Saturday, June 11, 2016

Reppert's XYZ Question

Victor Reppert presents a challenge to everyone concerning the question of what constitutes evidence.  He states it this way:
It goes like this. X is evidence for Y just in case Z.

I've answered the question here at various times, and people are dissatisfied with my answer. Fine. I want to know your XYZ answer. If you are going to tell me I don't have any evidence, then apparently you have a different answer to the XYZ question than I do. But when I ask people what their answer is, I never find out.
Presumably, he wants people to provide some substitution for Z such that it makes a reasonable definition of 'evidence'.  Now this is a topic that he has written about before in his article Victor Reppert on the No Evidence Charge.  His own answer is "X is evidence for Y just in case X is more likely to exist given Y than given not-Y."  This may sound reasonable on its face, but it leads people down the wrong path before they ever have a chance to address the question of "What is evidence?"

I addressed Victor's earlier article in my own post, called Evidence For Dummies, and obviously Victor has ignored what I had to say about it, if he ever saw it.  In the comments to his current post, Cal has said pretty much the same thing as I did, and I'll wager Victor will ignore that, as well.  Why?  I can only speculate that by sticking to an unscientific approach to defining evidence, he can continue to justify his superstitious beliefs.  With a more scientific approach to looking at evidence, the epistemic justification for belief in God and all things supernatural is greatly diminished.

To put it briefly, it is not really correct to say that there is evidence for something.  It is more correct to say that evidence is what we can observe as a result of the real state of affairs, whatever that state of affairs might be.  We might have various hypotheses about what that state of affairs is, but it is wrong to take a single observed fact and say that this is evidence for a specific hypothesis.  The scientific approach is to gather all the available observed facts (or evidence) and produce a hypothesis that does the best job of explaining everything that is in evidence.

Let's examine Victor's approach more carefully.  He says, for example, "the evident fact that there are thinking beings in our universe is evidence that God exists."  Using his formulation of the definition of evidence, I might agree that the existence of thinking beings is more likely given the existence of God.  but I can't agree that the existence of thinking beings constitutes evidence for God.  Why not?  Because by looking at all the evidence as a whole, I think it is most likely that thinking minds arose by natural evolution.

What if we postulated that there is a supernatural being (not God) who picks up water from the ocean in a big bucket and deposits it into the clouds to produce rain.  Let's call this being the Waterlifter.  We could use Victor's approach to defining evidence, and say that rain is evidence for the Waterlifter, because it may well be true that rain is more likely to exist if the Waterlifter exists than if he doesn't exist.  But that's just plain stupid.  We know why rain exists, and we know that there is no Waterlifter.  There are plenty of other facts in evidence that give us a better picture of how water actually finds its way into the clouds.  When we look at all the evidence as a whole, a more realistic picture emerges.  The evidence does not tell us that there is a Waterlifter.  In truth, rain is not evidence for a Waterlifter, even though it complies with Victor's definition of evidence. 

The problem is that Victor is using a bad definition for evidence - one that allows him to justify all kinds of beliefs that are not supported by real evidence.  To isolate a single fact and call it evidence for some particular belief does not help us to understand the actual state of affairs that results in the complete body of evidence we observe.  We have to look at all of the evidence.  Failure to do so is unscientific, and leads to conclusions that can be contradicted by other facts that have not been considered.  This is the favored approach of religionists.  Take isolated facts and call them "evidence" for beliefs that can't be justified when all that facts are taken into account.

As long as people like Victor continue to take an unscientific approach to looking at evidence, they will go on claiming they have evidence on their side.  They are using parlor tricks - playing games with the facts - as a means of fooling themselves into believing that their beliefs are justified.  But they're not fooling everyone.


  1. if he ever saw it

    I doubt seriously that Victor reads your blog. Just a hunch.

    He's never commented here, nor has he ever addressed anything written here. As far as I can tell, not once.

    I count that as evidence that he doesn't visit this site.

    1. I count as evidence the fact that he has commented on several occasions, including as recently as two weeks ago.

      Furthermore, he has made his own posts specifically in response to mine.

  2. I've learned that when apologists say, "There is evidence that God exists," what they usually mean is more like:

    There is evidence (under a certain definition of "evidence") that God (for a certain definition of "God") exists (for a certain definition of "exists").

    And you mustn't make the naive assumption that any of these words are being used in the commonly-understood way. No, it isn't evidence in the courtroom sense; no, it isn't the jealous, angry God of the Septuagint but rather an ethereal "necessary being;" and no, this being doesn't exist as a physical phenomenon but "exists necessarily" or "exists metaphysically." which point the claim that "God exists" has been so thoroughly divorced from human experience and so completely evicted from the observable universe that, frankly, it's difficult to care.

    1. Excellent commentary Adam. You have eruditely captured the definitional post hoc rationalisation around which much of the contemporary Christian apologetical process is founded.

  3. I simply could not believe what my eyes were reading on Victor's blog about the purported XYZ of evidence he recently put up. It is truly a sad indictment how an intelligent person has lowered one's expectations to such an extent, reduced to sustaining a fool's paradise when the many years of academic training and scholarship should have inured him against such folly. His God belief is a proposition in desperate search of matching 'evidence'. And the best he can offer in search of that evidence is to redefine the word.

    1. I don't know if he made that definition himself, or if it is a standard one used by theists, but it seems obviously contrived to meet the needs of theists.

  4. I just saw J J Lowder's comment under Evidence for theism from the existence of thinking beings (cited in my OP). In defending Victor, he says "This is such a basic example of how to think about evidence in a philosophical sense it is, well, almost a textbook example of how to think about evidence." And then in his next comment, he refutes himself.

    It's a textbook example of how PoR muddles your thinking.

    1. Lowder did not refute himself. Lowder's two posts are logically compatible based on how he interprets the term "evidence".

      Definition of "evidence" - For any hypothesis H and its negation ~H, some data D is evidence for H over ~H when the probability of D given H is greater than the probability of D given ~H. Or in other words, D is evidence for H if Pr(H|D) > Pr(H).

      Here is Lowder's first point:

      M = a mental something exists.
      T = theism.
      N = naturalism.

      Pr(M|T) = 1 whereas Pr(M|N) < 1, so Pr(M|T) > Pr(M|N). As a result, M is evidence for T over N.

      Lowder's second comment asserts two things:


      P = physical things exist.

      Pr(P|N) = 1 whereas Pr(P|T) < 1, so Pr(P|N) > Pr(P|T). As a result, P is evidence for N over T.


      The fact that P is evidence for N cancels out the fact that M is evidence for T unless the weight of evidence from M for T is greater than the weight of evidence for P for N.

      There is nothing contradictory about what Lowder said. By his definition of evidence, a very common one among philosophers of science, proper subsets of our data can be evidence for a proposition even though the full set on its own is not. As a result, claiming there is evidence for a proposition and against is not a contradictory position based on how "evidence" is being interpreted. This is standard among Bayesian confirmation theorists.

      Lowder's use of "Evidence" is not from PoR. It is very common in Bayesian circles. It is actually a fascinating concept:

    2. I won't say the Bayesian concept of evidence necessarily corresponds to how "evidence" is used within science, but it is popular among philosophers of science, and popular among many scientists (especially scientists working in physics).

      From philosophers of science/logic, people should read this:

      From a physicist, people should read this:*Version*=1&*entries*=0

      Both books can be found online if you look in the right spots. If either book seems too complicated to understand, well that sucks because science is complicated to understand.

    3. First, I want to thank you for your inputs. I appreciate the things you have to say

      Lowder did not refute himself in the sense of presenting logical arguments that are contradictory. But in a very real sense, he refuted his own statement that this is how to look at evidence. When he presents two arguments that each use isolated evidential facts (while ignoring the greater body of evidence that exists) to reach opposite conclusions that supposedly cancel each other out - this should serve as a strong indicator that this approach to the use of evidence is fundamentally wrong. The evidence always supports the true state of affairs, but it must be viewed properly.

      From the Stanford article: "In order to be justified in believing some proposition then, it is not enough that that proposition be well-supported by some proper subset of one's total evidence; rather, what is relevant is how well-supported the proposition is by one's total evidence."

      This seems to be in complete agreement with what I have said. But what we see so often in PoR is a refusal to look at the total body of evidence. Bayesian analysis can be abused in this manner, and that's what we see time and time again.

      If either book seems too complicated to understand, well that sucks because science is complicated to understand.
      - I don't want to get unpleasant, but please spare me your condescension. I'll wager I have studied far more mathematics than you - not to mention science.

    4. Lowder has now said that Reppert commits what he calls the "fallacy of understated evidence". So it appears that he also agrees that the total body of evidence must be considered. I suppose I may have misread his intent when he presented his argument that "cancels out" Victor's argument based on an isolated fact of evidence. But I still think it's wrong to say that this is the right way to view evidence. I would agree to that only in a limited sense - when examining an individual fact as part of a broader examination of all the relevant facts.

    5. Yes, Jeff and Bayesians think our beliefs should be proportioned to the probability of propositions given our total known relevant data. So there can be some evidence for some H over some ~H, but our belief that H is true must be equal to or no greater then the probability of H given our entire known relevant data (And of course our prior probabilities).

      When examining specific elements of data sets, we could use a different term than "evidence". Basically we would limit our universe of discourse for "evidence" to entire sets of data rather than proper subsets. The only downside is that this might mislead people into thinking that there is no data at all which might support a proposition that is probably false given our entire relevant data. i.e. even if H is made unlikely given D whereas ~H is made to be likely, a person might be led to believe that no data supports H over ~H if we claim that there is no evidence for H. A simple change in natural language would theoretically fix that confusion.

      And yes my comment was condescending which it should not have been. I'm sometimes driven crazy by fellow atheists online when they act as if they are champions of science yet don't understand high school algebra. I'm not saying this happened here though. I don't know how many dozens of times I have made comments using probability/statistics to back up a point and fellow atheists have said my point is obscure gibberish despite probability/statistics being one of the languages of science.

    6. There is no question that probability calculations can be tricky. What is the complete space of possible outcomes? Are they all equally likely? Are there dependencies? Often, people make assumptions that aren't justified, and in this way they make baseless probability claims.

      For example, when we consider the question of fine-tuning of physical constants, we have absolutely no basis for claiming that there is a large number of possibilities, or that the variables are all independent. We simply don't know those things.