mind fuck

So, suppose science and computing technology someday advance to the point that they can make a print out of exactly what thoughts are going to run through your head at what times.

(yeah, I know that's ridiculous, but it seems at least logically conceivable, though definitely not possible in our universe)

Ok, so they print out all your thoughts and then give it to you to read. So you're reading the thoughts that you're thinking as you read. So the predictions they made, which are really mental states, have been encoded in words and are now being decoded into perceptual information. It's like when you point a camera at the TV it's hooked to, and you get a crazy spiral feedback loop. In this case your mental states would loop around and become input to the computations that calculate the mental states from perceptual input.

HOOOOLY SHIIT!

 

I assume that the universe is mostly deterministic (except maybe at the quantum level) and that the quantum indeterminacy plays no important role in the functional relationships involved in consciousness. Of course that seems unlikely to work in the long term due to chaos, but we could at least get statistical predictions which would hopefully be accurate in the short term. And, of course, this would require more computing power than exists in the universe, but it's just a thought experiment. It's logically conceivable that a consciousness could have as input encoded versions of the mental states it experiences as it's reading the input. I'm just curious what the hell it would be like to be a consciousness in such a situation .

The laws of physics actually don't really matter, since I can just retreat to a logically conceivable thought experiment that can't be carried out in any imaginable world with natural laws remotely similar to ours. There's no logical contradiction about being a consciousness in such a situation. Sure, it's not ever gonna happen, but we can still wonder what it would be like to be in that situation.

 

I just thought of an interesting modification of this problem. Suppose we have a computer which randomly generates digital movies which are 800x600 pixels, 30fps, and 10 minutes long. There's a finite number of such movies (I think around 2.2 trillion). We randomly generate a movie and then watch it. If we keep at this long enough, it seems like we'll eventually watch a movie which shows exactly what we will do in response to the movie itself, but some small amount of time in advance. Naturally, to test out whether we are watching this movie, we'll wave our arms around as we watch the movies. When we get to this special movie, we will see ourselves wave our arms around in the film in exactly the same way we then wave them around.

The question is, then, is there any such video which will show us, in advance, exactly how we respond to the video itself? Couldn't we, for example, notice what the video predicts we will do, and then deliberately do something else just to falsify its prediction?

It reminds me of the problem of coming up with quines in computer programming.

I think one way we might imagine the possibility of such a video is to have the video include itself recursively. That is, the video would show us, some small amount of time in advance, watching the video on a TV screen, including a depiction of what exactly will be on the TV screen in a small amount of time.

This would of course create a sort of modified Droste effect-there is a series of nested video images contained in the displayed video image. We can appeal to the possibility of an movie which meets the desired criteria by reverse induction. Let I_0 be the image we wish to generate for a specific frame, and let I_1 be the nested image in I_0 which depicts the next frame. I_1 will contain a nested image I_2, which depits the next frame after I_1, etc. Assume that I_(n+1) can be drawn successfully. All we have to fill in, then, to draw I_n, is what you are doing when you see I_(n+1) on the TV. Since there are only a finite number of depictions of possible things you could do, one of these will work. So if we can draw I_(n+1), we can draw I_n

What's the base case for this induction? Well, there will be some k large enough that I_k is so small in I_0 that it must be depicted by one pixel. If we assume that the image I_0 is large enough, this one pixel will make absolutely no cognitive different to you as you watch the movie. We can choose any value for this pixel, and it will work just fine at the time, since you won't even notice it. We can then proceed via reverse induction to establish that I_0 itself is drawable.

This seems to imply to me that while it could be impossible to randomly generate an entire video beforehand which matched, it WOULD be possible to generate a video dynamically, as you were watching it, which would match.

What the hell would it be like to watch yourself doing stuff on TV and then do the exact same stuff a second later? There have been some optical illusions performed by researchers where people were convinced via illusion that other people's body parts belong to them and are under their control.
It's as if the optical illusion shatters the conceptual illusion we have that there is some deep causal connection, rather than just constant correlation, between our willing our body parts to move and their moving. Would watching a video of what you're about to do shatter the illusion of free will, and give one a direct experience of total determinism?