Friday, June 13, 2014

A shorter discussion of evidence

About a week and a half ago, I finished and published a ridiculously long essay about what constitutes evidence, totalling more than 10,000 words, because I felt like there was a whole lot of ground to cover to shore up the many things I was trying to address and make sense of given the topic. The original essay about evidence can be read here. My purpose here is to give a shorter, summarized version of that very long discussion.

How I think we should use the word evidence

In an effort to blend the lay uses of the word "evidence" with the more careful scientific use of the word, I'm offering the following as a prototype for a philosophical characterization of the term, hoping to replace what I feel are far worse interpretations of the word (e.g. "any observation that increases the probability that a hypothesis is true," which is broken to pieces and very misleading).

My suggested meaning for the term "evidence" is
A body of observations O is evidence for a hypothesis H if, and only if, it is a consistent part of a larger body of observations called evidential closure of O, comprised of all observations bearing significantly upon H, such that the probability that H is true given O (plus its evidential closure) is sufficiently great to warrant justified belief that H is true. In this case, we could call an observation A in O an evidential observation.
The key points to note about this characterization of the term "evidence" are:
  • "Evidence" refers not to individual observations but rather to a body of observations;
  • A body of observations only constitutes "evidence" for a hypothesis if that hypothesis is actually (provisionally) true (I will elaborate upon this below);
  • A body of observations cannot constitute "evidence" if it is cherry picked from a broader range of observations that collectively fail to render (provisional, qualified) acceptance of a hypothesis;
  • What we call "evidence" can be qualified by the statistical confidence with which we are claiming provisional acceptance of the hypothesis (i.e. O is evidence for accepting H at the 95% confidence level);
  • There is no evidence for anything we have really good reasons to think is false.
Understanding the idealization

To really get a keen idea of what I'm going for with this definition, it is helpful to examine its idealized expression, the kind of expression we would use if we had some way to apprehend clearly the truth instead of the provisional truths we're often obliged to rely upon.

Idealizing this definition would read something like this:
An observation A is evidential for a hypothesis H if it raises the probability that we're right to think H is true and H is actually true.
The idea here is to mirror the Platonic definition of knowledge: justified true belief. As I'm putting it, evidence supports knowledge, and knowledge requires that the belief in question is true. Thus, there is no evidence for anything false. That is, knowledge has to accurately reflect reality.

Evidence for a hypothesis?

The phrase "evidence for a hypothesis" is a common abuse of terminology. In reality, we have observations that are consistent with or not consistent with various hypotheses. Hypotheses represent a very specific kind of abstract construction by which we attempt to understand the world, and they should not be confused with the reality that we are attempting to describe. When I use the phrase "a body of observations is evidence for a hypothesis...," what I mean is that the given body of observations is consistent enough with a hypothesis that is sufficient for us to determine at some reasonable level of confidence that the hypothesis is provisionally true.

Range of applicability

I don't want to go overboard with this section, but it is important to realize that "evidence," like the hypotheses it supports, necessarily has a range of applicability. Many of our everyday observations constitute splendid evidence for Newtonian mechanics--which is to say that Newtonian mechanics will give spectacularly accurate predictions for the phenomena we're observing--even though we know that Newtonian mechanics isn't the whole story. On its range of applicability, which can be delimited with regard to special and general relativity and quantum mechanics, though, Newtonian mechanics is provisionally true with very high confidence, and thus we can call the body of observations in that range "evidence" for Newtonian mechanics (on that range).

The original

Once again, I encourage you, O Reader, to give the longer essay on evidence a look if you want to understand my thoughts and motivations better--and I do think they're important, which is why I  put the time into a 10,000-word essay on the topic in the first place. This is the TL;DR summary of the original. 


  1. I'm sure I will be linking to this in the future -- I'm very thankful for having a resource to which I can refer, as I think that defining "evidence" is going to be one of the most productive ways and efficient ways to cut down on a lot of time-wasting religious nonsense.

    I'll try and think of some other ways to make some of the language shorter and "friendlier" as well. You've done a great job of shortening your earlier blog-treatise, but I think that something as important as all this needs to have an "elevator version," as well as probably some rhetorical sound bite versions. Tbd.

    Also, I thought one of the more interesting observations in the great treatise version was the brief handling of the common conception of "evidence" coming from exposure to the term in legal proceedings -- an unfortunate fact, because this seems to reinforce the poor philosophical definition. I think that that is just one more way lawyers, like metaphysicians, have f'd everything up for the rest of us. :)

    1. Thanks, Cal.

      You're welcome to run with an elevator version and sound bites. My favorite is this one: "There is no evidence for anything false, only observations that have been misattributed to a bad idea." The more on-point version is: "There is no evidence for God, fairies, or any other imaginary thing, only evidence/observations that has been misattributed to them."

      If you write an "elevator version" and don't have anywhere to put it, I'll be glad to post it here with full credit if you'd like.

      The thing with law and lawyers is its strong reliance upon tradition. A great deal of our legal language and methodology goes way back now, and so a less careful (pre-scientific, in fact) conception of "evidence" became the standard and tradition there. Habit, pomp, and circumstance, and probably not worth fussing about too much.

      (By the bye, I don't have any interest whatsoever in overturning or even significantly modifying the lay use of the term. I don't think such efforts are terribly productive.)