Something bugs me about the idea of “rationality”. I don’t think truth is best thought of as an absolute concept. In practice, the way people determine truth varies depending on the sitation. Pay attention to yourself over the course of a day or a few hours. How many different methods do you use to conclude that something is true? I am going to try to categorize some.

Experiential Truth

The back of my head itches right now. Do I have any doubt about that? Am I going to engage in debate? No. I am absolutely confident without a doubt that the back of my head itches, because I feel it directly.

Experiential truth is great for what it does, but it can’t really be communicated. When I tell someone else that my head itches, I could be lying. There is no way for them to know the difference. The domain is also pretty limited to the things you can directly observe.

Mathematical Truth

Mathematical truth is extremely nitpicky about not permitting any errors. If a line of reasoning works in 99.9999% of cases, it is not good enough to use in a mathematical argument. If there is a single flaw in an argument, the whole argument must be thrown out. Mathematicians like to call these arguments “proofs” to distinguish them from the more-human sorts of arguments.

Mathematical truth isn’t very useful in everyday life. There just isn’t anything you can say mathematically about, for example, apples. Do they have a certain weight or color? Are they mostly red or green? Well, maybe in a million years all apples will be blue. Why is that impossible? Since that could happen, we can’t really say anything at all mathematically about the color of apples.

As a software engineer, mathematical truth is occasionally useful. You can prove mathematically that a certain algorithm will take exponential time. But it’s pretty limited by itself.

Scientific Truth

Scientific truth is what you get when you take mathematics and you add a small willingness to accept mistakes. If something is true the vast majority of the time, that’s usually okay in science.

The scientific method isn’t precisely how scientists work, but it’s roughly the right idea. You construct a hypothesis, test the hypothesis with some experiment, gather data, and if the vast majority of the data supports your hypothesis, it’s good scientific evidence.

Scientific truth is still not that useful in everyday life. What can I say scientifically about apples? I am not going to spend time gathering data on apples.

Often scientific truth builds on mathematical truth. Science teaches us laws of physics, and we can apply those laws of physics to many different situations using mathematics. It doesn’t work the other way around, though. A true mathematical statement is also true by the standards of scientific truth. But a true scientific statement isn’t necessarily true by the standards of mathematical truth.

How much error is acceptable for something to be a scientific truth? There is a tradeoff. Physics experiments require a very high confidence, and in return it is pretty rare that mistakes are found in accepted physics. Social sciences like psychology are on the other end of the spectrum. It is quite common for multiple psychology papers to come out that contradict each other, and the state of the art is uncertain which to accept. Chemistry, biology, and medicine are somewhere in between.

Rhetorical Truth

Rhetorical truth is something that you believe because you heard a convincing argument for it. These arguments don’t necessarily involve data and statistics. When they do involve data, they are often combining a statistical fact about the data with a more humanistic conclusion that could be derived from the fact.

The power of rhetorical truth is that it can operate in basically any human domain. Think of public policy, economics, ethics, or history. When it’s hard to run an experiment and gather data, the scientific method doesn’t really work.

The downside of rhetorical truth is that it can be proven wrong more often than scientific truth. When you read two opposing scientific papers, you can usually drill down and figure out an experiment that will determine which of the two is correct, and use that to convince other people as well. When you read two opposing pieces of rhetoric, you can come to a conclusion about which one you find more convincing, but some people might come to a different conclusion, and there isn’t necessarily anything that will prevent that.

Rhetorical truth can build on scientific and mathematical truth. When I believe in global warming, it’s not because I myself measured the amount of carbon in the atmosphere. And it isn’t because some people ran an experiment where on some planets they put carbon dioxide in the atmosphere and on the control group they kept the atmosphere as is. It’s because I read some convincing arguments that used scientific analyses as their source material.

Memetic Truth

I wasn’t sure whether to call this “mimetic truth” a la Girard or “memetic truth” a la Dawkins. I went with “memetic” because that seems more associated with “truth”, and “mimetic” seems more associated with “desire”. It is basically the same concept, though. Memetic truth is something you believe because other people also believe in it.

Your instinct might be to think, oh yuck that’s bad, that’s how you get big groups of people thinking stupid things. Well, maybe. But first you should appreciate that memetic truth is massively useful in everyday life. Memetic truth is the main form of truth that lets us live our lives as intelligent human beings.

How much science do you do yourself? Some of you are scientists and the answer will be nonzero. But the vast majority of scientific truth, you can’t get it straight from the scientific method yourself. You need to trust other people. You trust your chemistry teacher without having to redo every experiment to check their claims.

I’m a software engineer. Sometimes computer science will tell us useful things. Like if an algorithm is quadratic, you can often scientifically determine that it will be inefficient in some cases, and then you know not to use it. But mimetic truth is useful in far more cases. Maybe I’m picking which encryption library to use. Do I analyze them all and run tests? That will take forever. I’m better off asking someone who’s done it before and taking their advice.

I also have little kids. The vast majority of things that little kids believe, they don’t believe it because of any sort of argument whatsoever. Little kids are not yet equipped to handle philosophical debate. They believe things because they are copying other people that believe things. That can be believing adults that it’s dangerous to run into the street, or believing other kids that today is Opposite Day.

Memetic truth can build on all the other types of truth. I tell my kids that cigarettes cause cancer, and they believe me, even though they don’t understand any of the research behind it, and really I don’t either, I just believe the medical establishment.

Memetic truth can be proven wrong all the time. Something that seems memetically true in California can easily be memetically false in Texas.

Which one is the best?

There’s a spectrum here. On one end, you can be the most confident in experiential truth and mathematical truth. But those types of truth are not useful in very many areas of human activity.

On the other end of the spectrum, rhetorical truth and memetic truth apply to almost anything. If a question can be phrased in English, there is probably a way to come up with a rhetorical answer. And there is definitely a way to find out what someone else thinks. But you can be the least confident in these forms of truth.

I think that confusion between the different levels of truth accounts for a lot of disagreement about things like, how much should we believe social science and how much should we incorporate the claims of scientists into public policy. But those are probably issues for another blog post.