Is there a principle in philosophy that tells you not to look at evidence in isolation?

Question

Suppose one is testing whether or not a person is a psychic who says that God has given him special powers. He guesses a number between 1 to 10 that you’re thinking of where you try to think of a fresh number each time. He does this 10 straight times. This seems extremely unlikely, if not miraculous by chance. Of course, we don’t think of numbers in a uniformly random fashion, but if we did, the chance would be low. Very low. 1 in 10^10 low. That’s 10 zeros. A one in 10 billion chance.

When looking at just this event in isolation, it seems hard to think that this was caused by chance. But when looking at the entire history of the world, this seems less obvious. What is more likely? That a God caused this psychic to exhibit magical powers yet chose to remain absent and hidden throughout the entire history of the world or that the entire history of the world is truly explained by naturalism and so this one being a one off event would fit in with that picture of the world?

Depending on how you frame the evidence, the answers can seem wildly different. In Bayesian terms, I suppose this equates to looking at prior probabilities. But is there a general principle that says to look at evidence as a whole and not in isolation?

Answer 1

As I've said before (and will likely say again), randomness as used in analysis is a model, not a thing-in-itself. When we use a random variable, that random variable is merely a way to account for things and events we have no knowledge of, i.e.:

• Truly random forces and effects, like quantum fluctuations (should such exist)
• Forces and effects that we have not measured, because they are too complicated, subtle, or numerous (should such exist)
• Forces or effects that we have no knowledge of whatsoever, because they lie outside our current scientific understandings (should such exist)
• Social forces and effects, like subterfuge, bias, or lies, that might lead us astray
• Forces and effects that are non-analytic, like Divine intervention or free will (should such exist)

A random variable is merely a way of putting numbers to a giant shrug, to make that shrug more manageable. It's an effective technique, but ultimately it's numerical sleight-of-hand; don't let it fool you.

Every random variable comes with a 'scope' — the class of events that the variable applies to — and changing the scope on the fly destroys any analysis. In your case, the scope would be something like "sequential predictions of a single individual", and any way you cut it getting it right ten times in a row is unusual. I mean, there are currently something like eight billion people one the earth, which means that we would expect eight people from around the globe to have made ten sequential right predictions once in their life, just by accident. The fact that one of those eight people is right in front of us and seems to be able to do it on command is pause for thought. Trying to broaden the scope to something else, like 'the history of the world' would render all of our calculations meaningless.

Scope is like the denominator in a ratio; you can't just change it at whim to make your ratio look better.

The fact that this one person accomplished this is good evidence that something unusual — some force or effect that we are unaware of — is occurring. It does not tell us that this person is correct in asserting God-given special powers; that's a theoretical issue that lies outside the province of experimentation. But since it's clearly not something that would happen under normal assumptions, there must be something off in our normal assumptions.

Answer 2

Let's say you are trying to determine whether any Tasmanian Devils still exist on a particular remote island.

There might be a range of information (various 'evidences') which might contribute to any conclusion you make. For example:

1. One sample of Devil scat (droppings) as recent as three days old,
2. A Devil skeleton (as recent as three days old),
3. Possible Devil dens (without ambiguous signs of recent habitation),
4. A lack of typical Devil prey,
5. An abundance of atypical Devil prey (and so on).

If you take (4) in isolation, you might be led to decide that there is insufficient food available to sustain a Devil population.

If you take (3) and/or (5) into consideration, you might decide you have insufficient evidence to draw a conclusion.

If you take (1) and (5) in isolation, you might decide it is probable at least one Devil remains extant.

Without taking (2) into consideration, you might not consider that there was recent Devil habitation, but the last Devil may have died 3 days prior.

If you take all into consideration, you have a greater data set from which to draw a conclusion. It might be insufficient to provide you with certainty, but it remains a more complete picture from which you are more likely (not necessarily) to draw a correct conclusion (including that there is insufficient evidence to draw a reliable conclusion) than had you limited yourself to one piece of evidence.

Examination of all the available evidence should be considered mandatory, a principle, in any discipline in which the task arises to reliably maximise your chances over time of arriving at conclusions which are accurate reflections of reality.

In the absence of any other evidence, nothing about the presence or lack of presence of Tasmanian Devils on an island (no matter how unlikely their presence may be) provides you any insight into whether a god for which you have no definition and no evidence might exist. Until you define a god, you have no idea what might constitute evidence for such a god.

It's an interesting question to ask yourself:

"What would convince me of the existence of a god?".

The answer might be more elusive than you expect.