Expert: Jeffrey Eldred - 2/4/2011

Question
Dear Jeffrey:

What we call as science is essentially stuff that can be independently verified and repeatable anywhere in the world at any point in time and the results will come out identical.(I have no problems with this on a practical level.)

Yet, isnt this a *claim* or *belief*? It is not clear to me *at a philosophical level* whether this is self-evident. This arose as I was thinking through the atheist position that the burden of proof of God is in the believer's court. Likewise, *at a philosophical level* isnt it true that the entire epistemology of science is based on the repeatability of experiments and the claim that the results will be the same everytime? Is it fair to have the atheist shoulder the burden of the proof of this claim? And how would an atheist go about proving this?

I appreciate your time.

Answer
Problem of Induction:
What repeatability really depends on inductive reasoning: Something that has operated a certain way in the past will operate that way in the future and therefore inconsistent results point to inconsistent procedure. The dilemma you are proposing is known as the Problem of Induction [1]. It should be noted that there are no academics which actually doubt that induction is true, there are only philosophers who think it would be useful to identify why we can rely on it. Similarly if we are talking about theology, the notion of an eternal God only makes sense with induction - because if we considered all the possible lifespans that God could have than the probability that “he'd always have been and always be” is actually zero. An argument that appeals to any kind of empirical evidence for God is similarly plagued by the Problem of Induction. I know of no proof for God that can avoid induction altogether, but even if there was it would not make any practical difference except where induction was applied to real world principles. In this sense if repeatability is at a philosophical level an unjustified assumption than it is a not a unique burden on the atheists but rather on the whole of human population.

Hume says that it is a matter of practicality that forces humans to rely on induction. If you put your hand on a hot stove and it hurts, who would put their hand on that hot stove again? Whenever someone grabs an object, they make the assumption that they will be able to touch it and pick it up rather than have their hand pass through it. Hume however says that he cannot justify the principle of induction, but it is so necessary that even if induction were not true we'd still have to use it.

But it's not even just the case that we cannot accomplish anything without induction, it's that we are physically incapable of removing the notion from our brain. Biologically speaking, induction is the same thing as continuity, object permanence[2], or the gestalt psychology[3]. You expect an object to remain these even if you blink, you expect that if you see part of a well-known object that the rest is there, and you expect empty space to remain empty as well. Our visual system has this effect built into it at a very low-level. The eye actually has a blind spot right in the middle that you don't see because your eye is always moving slightly and displays what was there when it last looked (not unlike a framebuffer[4]). There are similar effects in other human senses, but they are generally less understood and harder to explain.

If the notion of being forced to put up with induction without justifying it is a troublesome thought, than there may be a way to justify induction with probability. Psychologically speaking this is one of the aspects of probability that it is hardest for humans to understand[5][6][7], so hopefully my explanation is clear. Suppose you knew a barrel contained 100 hundred balls that were either blue or red, but you didn't know what proportion the balls actually made up the barrel (Y). All you knew is that you drew a sample of 10 balls from the barrel (with[8] or without replacement[9], the analogy works either way) and you find that 1 of the balls is blue and the other 9 are red (X). At this point you are probably imagining that there barrel was composed out of 10 blue balls and 90 red balls but you don't know that for certain (given X, what is Y?). Instead it could have been fewer than 90 red balls but you just happen to draw an unusually high number of red balls in your sample or it could have been fewer than 10 blue balls and you just happened to draw one of them. But 10 blue balls and 90 red balls is special because it is the proportion of red and blue balls in the barrel for which it would be most commonplace for you to draw 1 blue ball and 9 red balls (given Y, what is X). But still, 10 blue balls and 90 red, is only a good estimate of the proportions in the barrel if we assume that all proportions would have been equally likely before you drew your sample. But if I told you that the barrel came from a factory which attempted to fill each barrel with red and blue evenly then you would probably think that your barrel had something closer to 50-50 or at least 40-60. And if I told you that red balls were incredibly rare so there were only 90 in the universe than you would probably find it unlikely that they were all in this barrel despite the fact you drew 9 already. This component of the question is called prior probability (or base rate)[10] and more likely than not it itself is dependent on other prior probabilities (what is the probability that the factory will malfunction as it fills a barrel? What is the probability that any of the information that I gave you was a lie? etc) [11]. Induction is essentially the neutral, reasonable, assumption that we make in the absence of any information about prior probability (and in this case, the inductive answer would be that the barrel is composed of 10 blue balls and 90 red balls). But it's not always clear that there is a neutral reasonable assumption to every probability problem, and maybe there can be some legitimate philosophical arguments that could be made about general approaches to prior probability [12]. But even still, it seems like there is something objective and deductive that can be said about learning from past experience, i.e. induction.


An Overview of at My Epistemology:
Allow me to briefly construct the bigger picture in which induction and repeatability fits. To begin I have to say how I know that logic (of any kind) is a valid path to truth, without using logic to get there. I cannot actually imagine going without logic, so the question is somewhat moot. However in attempting to justify logic it is worth noting that logic contends that it is more reliable than the illogic whereas illogic makes no claim at all, and therefore any kind of analysis that takes both into consideration has to favor logic over illogic.

The next we observe is that we cannot deny the truth/existence of our senses. I can deny that I sitting in a chair, but I cannot deny that I have the sensation that I am sitting in a chair. Applying induction to the senses allows me to more accurately predict future sensations. For instances I predict that my hand will not pass through the wall when I go to put my hand on it, as I have done before. Even if all my sensations are fed to me by my subconscious[13] or by an external agent[14], than induction will still inform me of the most likely way my sensations will be fed to me in the future. I would support the notion of their being an objective reality however, by the fact that I can be surprised and that the rules in the universe seem to be indifferent to me. It seems worthwhile to act as though I have absolute certainly that the universe I perceive exists outside of my sensations, even if I cannot prove it[15].

If indeed there exists a universe outside of myself than my sensations of observing human beings which act like me would seem to indicate that other human beings are like myself with regard to the universe. They have sensations, thoughts, and feelings just as I do and as such I can use them to independently evaluate my ideas about how the world works. Over time a procedure of peer review, including repeatability, has been developed for this purpose.

As an aside, my understanding of individuals outside of myself also sets up some ethical notions. I realize the principle of mutualism, that is, that I can cooperate with other individuals for our mutual benefit. I also realize the Golden Rule and empathy, because I cannot come up with reasons to value myself that would not also apply to other people. Consequently I adopt a moral system that is concerned about the life and happiness for all humans and any moral proposition I accept after this point must be reconciled with these priorities.

Hopefully my long-winded response fleshed out all the concerns you had about systematic assumptions in the scientific worldview.


Links:
[1] http://en.wikipedia.org/wiki/Problem_of_induction
[2] http://en.wikipedia.org/wiki/Object_permanence
[3] http://en.wikipedia.org/wiki/Gestalt_psychology
[4] http://en.wikipedia.org/wiki/Framebuffer
[5] http://en.wikipedia.org/wiki/Base_rate_fallacy
[6] http://en.wikipedia.org/wiki/Monty_Hall_problem
[7] I was once at a particle physics conference where a physics asked a question that demonstrated that he failed to understand the relevance of the base rate to the problem. It can happen to anyone.
[8] http://en.wikipedia.org/wiki/Binomial_distribution
[9] http://en.wikipedia.org/wiki/Hypergeometric_distribution
[10] http://en.wikipedia.org/wiki/Prior_probability
[11] http://en.wikipedia.org/wiki/Bayesian_probability
[12] http://en.wikipedia.org/wiki/Skepticism
[13] http://en.wikipedia.org/wiki/Solipsism
[14] http://en.wikipedia.org/wiki/Descartes_demon
[15] http://en.wikipedia.org/wiki/Pragmatism
[16] http://en.wikipedia.org/wiki/Mutualism_%28biology%29