Related to: Reasoning Isn't About Logic, It's About Arguing; It is OK to Publicly Make a Mistake and Change Your Mind.

Examples of being wrong

A year ago, in arguments or in thought, I would often:

  • avoid criticizing my own thought processes or decisions when discussing why my startup failed
  • overstate my expertise on a topic (how to design a program written in assembly language), then have to quickly justify a position and defend it based on limited knowledge and cached thoughts, rather than admitting "I don't know"
  • defend a position (whether doing an MBA is worthwhile) based on the "common wisdom" of a group I identify with, without any actual knowledge, or having thought through it at all
  • defend a position (whether a piece of artwork was good or bad) because of a desire for internal consistency (I argued it was good once, so felt I had to justify that position)
  • defend a political or philosophical position (libertarianism) which seemed attractive, based on cached thoughts or cached selves rather than actual reasoning
  • defend a position ("cashiers like it when I fish for coins to make a round amount of change"), hear a very convincing argument for its opposite ("it takes up their time, other customers are waiting, and they're better at making change than you"), but continue arguing for the original position. In this scenario, I actually updated -- thereafter, I didn't fish for coins in my wallet anymore -- but still didn't admit it in the original argument.
  • defend a policy ("I should avoid albacore tuna") even when the basis for that policy (mercury risk) has been countered by factual evidence (in this case, the amount of mercury per can is so small that you would need 10 cans per week to start reading on the scale).
  • provide evidence for a proposition ("I am getting better at poker") where I actually thought it was just luck, but wanted to believe the proposition
  • when someone asked "why did you [do a weird action]?", I would regularly attempt to justify the action in terms of reasons that "made logical sense", rather than admitting that I didn't know why I made a choice, or examining myself to find out why.

Now, I very rarely get into these sorts of situations. If I do, I state out loud: "Oh, I'm rationalizing," or perhaps "You're right," abort that line of thinking, and retreat to analyzing reasons why I emitted such a wrong statement.

We rationalize because we don't like admitting we're wrong. (Is this obvious? Do I need to cite it?) One possible evo-psych explanation: rationalization is an adaptation which improved fitness by making it easier for tribal humans to convince others to their point of view.

Over the last year, I've self-modified to mostly not mind being wrong, and in some cases even enjoy being wrong. I still often start to rationalize, and in some cases get partway through the thought, before noticing the opportunity to correct the error. But when I notice that opportunity, I take it, and get a flood of positive feedback and self-satisfaction as I update my models.

How I learned how to do this

The fishing-for-coins example given above was one which stood out to me retrospectively. Before I read any Less Wrong, I recognized it as an instance where I had updated my policy. But even after I updated, I had a negative affect about the argument because I remembered being wrong, and I wasn't introspective enough to notice and examine the negative affect.

I still believed that you should try to "win" an argument.

Eventually I came across these Sequences posts: The Bottom Line and Rationalization. I recognized them as making an important point; they intuitively seemed like they would explain very much of my own past behavior in arguments. Cognitively, I began to understand that the purpose of an argument was to learn, not to win. But I continued to rationalize in most of the actual arguments I was having, because I didn't know how to recognize rationalization "live".

When applying to the Rationality Boot Camp, one of the questions on the application was to give an instance where you changed a policy. I came up with the fishing-for-coins example, and this time, I had positive feelings when remembering the instance, because of that cognitive update since reading the Sequences. I think this positive affect was me recognizing the pattern of rationalization, and understanding that it was good that I recognized it.

Due to the positive affect, I thought about the fishing-for-coins example some more, and imagined myself into that situation, specifically imagining the desire to rationalize even after my friend gave me that really compelling argument.

Now, I knew what rationalization felt like.

At the Rationality Mega-Camp, one of the sessions was about noticing rationalization in an argument. We practiced actually rationalizing a few positions, then admitting we were rationalizing and actually coming to the right answer. This exercise felt somewhat artificial, but at the very least, it set up a social environment where people will applaud you for recognizing that you were rationalizing, and will sometimes call you out on it. Now, about once a day, I notice that I avoid getting into an argument where I don't have much information, and I notice active rationalization about once every two days.

The other thing we practiced is naming causes, not justifications. We attempt to distinguish between the causes of an action -- why you *really* do something -- and myriad justifications / rationalizations of the action, which are reasons you come up with after the fact for why it made logical sense to do a thing.

How you can learn to recognize rationalization, and love to be wrong

These steps are based mostly on my personal experience. I don't know for sure that they'll work, but I suspect they will.

You'll do this with a close friend or significant other. Ideally they're someone with whom you have had lots of frustrating arguments. It would be even better if it's someone who also wants to learn this skill.

First, read these Sequences: The Bottom Line and Rationalization. Be convinced that being right is desirable, and that coming up with post hoc reasons for something to be true is the opposite of being right: it's seeming right while being wrong; it's lying to yourself and deceiving others. It is very bad. (If you're not convinced of these points, I don't think I can help you any further.)

Next, take 10 minutes to write down memories of arguments you had with people where you didn't come to an agreement by the end. If possible, think of at least one argument with this friend, and at least one argument with someone else.

Next, take 10 minutes to write down instances from your personal life where you think you were probably rationalizing. (You can use the above arguments as examples of this, or come up with new examples.) Imagine these instances in as much explicit detail as possible.

Next, tell your friend about one of these instances. Describe how you were rationalizing, specifically what arguments you were using and why they were post-hoc justifications. Have your friend give you a hug, or high-five or something, to give a positive affect to the situation and condition yourself.

This step is optional, but it seems like it will often help: actually work out the true causes of your behavior, and admit them to your friend. It's OK to admit to status-seeking behavior, or self-serving behavior. Remember, this is your close friend and they've agreed to do the exercise with you. They will think more of you after you admit your true causes, because it will benefit them for you to be more introspective. Again with the hug or high-five.

Next, rehearse these statements, and apply them to your daily life:

  • "When I notice I'm about to get into an argument, remind myself about rationalizing."
  • "When I notice illogical behavior in myself, figure out its true causes."
  • "When someone else states a position, ask myself if they might be rationalizing."
  • "When I learn something which doesn't make sense, say 'I notice I am confused' out loud."
  • "When someone else seems upset in an argument, ask myself if they might be rationalizing."
  • "When I notice rationalization in myself, say 'I was rationalizing' out loud."
  • "When I notice I've updated, say 'I was wrong' out loud."
  • "When I say 'I was rationalizing', ask for a high five or give myself a high five."
  • "When I say 'I was wrong', ask for a high five or give myself a high five."

At the very least, read these out loud to your partner. If you want to go further, you could try using Anki to learn these statements by heart.

And let me know in the comments how it goes.

New Comment
13 comments, sorted by Click to highlight new comments since:

Just to be pedantic: Enjoying being wrong probably not good.

Enjoying having been wrong, and now being (potentially) less wrong is good.

But the latter doesn't make as good a title :D.

Enjoying having been wrong, and now being (potentially) less wrong is good.

Better still: enjoying having been able to notice whether you were right or wrong.


If I may suggest replacing certain Less Wrongish jargon with more everyday language:

Over the last year, I've self-modified to mostly not mind being wrong

How about "I've changed myself to . . ."?

I had a negative affect about the argument

Consider "I felt bad about . . .".

Note that the phrase "feel bad about" is often understood to imply guilt, specifically, rather than negative affect in general.

True. Something more specific, like:

"Recalling the argument made me cringe."

(or whatever the actual conditions and type of feeling were). Both "feel bad about" and the original are vague.


I like Anna's perspective that being wrong is like admitting that you're smarter now than you were in the past. It helps me be willing to see my beliefs as things I used to believe and haven't rethought about from my new, improved perspective, and remember to try and update when appropriate.


I downvoted the article not because I think it is bad, per se -- even though you're other-optimizing -- but because I want to see less of this genre of writing on LW in general.

Can you elaborate (or link to an elaboration) of what you would like to reduce specifically and why?

you're other-optimizing

less of this genre of writing

While I agree with the first, I don't see how the second follows. Would an adjustment in delivery to be more like "These are methods for solving problem X that worked for me, in case you hadn't considered attempting something similar in solving X for yourself." be more acceptable?

Unless you're against the personal-self-help-story sort of writing in its entirety for other reasons?

I guess I'm just asking for an elaboration on why you wouldn't want to see this sort of writing.

ETA: Or... exactly what jsalvatier just said.


"These are methods for solving problem X that worked for me, in case you hadn't considered attempting something similar in solving X for yourself."

No, it's not the delivery that I take issue with. It's the unintended consequences. See this.

If LW must tackle self-help, I want to see meta-analyses of published science. I want sample sizes bigger than one. If a writer feels compelled to write down "what worked for them", then in the name of Bayes, at least set up an ad-hoc internet survey first. Gather some data about how people actually are, not how you are. Because you don't even know how you are.

That's the center of the other-optimizing problem.

I agree that the 3rd part of the article was a self-help style, something that only works for those who happen to self-optimize in the same way the author does. This is likely a small percent of the readers, but apparently large enough to provide glowing testimonials for published books.

However, the first two parts are simply a relevant personal experience to share, potentially interesting and maybe even useful for some readers and so worthy of a post, especially if it was (less moralizingly) named along the lines of "How I learned to stop worrying and enjoy being wrong, YMMV".

I'd also like to see some tangible benefits reported at the end of the article (otherwise, what's the point of trying to be rational?), but that's just me, not going to try to other-optimize.

I reread that post and now, I understand. I think I've been making this mistake a lot recently. I think 'other optimizing' is not a very descriptive name, though.

While I agree with the first, I don't see how the second follows.

If they did follow paper-machine would have been making an error. The grammatical structure is "not because (), but because()".