According to the University of Chicago:

"Bias is a pre-formed negative opinion or attitude toward a group of persons who possess common characteristics, such as skin color, or cultural experiences, such as religion or national origin."

 

So, should we ever be biased?  And if the answer is yes then should we hide our biases for signaling reasons?

 

Or should we take into account that many people have irrational biases against those who possess different skin colors, religions or national origins.  So perhaps a high percentage of any negative biases readers of this blog have are irrational and so maybe the best course for us rationalists in training is to work against having any negative biases.

New to LessWrong?

New Comment
23 comments, sorted by Click to highlight new comments since: Today at 4:48 AM

Yes, we should hold true negative beliefs about groups. No, we should not hold false negative beliefs about groups. What else is there to say?

many people have irrational biases against those who possess different skin colors, religions or national origins

Among intellectuals, irrational bias (in this sense) isn't obviously more common than irrational lack of bias.

The definition referenced in the post is a special case. It talks about a situation where people form a cached overly negative opinion, and apply it to a whole group. This makes them systematically wrong about the individuals of that group, who are not as negative as the stereotype, and thus this situation constitutes a bias, systematic error of judgment.

The post is poorly written, its topic or contribution are unclear.

So, should we ever be biased? And if the answer is yes then should we hide our biases for signaling reasons?

The only logically consistent answer to give to the second question is "no".

The definition of bias given in the article is, I believe, quite different from the definition accepted here. LW/OB largely thinks "bias" means inaccurate opinion, not socially unacceptable opinion. There's no natural law saying negative opinions or pre-formed opinions are always inaccurate. But they often are, especially when we're thinking about clumps of people different from "us".

Now for my personal answers to your questions, that may or may not be shared by others here. Yes, you should keep the capability to be biased if you want to function more efficiently. Yes, you should hide it for signaling reasons. Yes, you should take into account the irrationality of other people's biases and reduce your own biases to an evidentially justified level of negative (or positive) "stereotype", but no closer to "neutrality".

"The definition of bias given in the article is, I believe, quite different from the definition accepted here. "

You are probably right but it is very dangerous to define a word differently than most other people do when that word has extreme negative connotations for many people.

Talk to the University of Chicago about that - they are the ones who need to get a dictionary:

http://dictionary.reference.com/search?q=bias

Bias is a pre-formed negative opinion or attitude [...]

Here, the word "pre-formed" is doing the work of saying that the negative opinion is not evidentially justified ... i.e., the negative opinion is inaccurate. So the University's definition and LW's definition aren't that different.

To explain my downvote: your leap from "pre-formed" to "inaccurate" is too poorly justified. Humans hold many accurate pre-formed (even inborn) opinions, like "being thirsty for a long time is bad for you".

I'm glad you explained the reason for your downvote, because it's easy to argue with.

I was not asserting that "pre-formed" means "inaccurate". I was just explaining that in this context, the University chose the words "pre-formed" to convey this meaning. Don't blame the messenger.

Later edit: If I was saying that "pre-formed" has the same meaning as "inaccurate" then why would I have used the nuanced phrase, "does the work of saying"?

Don't you mean "Should we be biased?"

Thanks, change made.

It's not just the title; you do it again in the text.

So, should we ever be biased?

Are you asking whether it would ever be justified to

(a) hold negative opinions against a group, even if the individual group members may vary on the negative quality

or

(b) hold a negative opinion that isn't based on any evidence?

For example, being afraid of all spiders because some are poisonous would be (a), while being afraid of spiders because you believe they have mind control powers would be (b).

Assume that on average members of a group are more likely to have a negative quality than are members of the general population.

No kidding.

There's nothing I like less than bad arguments for a view that I hold dear.

--Daniel Dennett (hat tip: Steven).

The judgment is the bottom line. What you write above it doesn't change the nature of the judgment, doesn't change whether it's biased, systematically mistaken. You may be accurate in deciding to be afraid of spiders, even if you do that only because you believe that God told you to do that.

This is the second time you've criticized what I've written due to some semantic quibble. Are you saying that being afraid is not an opinion? Are you asserting some weird distortion of the meaning of "accurate"?

(Just so you know: I agree that a bias not based on evidence could be accurate by accident, but I'm still going to call it inaccurate.)

Just so you know: I agree that a bias not based on evidence could be accurate by accident, but I'm still going to call it inaccurate

That's exactly how I interpreted your message (modulo s/bias/opinion, I assume, bias by definition can't be accurate, however heuristic can). My reply is an explanation of why I disagree that one should call that opinion inaccurate.

Well, we understand each other then. Maybe you would be happy (or less unhappy?) to know that everywhere in my draft I had written "inaccurate until substantiated" instead of "accurate" before I edited it out.

The edit wasn't because I'm biased against statements that are not substantiated. But, really, I didn't think the extra precision was worth the cost in simplicity.

Maybe you would be happy (or less unhappy?) to know that everywhere in my draft I had written "inaccurate until substantiated" instead of "accurate" before I edited it out.

Well, since the accuracy of a statement doesn't causally depend on whether it's substantiated, so you can't flip the accuracy of a statement by finding substantiation, I don't see how that helps.

So "inaccurate until substantiated" isn't good enough either.

How about "inaccurate unless possibly substantiated"?

Or, "inaccurate unless possibly substantiated or just happens to be true but is unverifiable"?

Do you really love accuracy this much?