# The Bayesian Case for Reincarnation

### A discussion of a paper from Michael Huemer entitled "Existence is Evidence of Immortality"

Michael Huemer of the University of Colorado at Boulder has recently published a philosophy article entitled “Existence Is Evidence of Immortality,” in which he argues that either time is finite or people are able to live multiple times. As a secular person, I did not expect to be convinced of reincarnation, but I was. The argument rests on a few premises that are not so absurd and some mathematical reasoning.

The argument starts with the concept of time. The idea that we have an infinite future is widely excepted but the idea of an infinite past is controversial. A good reason for thinking that the existence of the beginning of time is impossible is that we can think of time somewhat analogously to space. If there was an edge of space which you could stand at, couldn’t you imagine sticking your arm out past it? Similarly, if we think of the moment that time started, it makes sense to think about one moment before it. There was a Big Bang that occurred 14 billion years ago but it is reasonable to suspect that things did not just appear in this unusual state for no reason. There could have been time before the Big Bang, potentially the big bang could be one of many in a never ending cycle of expansion and contraction.

The doctrine of Eternal Recurrence is the belief that everything that occurs will repeat itself in the future, infinitely many times. In a universe in which everything expands out and never contracts, there would not necessarily be recurrence. However, if the universe goes through expansions and contractions, there may be recurrence. Or if an infinite multiverse gives birth to daughter universes, then our particular universe may not return to old states, but some future one will approximate to old states of the current universe.

The argument made for reincarnation rests on some Bayesian reasoning, so Huemer explains the *Bayesian Conception of Evidence*. The Bayesian Conception of Evidence is E supports H if P(E|H) > P(E|~H). If you have some evidence, it supports the hypothesis if the chance of having the evidence given the hypothesis is true is greater than the chance of having the evidence given the hypothesis is false. The above inequality entails P(H|E) > P(H), which means that the evidence provides support for the hypothesis.

It is likely that there will exist persons that are very similar to you in the future, possibly simply as a consequence of the fact that rearranging particles will result in a recreation of a being similar to you. Is it the case that this clone of you is an incarnation? One could formulate a theory of persons in which a reconfiguration of you is similar enough that it counts as you. Huemer calls this a “Permissive” view. Another could say that spatiotemporal continuity is necessary for it to truly be you. Huemor calls this view “Restrictive.” We can assume at least non-zero probability that the correct view of persons is the view that people are repeatable, the permissive view.

Your existence counts as evidence that you are alive right now. It is also evidence that some person having some qualities lives at some time. However, the correct way to characterize evidence is specifically rather than broadly. We should’t strip down our evidence to have less details, so we shouldn’t characterize your existence as just evidence that someone existed.

The fact that you exist now is evidence of a permissive view of persons because P(L|H) > P(L|~H) where L = [You live now] and H=[Some permissive view of persons is correct]. If time is infinite, the probability of you existing in any given century is 0. If you believe it to be non-zero, maybe something like 1%, then over the course of 101 centuries, it would add up over 100% — an impossibility. You can do this with an increasing number of time intervals until you see the correct answer must be 0.

P(H|L) = P(H)*P(L|H) / [P(H)*P(L|H) + P(~H)*P(L|~H)]

= P(H)*P(L|H) / [P(H)*P(L|H) + P(~H)*0] (remember P(L|~H) is 0)

= P(H)*P(L|H) / P(H)*P(L|H)

= 1 (provided P(H), P(L|H) are nonzero)

You should dismiss a theory of the restriction view because given that you exist now, it is certain the restrictionist view is false. This means that the permissive view is the correct one. Therefore, persons are repeatable and reincarnation is real. This means you have lived before and you will live again.

*If you’re interested in more, check out this video explaining the argument, this blog post and the paper which goes into much more detail.*

I don't think this is correct. If we're going to allow actual infinities, we should allow infinitesimals. So the probability of you existing in any given century is infinitesimal, but non-zero.

> it is reasonable to suspect that things did not just appear in this unusual state for no reason.

Okay, but ... um ... how can there be an infinite past? The state of the universe at every point in time depends on the state at the time just before that. It simply doesn't make sense to say that there is no initial state. Because without a first state, the universe cannot determine its second state.

It's not like a number line in which every number is *independent* so we are free to imagine a negative infinity (and note that negative infinity isn't proven to exist in any physical sense; it's only a mathematical idea).

It's not relevant to your thesis, but I also disagree about the future being infinite. My model is that although the edges of space and time are both unimaginably vast and unreachable,

1. time is physically finite (rationale: it must have a beginning in *some* sense because causality exists, and "year infinity" will never happen, so clearly every future point in time is a finite point) and

2. space might well be finite also (under a Wolfram-like model of spacetime, which is hard to prove but has nice intuitive properties for us computer engineers). It's simple enough for space to be finite while having unreachable and invisible edges: it just needs to have an outer surface that always expands at the speed of light. So just like time, space can go on and on forever without literally being infinite.

> The doctrine of Eternal Recurrence is the belief that everything that occurs will repeat itself in the future

Is this like the Futurama episode? https://futurama.fandom.com/wiki/Forwards_Time_Machine

It doesn't look like the universe will collapse and produce another big bang.

> If time is infinite, the probability of you existing in any given century is 0.

I think this is where your reasoning went off the rails. Sure, it's wrong in my model where time is finite, but it's also wrong in general.

Suppose X is unknown to you, but its value actually happens to be 7^51 trillion googolplex. Since you don't know the value of X, and it could be anything on an infinite number line, you reason that the probability that it is any particular number (including 7^51 trillion googolplex) is zero. But actually X = 7^51 trillion googolplex, so you were wrong to discount that possibility.

Also, zero and one are not probabilities: https://www.lesswrong.com/posts/QGkYCwyC7wTDyt3yT/0-and-1-are-not-probabilities