Faith, Trust, and Evidence

A Question of Definition

In #Unbelievable Project

I was listening to the Unbelievable podcast debate between Tim McGrew and Peter Boghossian on the definition of faith and whether or not faith is a valid way of knowing. In this discussion, Boghossian's main point - that "faith is belief without evidence" or "pretending to know things you do not know" - is derailed quite early and he never truly addresses the counter definition other than to claim that few people use it. The discussion then devolved into a back and forth with both sides claiming that "all the people I know use my definition", and got nowhere. Let's look at these definitions, and see where it goes. {.me}

The Definitions

Peter - "Belief without evidence."{.aq}

Peter - "Pretending to know things you do not know."{.aq}

OED - "complete trust or confidence in someone or something"

Tim says that very few Christians (less than 1%) would use Peter's definition{.theist}

I agree that no Christian would articulate this definition of faith, however they may be functionally using it, which we will address later. Peter really should have said that.{.me}

Tim - "Faith is trusting in, holding to, and acting on what one has good reason to believe is true in the face of difficulties. The difficulties may be where you have to take an action where the outcome is beyond your control."{.tq}

The example Tim gives is jumping out of an airplane, where you have faith in your instructor that he packed your parachute properly. Your act of jumping makes faith more than simply hope (if you just hoped your instructor packed it, you wouldn't jump), and the decision is made in the face of evidence, not despite it or without it.{.me}

When Peter asks Tim what people mean when they accuse someone of having "faith in evolution", Tim responds {.atheist}

You're trusting in something that you cannot completely verify because it doesn't lie open to your senses. {.tq}

When Peter asks Tim what people mean when they say "I don't have enough faith to be an atheist", Tim responds{.atheist}

Belief in something in the face of certain difficulties, where the weight of the difficulties is greater on one side compared to the other.{.tq}

When Peter asks Tim "Why don't we say that we have faith in the existence of chickens?", Tim responds{.atheist}

We are venturing nothing on the existence of chickens. When I believe that chickens exist and I act on that belief I am not taking any step that places outcomes I care about beyond my direct control. [In the case of religion], people are placing the outcome of their eternal soul out of their control. They are taking a risk where the outcomes matter. The decision itself is evidenced but the outcome is uncertain.{.tq}

When Peter asks Tim "Do you have faith or evidence that Islam is false?", Tim responds{.atheist}

Why would I use the word "faith" when I am venturing nothing on Islam? I am a little bit confused about the framing of the question that way. I think I have evidence that it is false, but since I am not venturing on Islam, I'm not sure why the word faith would come in. {.tq}

Addressing the Definitions

I've said this before, but it bears repeating. The word faith conflates two related terms - probability and utility. Probability is just the value we assign to the strength of belief, and utility is the value we assign as the reward (or penalty) we receive when taking an action. They both have well-defined mathematical relationships and procedures (e.g. Bayes theorem, decision theory, etc...) A rational actor should try to maximize expected utility. From Wikipedia on Decision Theory,{.me}

The idea of expected value is that, when faced with a number of actions, each of which could give rise to more than one possible outcome with different probabilities, the rational procedure is to identify all possible outcomes, determine their values (positive or negative) and the probabilities that will result from each course of action, and multiply the two to give an expected value. The action to be chosen should be the one that gives rise to the highest total expected value.{.me}

Clearly claims using expected utility require that probability assignments have already been made, so claims of utility must necessarily be probability claims as well. When translated into these more precise terms, both Tim's and Peter's claims begin to make more sense. It will also show that Tim is in fact using the definition, in some cases, like Peter and is therefore inconsistent.{.me}

You have faith in your instructor that he packed your parachute properly. Your act of jumping makes faith more than simply hope (if you just hoped your instructor packed it, you wouldn't jump), and the decision is made in the face of evidence, not despite it or without it.{.tq}

The equations are:{.me}

  • \(\langle U(J) \rangle = U_I \times P(K_I) + U_B \times P(K_B)\)
  • \(\langle U(N) \rangle = V_I \times P(K_I) + V_B \times P(K_B)\)

with the notation{.me}

  • \(K_I\) is the proposition "my instructor packed my parachute"{.me}
  • \(K_B\) is the proposition "Peter Boghossian packed my parachute"{.me}
  • \(U(J)\) is the utility gained from jumping{.me}
  • \(U(N)\) is the utility gained from not jumping{.me}
  • \(U_I\) (\(V_I\)) is the utility gained if you jump (don't jump) and the instructor packed the parachute{.me}
  • \(U_B\) (\(V_B\)) is the utility gained if you jump (don't jump) and Peter Boghossian packed the parachute{.me}

We have the following approximate values for the variables:{.me}

  • \(P(K_I)\sim 1\) (nearly certain the instructor packed the parachute){.me}
  • \(P(K_B)\sim 0\) (nearly certain that Peter didn't pack the parachute){.me}
  • \(U_I\gg 1\) (good benefit from jumping, with instructor packing the parachute){.me}
  • \(U_B\ll 0\) (very bad penalty for jumping, with Peter packing the parachute){.me}
  • \(V_B\sim V_I \sim 0\) (neutral gain for not jumping in either case){.me}

Notice, for Tim to have "faith in his instructor", two things must be true:{.me}

  1. \(P(K_I)\sim 1\) (nearly certain the instructor packed the parachute){.me}
  2. \(U_I\gg 1\) (good benefit from jumping, with instructor packing){.me}

Tim wants to focus on point (2), while Peter wants to focus on point (1). Once seen this way, it is very easy to understand the responses. {.me}

Why don't we say that we have faith in the existence of chickens?{.atheist}

Because \(U_{\rm chickens}\sim V_{\rm chickens}\sim W_{\rm chickens}\sim \cdots\sim 0\), for all choices with utilities \(U, V, W, \cdots\) even if we are confident that chickens exist (i.e. \(P({\rm chickens})\sim 1\)). {.me}

Do you have faith or evidence that Islam is false?{.atheist}

Tim has evidence that Islam is false, \(P(I|{\rm data})\ll 1\), but is not venturing anything on Islam (or more accurately, on his choice to not follow Islam), \(U_{\rm not-Islam}\sim 0\). Again, Peter sees the first part, yet ignores the second part. {.me}

Not all is well with Tim

Going back to this little exchange, we note that Tim is also using Peter's definition:{.me}

When Peter asks Tim what people mean when they accuse someone of having "faith in evolution", Tim responds {.atheist}

You're trusting in something that you cannot completely verify because it doesn't lie open to your senses. {.tq}

Now I can think of no way to understand this statement from the perspective of Tim's definition:{.me}

"When I act on that belief I am taking some step that places outcomes I care about beyond my direct control"{.tq}

What outcomes are you placing beyond your control believing in evolution? What obvious utility are you weighing in this case? As far as I can see there is none, and so faith is in fact being used here as "belief without sufficient evidence". {.me}

Priors and Faith

I think this brings in another aspect of faith, which I believe applies to all of the cases explored so far, and that is that faith is used only in contexts with low prior probability. In this conversation, they spoke of faith in the context of the supernatural, extreme activities (i.e. jumping out of planes), events beyond our immediate senses - all of which coincide with lower prior probability, and need more evidence than is typical to overcome them. They may, or may not, also have high utility. We don't have faith in the existence of chickens because the existence of chickens has high prior probability. {.me}

The Empirical Claim

I'll have to address the empirical question Tim raises in another post. Is it empirically true, as Tim claims, that less than 1% of the people use the term faith in the way that Peter defines it (i.e., belief without [sufficient] evidence)? {.me}

Even in Tim's case, however, I have found he doesn't use it all the time in the way he claims, and reverts to Peter's definition when convenient. That is the problem with the word - since it is referring to two distinct components, the apologist can easily switch between them without even noticing themselves. That's why I don't like using the word, and simply try to address the question, "what do you believe and why?".{.me}

Conclusions

All in all, Peter does a lousy job and seemed completely unprepared to deal with this rather common understanding of the word faith. He also insisted on using the "belief without evidence" instead of the more honest "belief without sufficient evidence". When people say there is no evidence for something (like God, UFOs, astrology, psychic phenomena, etc...), they really mean that there is terrible evidence for something. There is some evidence for astrology, it's just that it is terrible evidence. The exaggerated, more simple, phrase of "belief without evidence" is counterproductive, especially when the more accurate phrase, "belief without sufficient evidence", is nearly as simple. {.me}

Peter does say that by claiming faith, one is "assigning a confidence value to a belief higher than what is warranted by the evidence." This doesn't roll off the tongue quite as well as "pretending to know things you do not know", and he should have made clear the distinction between the precise language and the slightly hyperbolic simplifications.{.me}

I also feel that words like faith are so slippery as to be useless, and in fact dangerous in a debate or discussion. It is too easy to slide from probability to utility and back when using the word faith. Although I might advise one to stick to the math, I think the equations are just more concise representations of the statements that could be made in plain English. {.me}