BBDD: I'm increasingly of the opinion that truth as correspondence to reality is a minority orientation.
PeerGynt: I've spent a lot of energy over the last couple of days trying to come to terms with the implications of this sentence.
Let me give you the basic outlines of what I am thinking. It has been a gusher of explanation and clarity for me.
First consider the basic distinction around here between between epistemic rationality and instrumental rationality.
Epistemic Rationality:
The art of obtaining beliefs that correspond to reality as closely as possible.
But we recognize the broader skill of Instrumental Rationality:
The art of choosing actions that steer the future toward outcomes ranked higher in your preferences. On LW we sometimes refer to this as "winning".
Winning is about more than epistemic rationality, though epistemic rationality can be pretty dang handy.
Second, consider a Truth. What is it? At least some Truths are statements (don't want to deal with algorithmic or model based truth today). Consider Truths as the winning statements, the statements that allow you to do something that "steer the future toward outcomes ranked higher in your preferences". We do lots of things with statements. We repeat them. We use them to generate other statements. We agree or disagree with others who say them. And, sometimes, we use them to more accurately map the world. But only sometimes.
Third, consider the etymology of the word "probability". On reading Ian Hacking's book "The Emergence of Probability: A Philosophical Study of Early Ideas about Probability, Induction and Statistical Inference ", I came on what seemed an odd fact. Once upon a time, probable wasn't about frequencies or likelihood of events, it was about the standing, credibility, and authority of the speaker. We can interpret that as the way they identified the statements with the better numbers, but maybe that's not what it meant to them - it really just meant a quality of the speaker, and instead, through time, people found that it was a more winning way to characterize the meaning in terms of frequency and likelihood. The criteria for choosing the statements you wanted changed.
Fourth, consider Haidt's recent work on moral modalities. http://www.moralfoundations.org/
the theory proposes that several innate and universally available psychological systems are the foundations of “intuitive ethics... The five foundations for which we think the evidence is best are:
Morality has numerous dimensions Care/harm, Fairness/cheating, Loyalty/betrayal, Authority/subversion, Sanctity/degradation,
We think there are several other very good candidates for "foundationhood," especially:
Liberty/oppression
It's interesting to think of these moral modalities as innate, as biological pattern recognizers that evolved and don't need to be learned. The morality of something is how hard it pings these pattern recognizers, and there is wide variation in the pattern of pings between different people - people weight the different modalities very differently.
Finally, think of a Truth detector, a filter between the winning Truths and the losers, as another candidate for evolutionary development, similarly with different Truth modalities as pattern recognizers, and similarly with widely varying weights between people.
What kind of Truth modalities would you expect? Certainly, correspondence with reality would be a good one. But it's not the only one. Spoken by those in power. Spoken by authorities. In consonance with the tribe. With parents. Quieting a disagreeable confusion.
It's not that people have no conception of correspondence with reality. It's that that pattern recognizer just doesn't ping that loudly, and so is drown out by the others when they ping. Certainly, some of those other pattern recognizers don't ping so loud for me, but seem to ping pretty loud for other people.
Years ago, arguing about God with a Christian girl I knew, she said something that just struck me as bizarre. "I just decided my life would be better if I believed in God." What? What does that have to do with anything? That doesn't make it true. That doesn't mean it corresponds to reality.
It's taken me decades to catch up with her. She seemed to have the idea of "Truth as winning statements", but being a fanatic for "truth as correspondence", I just didn't get it.
Would an accurate summary of this be "humans have a generic, intuitive, System 1 Truth-detector that does not distinguish between reality-correspondence, agreeability, tribal signaling, etc, but just assigns +1 Abstract Truth Weight to all of them; distinguishing between the different things that trip this detector is a System 2 operation"? That seems...surprisingly plausible to me. It also seems like something one could test, with whatever it is scientists use to look at brain activity.
Hook a person up to a brain scanner. Give them true and fals...
A couple of days ago, Buybuydandavis wrote the following on Less Wrong:
I've spent a lot of energy over the last couple of days trying to come to terms with the implications of this sentence. While it certainly corresponds with my own observations about many people, the thought that most humans simply reject correspondence to reality as the criterion for truth seems almost too outrageous to take seriously. If upon further reflection I end up truly believing this, it seems that it would be impossible for me to have a discussion about the nature of reality with the great majority of the human race. In other words, if I truly believed this, I would label most people as being too stupid to have a real discussion with.
However, this reaction seems like an instance of a failure mode described by Megan McArdle:
With this background, it seems important to improve my model of people who reject correspondence as the criterion for truth. The obvious first place to look is in academic philosophy. The primary challenger to correspondence theory is called “coherence theory”. If I understand correctly, coherence theory says that a statement is true iff it is logically consistent with “some specified set of sentences”
Coherence is obviously an important concept, which has valuable uses for example in formal systems. It does not capture my idea of what the word “truth” means, but that is purely a semantics issue. I would be willing to cede the word “truth” to the coherence camp if we agreed on a separate word we could use to mean “correspondence to reality”. However, my intuition is that they wouldn't let us to get away with this. I sense that there are people out there who genuinely object to the very idea of discussing whether a sentences correspond to reality.
So it seems I have a couple of options:
1. I can look for empirical evidence that buybuydandavis is wrong, ie that most people accept correspondence to reality as the criterion for truth
2. I can try to convince people to use some other word for correspondence to reality, so they have the necessary semantic machinery to have a real discussion about what reality is like
3. I can accept that most people are unable to have a discussion about the nature of reality
4. I can attempt to steelman the position that truth is something other than correspondence
Option 1 appears unlikely to be true. Option 2 seems unlikely to work. Option 3 seems very unattractive, because it would be very uncomfortable to have discussions that on the surface appear to be about the nature of reality, but which really are about something else, where the precise value of "something else" is unknown to me.
I would therefore be very interested in a steelman of non-correspondence concepts of truth. I think it would be important not only for me, but also for the rationalist community as a group, to get a more accurate model of how non-rationalists think about "truth"