observers will eventually vastly outnumber the normal observers. Whatever specific experiences and evidence I take myself now to have, according to this theory, to any finite degree of precision, there will be an infinite number of randomly generated Eric Schwitzgebel clones who have the same experiences and apparent evidence.

Can I prove that I am not a freak observer by counting "1, 2, 3, still here"? Seemingly no, for two reasons: (1.) By the time I reach "still here" I am relying on my memory of the "1, 2, 3", and the theory says that there will be an infinite number of freak observers with exactly that false memory. (2.) Even if assume knowledge of my continued existence for three seconds, there will be an infinite number of somewhat larger freak observers who congealed simultaneously with a large enough hunk of environment to exist for three seconds, doing that apparent count. If I am such a one, I will very likely perish soon, but it is not guaranteed that I will perish, and if I don't perish and thus conclude that I am not a freak I have ignored the overwhelming base rate of freaks to normal observers.

Suppose that given the physical evidence such a cosmology seems plausible, or some other cosmology in which freak observers vastly outnumber normal observers. Should I conclude I am probably a freak observer? It would be a strange conclusion to draw!

One interesting argument against this conclusion is the cognitive instability argument (Carroll 2010; Davenport & Olum 2010; Crawford 2013): Suppose that my grounds for believing that I am a freak observer are Physical Theory X, which I accept only conditionally upon believing that I have good empirical evidence for Physical Theory X. If I am a freak observer, then, contrary to the initial assumption, I do not have good empirical evidence for Physical Theory X. I have not, for example, despite my contrary impression, actually read any articles about X. If I seem to have good empirical evidence for Physical Theory X, I know already that that evidence is almost certainly misleading or wrongly interpreted -- either I do have the properly-caused body of evidence that I think I have, that is, I am not a freak, and that evidence is misleadingly pointing me to the wrong conclusion about my situation; or I am a freak and I don't have such a body of properly-caused evidence at all.

For this reason, I think it would be irrational to accept a cosmological theory that implies that almost all observers are freak observers and then conclude that therefore I am also a freak observer.

But a lower-confidence conclusion seems to be more cognitively stable. Suppose our best cosmological theory implies that 1% of observers are freaks. I might then accept that there is a non-trivial chance that I am one of the freaks. After all, my best understanding of the universe implies that there are such freaks, and I see no compelling reason to suppose that I couldn't be one of them.

Alternatively, maybe my best evidence should leave me undecided among lots of cosmologies, in some of which I'm a freak and in others of which I'm not. The possibility that I'm a freak undercuts my confidence in the evidence I seem to have for any specific cosmology, but that only adds to my indecision among the possibilities; it doesn't seem to compel elimination of the possibility that I am a freak.

Here's another way to think about it: As I sit here in my office, or seem to, and think about the scope of the cosmos, I find myself inclined to ascribe a non-trivial credence to some sort of very large or infinite cosmology, and also a non-trivial credence to the hypothesis that given enough time freak observers will spontaneously form, and also a non-trivial credence to the possibility that the freaks aren't vastly outnumbered by the normal observers. If I accept this conjunction of views, then it seems to me that I should also assign a bit of credence to the possibility that I am one of the freaks. To do otherwise would seem to commit me to near certainty on some proposition, such as about the relative nucleation rates of freaks vs. environments containing normal observers, that I wouldn't normally think of as something I know with near certainty.

Or maybe I should just take it as an absolutely certain "framework" assumption that I do have the kind of past I think I have, regardless of how many Eric-Schwitzgebelesque freaks the cosmos may contain? I can see how that might be a reasonable stance. But that approach has a dogmatic air that I find foreign.

If I allow that I'm not absolutely 100.0000000000000000000000000000% certain that I'm not a spontaneously formed freak observer, what sort of credence should I assign to the possibility that I am a freak? One in million? One in ten trillion? One in 10^100? I would like to go low! But I'm not sure that it's reasonable for me to go so low, once the possibility occurs to me and I start to consider my reasons pro and con. I'm inclined to think it is vastly less likely that I am a freak observer than that this ticket will win the one-in-ten-million Lotto jackpot -- but given the dubiety of cosmological theories and my inability to really assess them, should I perhaps be considerably less confident than that about my non-freakish position in the cosmos?


JK said...

We are 100% certain of the composition of living cells. However, there are limitations to human measurement. There is a possibility that at some future point there is something even more microscopic within mitochondria, say, that is even smaller than we are currently capable of discerning. I'd give that a ten percent credence. I don't think being a cosmic freak is anything quite so large, based on the nature of the contemplation. Especially, since one could argue that we all might be rather an intersection of unique elements.

Zach Barnett said...

I am not 100.00000000% certain I am not a freak observer. But in general, I assign it a *very* tiny probability, even conditional on the most Boltzmann-friendly cosmological view we can pick (within reason).

I wrote a comment about this a while back, and I think you found it somewhat persuasive. But here's take 2.

Suppose you and I are sitting outside of the universe watching things unfold. You challenge me to a bet. You choose a big hunk of space-time, and I pick one of equal size, and then we watch our respective regions unfold. Whoever's region produces more "Schwitz-seconds" wins. (If five Schwitzgebel brains exist for 12 seconds each, that would total one Schwitz-minute.) Suppose we pick regions big enough to make it likely that both regions will produce non-zero Schwitz-time.

Perhaps some of our Schwitz-time will come from lucky, Boltzmann-style, highly deceived Schwitz brains, and some of it will come from ordinary non-deceived Schwitzes living out their full lives.

The issue is whether one should expect more of our Schwitz-time to come from ordinary non-deceived Schwitzes, or from Boltzmann deceived Schwitzgebel brains. (Expected Proportion Principle: If p is the proportion of non-deceived time to deceived time we are rational expect, then p is the credence (real) Schwitz should have that he is not deceived. Not sure about this, though.)

There are a few reasons to expect that the vast majority of our Schwitz-time will come from non-deceived Schwitzgebel brains.

First, if we are lucky enough to have a non-deceived Schwitz brain pop up in our region, it is going to net us a lot more Schwitz-time than even a slew of Boltzmann Schwitz brains, of which the vast majority will perish within nanoseconds.

Second, just compare the expected total number of evolutionarily arising Schwitzes with the expected total number of Boltzmanian Schwitz brains. I think it is much 'easier' to create one of the former than one of the latter. The reason is that you get to make use of "checkpoints." First, get a planet the right distance from a star. Next, get the raw ingredients to the planet. Next, get just a single living thing. Next, get the right sort of species to evolve. Then, get a Schwitzgebel to be born. With each successive step, you will get many, many opportunities for the next step to occur. This succession of mini-miracles seems much more likely to me than the occurrence of an Uber-Miracle. I don't really have a new argument for this, though, beyond what I said in my old post.

Eric Schwitzgebel said...

Zach: You put the case nicely, but it really does turn on some cosmological assumptions that might not be correct, such as that universes don't basically collapse into an infinitely enduring de Sitter-ish space after a finite period, nucleating stable environments at a much lower rate than they nucleate freaks. *Probably* that's right. I find it an attractive and plausible view. But if I give even a small slice of cosmological credence to a cosmology closer to the one articulated in the post, then I've got Boltzmann brain worries.

Zach Barnett said...

Hmmm... I see. Sorry about the confusion on my part. I didn't see what you were saying in the post.

I think you are right. But I don't want Boltzmann worries! How might we be able to address this?

Maybe one way to resist is to question the assumption that the probability of Boltzmann Schwitz's arising from 'thin chaos' will forever exceed some tiny probability bound. Maybe as the 'thin chaos' becomes more and more chaotic, the probability of your arising decreases more and more, dropping below any bound we could choose.

If this is true, then the number of freak-observers wouldn't have to dwarf the number of finite observers, even over an infinite duration. But this is way over my head.

Eric Schwitzgebel said...

Zach, on current cosmological theory, as I understand it, the most likely outcome of an infinitely enduring universe like ours is some de-Sitterish thin vacuum that doesn't have the asymptotic feature you describe. So De Simone et al 2010, for example, try to escape this by some combination of vacuum decay (the vacuum at some point collapses into nothing or into something that can't host freak observers) and the nucleation of new normal-observer-friendly environments (e.g., big bangs). If the ratios of those things are right and you compare the infinitudes in the right way, you can avoid freak-observer domination.

But here I think a bit of a long-term philosophy of science perspective is the right thing to do. Cosmology has changed a lot over the decades, and I don't see reason to invest high credence in even our best current thinking about these matters. I think the most reasonable reaction is not to radically adjust our prior virtually-zero credence in being a freak observer in light of dubious current science, but rather to tweak it just a tiny but. Current science suggests that the possibility is not as totally ridiculous as I had previously thought. My own inclination is to invest about 0.2% of my credence in the disjunction of all radically skeptical cosmological scenarios combined (of which the Boltzmann brain scenario is just one). That's still, actually, I think, quite high. But it's not crazy-high.

clasqm said...

"Given infinite time, the number of spontaneously formed freak observers will eventually vastly outnumber the normal observers."

Well, let's ignore for now the chauvinistic rhetoric, except to note that the Boltzmann Brain Student Interest Group on Campus may be posting a strongly worded note of protest to the Dean of Students, objecting to your use of the word "freak".

They do have a point, though. If the BB's are in the vast majority, then what basis do we have to declare that the planet-bound brain (PBB) is the original of which they are pale copies?

Indeed, how can we even know that even a single PBB actually exists? BB life may well be the only form of sentient existence in this universe. Your probability of being a BB would then be 100% and you should worry if you are in that tiny minority of strictly hypothetical planet-bound freaks.

Oops, did I just say that?

UserGoogol said...

Regarding the "1,2,3: still here" case, even though there would be infinitely many people with those memories, there would be even more people who would have memories which in no way matched up with their current actions. Having false memories and present experiences match up is a staggeringly unlikely coincidence for a freak observer, but quite likely if your memories are the causal result of a long series of events building up to your current experience. (And similarly for longer term memory, although things get more ambiguous the further back you push.) So that helps too, I guess.

Eric Schwitzgebel said...

I agree, UserGoogol. Nice point. But it only helps if the base rate isn't infinity to one!

Eric Schwitzgebel said...

Michel: LOL. I love it!

Zach Barnett said...

I think you have a point. Sorry if I was a bit of a nuisance!

Eric Schwitzgebel said...

Not a nuisance at all, Zach! Your thoughtful comments are very interesting pushback, forcing me to think more about what exactly is going on in these arguments.