People Being Bad at Logic

I was talking with someone about people not being very good at math and logic and that being relevant to philosophy errors. Also from the conversation they brought up the book Governed By Affect by Michael Pettit. I found something potentially relevant there: a logic error by Pettit. Here’s the first paragraph I read (from early in chapter 5; I saw something familiar in the table of contents and skipped to it):

The famous Linda Problem, first introduced in 1983, exemplified Kahneman and Taversky’s approach to understanding the fallacies of human judgments when the mind largely runs on automatic. They presented participants with the following description of a fictional person: “Linda is 31 years old, single, outspoken and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear demonstrations.” They then asked participants to rank a variety of statements about Linda in the order of their likelihood. Respondents tended to rank the conjunction of attributes “Linda is a bank teller and is active in the feminist movement” as more likely than either attribute appearing in isolation. The conjoined statement appeared to be more representative of her person as a whole, even though statistically speaking, she more likely possessed only one of the attributes.[4] People failed at reasoning by succumbing to quick, appeasing, intuitive shortcuts instead of slowing their thinking sufficiently to contemplate the problem to solve it correctly. [bold added]

Here, Pettit is talking about laypeople being bad at logic. But he made his own mistakes.

Pettit compared having both attributes to possessing “only one” of the attributes. That’s not the correct comparison. He’s comparing

  1. bank teller and feminist (this is the conjunction they’re talking about because it uses “and” to conjoin two attributes)

to

  1. bank teller and not feminist (wrong)

rather than comparing to

  1. bank teller and may or may not be feminist (right)

which was originally presented as

  1. bank teller (means the same thing as 3)

In the correct comparison, the conjunction option (1) is strictly worse (3 is higher probability than 1). The argument is that, since people prefer a strictly worse thing, they must be irrational, biased, or something else bad.

In Pettit’s version, he says having “only” one trait is more likely than having both. But that depends. X&Y can be more likely than X&!Y. By saying “only” one specific thing, in the context of two things, he’s actually comparing two conjunctions. Statement 2 is both an inaccurate description of the research and also not higher probability than statement 1 as Pettit claims.

Here is the original paper for the Linda problem:

They even tried wording it as “Linda is a bank teller whether or not she is active in the feminist movement” and people still thought that was less likely than “Linda is a bank teller and is active in the feminist movement”. Despite this, Pettit somehow thought the alternative was that Linda was only a bank teller (not a feminist), and then he incorrectly judged the comparative probability for that.

The original research involved multiple additional efforts to clarify this (to study participants, not directly to people reading the paper like Pettit). I wonder if Pettit read the paper while he was writing the book, years earlier, or never. But you can’t explain this as just bad research methods and bad memory. Even setting aside Pettit’s incorrect statement about what the original research said, he also made his own incorrect statement about probability. His logic/math error is visible without reading the original paper.

Pettit continues:

Such simple demonstrations of people’s incapacity to reason properly galvinized a wide of range of psychologists by offering them an exciting means of intervening in the most pressing issues of the day. Gone was the layperson functioning as a deliberative chess expert carefully testing hypotheses about the world.[5]

I don’t like it when experts like Pettit talk about the incapacity of laypeople while making their own similar errors. Maybe you shouldn’t ever be condescending, but you at the very least shouldn’t be condescending if you can’t do a lot better than the people you’re disrespecting.

I’m not trying to single out Pettit as worse than his peers. For me, quickly finding pretty basic errors is just a typical experience when reading books and papers by experts. I think it’s a widespread problem.

From your recent How I Write a Lot article:

In my experience, confident people talking about hard topics often actually make a lot of reading and writing errors and that studying grammar and logic could help with.

When I read that, I wondered whether you meant studying formal logic (such as by taking Peikoff’s logic course).

I was skeptical of how valuable formal logic is because I didn’t think that people often make mistakes like affirming the consequent or whatever. So I wasn’t sure whether formal logic would be worth studying at some point. (Though I recognize that informal logical fallacies are quite common.)

But maybe formal fallacies are more widespread than I imagined.

Though I feel like the “only” error that you mentioned is relatively straightforward (at least once it’s pointed out—I almost certainly wouldn’t have noticed it myself). I’m not sure how widespread more complex formal logical fallacies are.

It looks like he also made typos while dunking on laypeople. (Misspelled galvanized and added an extra “of”.)

A ball and a bat cost a total of $1.10. The bat costs $1 more than the ball. How much does the ball cost?

[…] [skipping many paragraphs]

The first time I heard the story of the ball and the bat, it was from a friend who was studying cognitive science at Princeton. She had just read Kahneman’s book and wanted to do the test with me.

Like most people, I gave an instinctive response. I listened to my System 1 without knowing it was called System 1. Without thinking, without doing any calculations, I gave the first answer that popped into my head: “5¢.”

I felt that my answer annoyed my friend but I didn’t immediately know why. She took the time to explain what was up. I was supposed to answer “10¢,” or at least take a few seconds before answering “5¢.” At any rate, there was no way I was supposed to answer “5¢” immediately, without taking the time to think about it. That was simply not allowed. A guy had even won the Nobel Prize for showing it was impossible.

The author is a mathematician and has automatized some math skills such that he could get the correct answer intuitively.

The article is about how it’s possible to train your intuition.

Nothing is counterintuitive by nature—something is only ever counterintuitive temporarily, until you’ve found means to make it intuitive.

This quote fits with this article by Elliot: Rational Confidence and Standards for Knowledge. No topic is inherently difficult or counterintuitive, it’s about your knowledge and skill. A society with better math education could have almost every adult answer the ball and bat question intuitively and get it right.

If people are universal knowledge creators then the variations of skills and personalities people can have are infinite. When psychologist do studies they aren’t really testing human nature, but some specific culture. A possible type of person or society is one that can answer the ball and bat question intuitively and get it right. The same should apply to every single bias that might be claimed is part of human nature.

My numeric intuition isn’t all that remarkable. I wouldn’t be able to do the calculation if the ball and bat together cost $2,734.18 and the bat cost $967.37 more than the ball.

If people were asked this question they wouldn’t answer using their intuition. But with 1 dollar and 10 cents they have faulty confidence in their intuition. They’re not operating on proper knowledge. Judging when your intuition is reliable is an important skill.

A nice quote from the article on error correcting your intuition:

Nothing’s more exciting than a big glaring error: it’s always a sign that I’m not looking at things in the right way, and that it’s possible to see them more clearly. When I’m able to put my finger on an error in my intuition, I know it’s good news, because it means that my mental representations are already in the process of reconfiguring themselves.

I don’t think I would’ve detected the error without being primed either. And I don’t detect as many errors like Elliot does:

Which means I’m not good enough at logic! Otherwise I would’ve detected more like Elliot does. So knowing Elliot’s perspective on the commonality of logic errors, I’m more interested in learning formal logic and automatizing it.

I’ll share my experience reading this.

A ball and a bat cost a total of $1.10. The bat costs $1

I instantly thought the ball costs 10 cents.

more

oh ok. not 10 cents

than the ball. How much does the ball cost?

i had 5 cents by the time i finished reading

Like most people, I gave an instinctive response. I listened to my System 1 without knowing it was called System 1. Without thinking, without doing any calculations, I gave the first answer that popped into my head: “5¢.”

wait what? instinctive = wrong for these problems. so it’s not 5 cents. it must be a trick question. i went back and reviewed for 30 seconds to try to find the trick i missed. didn’t find it. continued reading.

I felt that my answer annoyed my friend but I didn’t immediately know why. She took the time to explain what was up. I was supposed to answer “10¢,” or at least take a few seconds before answering “5¢.”

oh lol

A guy had even won the Nobel Prize for showing it was impossible.

i’m not sure that’s really what his research says but it’s still a funny comment

can y’all not? am i the weird one here? real question.

i imagine the majority of people can’t do this intuitively but can the majority of smart people interested in rationality also not do it? that wouldn’t really surprise me but I don’t know how good people are at this and it would feel kinda risky and potentially condescending for me to conjecture that an individual smart person I’m talking with can’t do it.

I don’t know how effective formal logic study is. I just know I see lots of logic errors. Formal logic only covers some of what I view as logic errors. A lot of errors have multiple parts. Like the “only” issue in the book. Did he make an error with logic operators or an error with understanding the meanings of the words he used? Or I kinda suspect something else was involved besides those two things, like some confusion about the Linda problem itself.

Most logic errors I see don’t require fancy math to analyze.

Also, FYI, if I told the author about this, I can’t predict if he’d concede immediately or argue with me. From my point of view, it could easily go either way.

When people don’t concede immediately, it can be hard to explain it to them. Knowing some formal logic (and grammar and grammar trees) can help with that – it can make it easier to point out the issue in a precise way.

Hahaha ugh. I didn’t notice. I wasn’t close reading that part.

1 Like

No, when I read that the bat costed 1 dollar more than the ball, I immediately thought it was one of “those” problems. A problem with extra steps. Didnt think right away I could just use algebra to figure it out

Edit: im answering to No to the first question of the quote above

Edit: my answer came after intuitively guessing the ball costed 10 cents

1 Like

I had seen a youtube video (by Veritasium I think, but I couldn’t find it) about the question long ago so I knew the answer before hand. I don’t remember what I thought but I guess that I got it wrong.

Disclaimer: I don’t think I necessarily have a clear point here, I mostly just started thinking about this Linda problem and trying to understand what’s happening in it and wrote down what I was thinking. Definitely half baked.

This immediately jumped out at me when I read your quote. I immediately thought that either he was misrepresenting it, or the questions they asked people were confusing and not necessarily illustrating what he thought they were.

I haven’t read the paper about the Linda problem yet. But based on the information you gave, I still feel like Pettit’s error stands out.

The description given conjures a clear idea of a kind of person for most people. That kind of person strongly correlates with being a feminist and does not correlate with being a bank teller.

So like… I would challenge Pettit’s version of the question. I am not confident at all that Linda is more statistically likely to possess only one attribute.

I’m gonna revise the description to be more extreme to illustrate how I’m thinking about this.

Pretend the description of Linda was: “Linda is 31 years old, single, outspoken and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear demonstrations. Linda is a member of the feminist movement.

If your choice was between A) “Linda is a bank teller and a feminist” or B) “Linda is a bank teller” then A is more likely. Because we already know she is a feminist, and B is phrased in a somewhat ambiguous way that could be interpreted as exclusive. That is, when presented with them as a choice, people could assume B is implying she is a bank teller and not a feminist, even though it isn’t specifically saying that. So people (understandably IMO) would gravitate towards picking the thing they think they know.

The description is not quit this ironclad, of course, but to many people that description of Laura is going to almost necessitate her being a feminist, so IMO my rephrasing is not that far from how many people probably interpreted it.

Rephrasing my thoughts above to be more stripped down and just illustrating the logic:

Joe is A. Which is more likely?

  1. Joe is A and B
  2. Joe is B
  3. Joe is A

The way Pettit was writing sounds like he thinks it is an error to think 1 is more likely than either 2 or 3. I’m not sure about that. I think at the very least 1 and 2 are equally likely, and both are entirely dependent on the likelihood of B. Right? Or am I missing something?

But anyway it turns out they did account for this issue, and they tried to phrase questions to prevent people from assuming any exclusivity or otherwise misinterpreting the questions. So yeah it seems like the error is Pettit’s, not the people who did the study or wrote the paper.

I think it is an interesting topic. I just read some of the paper, too. I think there is something interesting going on here and I think I can relate to the error the respondents are making. It seems like… people intuitively want to focus on the stuff they are confident about. Any answer that doesn’t let them do that feels less likely to be true. Excepting from the paper:

Argument 1 is logically correct but I can see how it feels wrong to people because it basically ignores what we know about Linda. The respondents are confident Linda is a feminist. So a line like “some women bank tellers are not feminists and Linda could be one of those” feels wrong. She isn’t one of those, she’s a feminist. So pick argument 2 that explains why she must be a feminist (and also a bank teller, I guess).

My guess is that all of the social/cultural/personal elements make it harder to focus on the logical statements. Also, there’s another possible issue… Let me reuse my simplified example from before.

Joe is 99.9999999% likely to be A. Which is more likely?

  1. Joe is A and B
  2. Joe is B
  3. Joe is A

Logically, their likelihood should go 3, 2, 1. Because the tiny chance of him not being A has to be avoided and the unknown chance of him being B has to occur in order for 1 to happen. So Option 1 is less likely. It’s almost the same likelihood as 2, but not quite.

I think this may be counterintuitive for people. Especially if you take it out of an impersonal example like this and phrase it using regular language and real traits. I think the thing that they know is way more likely just feels like it ought to be more likely than the thing that is unknown, and it ought to make the unknown thing more likely by association in the conjoined descriptions.

Pettit says here that F (Linda is a feminist), using his interpretation where it’s F&!T instead of F, was ranked lower than F&T. I’ve just been skimming the paper but it looks like it says the respondents only ranked T (Linda is a bank teller) as lower than T&F. Does it also say F was ranked lower than T&F? If not then even correcting for “only one” then Pettit would still be misrepresenting the paper.

Given that the paper says):

More surprising and less acceptable is the finding that the great majority of subjects also rank the conjunctions (A&J and T&F) as more probable than their less representative constituents (J and T).

then I think the respondents did not rank F as lower than T&F, otherwise the paper would have mentioned that here.

They are equally likely since P(A) = 1. If the events are independent we get: P(A and B) = P(A) * P(B) = 1 * P(B) = P(B). If P(B) is dependent on A we get: P(A and B) = P(A) * P(B|A) = P(B|A). This is the same really because 2. in your example would be P(B|A) (means probability of B given A has occurred) instead of P(B).

Lol I fell for that. My initial answer was $0.10 haha. But it’s $0.05 + $1.05 = $1.10.

I like that quote! Thanks for sharing. It reminds me of Goldratt’s scientist attitude from The Choice.

I also saw this in the article that you shared:

Your intuition really is your most powerful intellectual resource.

It reminded me of what ET says about how cultivating a great subconscious is key to becoming a great thinker.

Also this idea of arbitrating conflicts of ideas and not suppressing one’s intuition (quoting from the article you shared):

When my intuition tells me A and rationality tells me B, I put myself in the position of a referee.

We’re told that our intuition is the mortal enemy of reason, that any dialogue between the two is impossible, and thinking means you have to submit blindly to System 2.

That’s so true. People really misuse the term human nature. From Altas Shrugged:

‘It’s only human,’ you cry in defense of any depravity

People sometimes say that stuff like vanity, deceit, envy, etc. are inherent in human nature because many people have had those traits for millennia. When in reality, they’re just ingrained values that could theoretically be changed (in those who have them).

I agree. That’d be a super cool society to live in.

Nice insight.

He has the subconscious of a great thinker :slight_smile:

1 Like

I noticed that too, forgot to include the quote in my post.

I agree. I’ve often thought that when hearing about the so called “cognitive biases”. I don’t doubt that I or anyone could build the correct intuitions if they tried. So I disagree they tell us about some human nature (which implies an unchangable characteristic of humans).

I got the bat/ball question wrong, I think. I didn’t read the posts in order though and saw elliots reply first where he cuts off the sentence:

It did take me a second to see the word ‘more’ when I reread the original quote. I don’t know that if I would’ve read the sentence properly the math would’ve been intuitive for me. I wonder if most people who make the error make an error in reading the problem or in the math problem itself.

1 Like