Boltzmann brain teaser

May 16, 2011 3:50 PM

In his TED talk, Sean Carroll very briefly discusses Feynman's explanation about how the universe we can see and experience is not a statistically lucky perturbation, before moving on to the rest of his lecture. In other words, Feynman seems to discount the "Boltzmann brain" hypothesis, that we're not just the ephemeral product of a lucky shuffle of a metaphorical box full of marbles. Can someone explain this to me in words other than Carroll's, i.e., what evidence Feynman was using to prove his argument?
posted by Blazecock Pileon to Science & Nature (7 answers total)

It doesn't seem to me like he was using "evidence" per se, except in the minimal sense of "when we see new things, they look like things we've already seen". Rather, he was just making an appeal to our sense of likelihood: It seems more likely that new things look like old things because there's something that specifically leads that sort of thing than because random chance made all of the billions of new things happen to look like the old things.
posted by Flunkie at 4:22 PM on May 16, 2011

I would guess that Feynman preferred to believe that improbable events had a logical explanation. No other evidence ...
posted by aroberge at 5:11 PM on May 16, 2011

It is much more likely that a smaller, simpler structure will form randomly than a more complicated one. It would be billions of times more likely for just you and the room you're in to form than for the entire planet Earth to form. Therefore, it should be much more likely that if you open the door, you will be greeted by (and promptly sucked into) a near vacuum. But that doesn't happen. Everywhere you look you do not see a near vacuum of entropy, but large amounts of order. Which is not what you would expect if you were a Boltzmann Brain.
posted by justkevin at 6:43 PM on May 16, 2011

Maybe this analogy will be helpful if you're used to thinking about computers:

Imagine that artificial intelligence is possible-- there's a configuration of bits in memory on a sufficiently large computer that can be self-aware. If the computer has an infinite amount of memory, and that memory is randomly populated by 1 and 0's, you would find many instances of self-aware programs, just by chance. Almost all of those programs would be in a sea of noise.

If a program found itself not surrounded by randomness, but instead found itself next to Microsoft Word, or someone's thesis or other AI programs, it could reasonably deduce that it was not a "Boltzmann AI"-- it had come to be in the computer by some other process than randomizing the memory.
posted by justkevin at 7:01 PM on May 16, 2011

In my understanding of the lecture, Feynman wasn't proving anything—he was merely stating the drastic unlikelihood of our Universe's nature to map to Boltzmann's brain hypothesis, and therefore discounting the possibility. The reasoning behind this discount (and ultimate dismissal) was that it's easier for a BB universe to spontaneously produce something rather than spontaneously produce all of the context necessary for that same something to develop within a larger configuration.

The example in the video is apple pie. Under the BB hypothesis, it's relatively easy for a universe to consist of an apple pie, but extremely difficult for a universe to consist of all of the things necessary to produce an apple pie. You need earth, a sun, an orchard, an oven, some sugar, and a pie crust, and for all of those things to coincide within a rather (ok, extremely) small area.

Another example is you. You'd be an easy universe to create just as you are, without anything else at all. Just you. But to make your mom and dad, arrange for them to meet and fuck and birth and raise you just so you're exactly like you is ridiculously difficult and improbable to a laughable extent.

Feynman says that because our observations of reality as it stands are so incredibly, inconceivably improbable when viewed within the BB hypothesis, it can't hardly be true.

Again, that's just what I heard while listening to the lecture. It didn't sit right with me entirely, so I share your skepticism of the dismissal. I like the BB hypothesis, and I like the idea that our Universe is maybe the most difficult and improbable iteration of a box of gas.
posted by carsonb at 7:29 PM on May 16, 2011

One issue with Feynman's hypothesis is that a Boltzmann brain wouldn't know that its beliefs about what supposedly-existing scientists supposedly discovered about the supposed rest of the universe actually happened, or if they're just artificial psuedomemories based upon the random creation, and momentary existence, of the Boltzmann brain.

That is, I don't actually know that there are billions of stars in our galaxy, or billions of galaxies in the universe. What I know is that I have a memory that sure seems like I heard some guy say so. In fact, this is not even limited to subject matter as esoteric as my supposed knowledge of supposedly distant supposed galaxies, or my supposed knowledge of the supposed laws of supposed physics; when it comes down to it, I don't actually know that there really is an "outside" of this room I'm in now. All I really know is that it sure seems to me like I remember going outside this morning.

But under the assumption that I'm a Boltzmann brain, popped into existence a moment ago and about to pop out of existence, those memories do not necessarily reflect reality.

And the appeal to probability breaks down once you realize that the fluctuation only had to create me with the impression of having memories, not me and a billion galaxies.
posted by Flunkie at 9:19 PM on May 16, 2011

I think it's the same logical process that you use when when you're trying to decide whether you're having a dream. You know the dream that I'm talking about: not an ordinary night's dream, but one of those dreams that just goes on and on for as long as you can remember. All the details are there, and all the details are right. Your grandmother is there. She's changed her hairstyle: it's not the frail, wispy old-lady hair that she had in the years before she died, but the long silky hair that she still had when you were little, the long silky hair that she would ask you to brush if you would sit quietly on the bed. Your grandmother is talking with your PhD adviser about growing avocados. They are sitting together under the avocado tree, its branches waving a little in the cool breeze, the dappled sunlight bright and dark on their faces. You didn't remember that your grandmother and your PhD adviser were friends. Didn't your grandmother die years ago? You are awake. It is still nighttime.

The Boltzmann brain hypothesis is subject to all the classical arguments about solipsism, but with an additional counterargument based on thermodynamics. Suppose that our reality (well, my reality, since you don't exist) is just a fictional recollection based on a thermal fluctuation. That fictional recollection appears to be growing longer and more complex as I get older and observe more things. There are many ways that I can continue to observe things that appear to be consistent with "laws of physics." But there are many, many more ways that my recollection could evolve to be inconsistent with anything. Outside of my fever dreams I don't know about anything in the latter category. The longer my solipsist delusion goes on without screwing up, the less likely the "solipsist delusion" hypothesis becomes.
posted by fantabulous timewaster at 5:13 AM on May 17, 2011