Read on Twitter
π
π¨π¨π₯ππ ππ² πππ§ππ¨π¦π§ππ¬π¬ π’π§ππ¨ ππππ’π¬π¦
A thought experimental thread
Does intuition enlighten or mislead? Depends. In ordered contexts with fast feedback (chess), intuition works well. In chaotic multivariate contexts (penny stock picking) it fails.
A thought experimental thread
Does intuition enlighten or mislead? Depends. In ordered contexts with fast feedback (chess), intuition works well. In chaotic multivariate contexts (penny stock picking) it fails.
The problem is that in the latter case you'll still think you're learning things, overfit transient patterns, and be overconfident in your future prediction ability. This is @nntaleb's main point in "Fooled by Randomness".
Now let's bring predictive processing into the picture.
Now let's bring predictive processing into the picture.
PP tells us there are three ways you make you predictions match sensory input:
1. Change your underlying models and their predictions based on what you see.
2. Change your perception to fit with what you predicted.
3. Act on the world to bring the two into alignment.
1. Change your underlying models and their predictions based on what you see.
2. Change your perception to fit with what you predicted.
3. Act on the world to bring the two into alignment.
In chaotic, complex and anti-inductive contexts, #1 doesn't work. There's no better model you can switch to to immediately improve your predictions. You're left with #2 and #3.
Telling whether people are good or bad at a glance is an example of a complex anti-inductive problem.
Telling whether people are good or bad at a glance is an example of a complex anti-inductive problem.
You're on a flight to planet X Γ A-12, and the magenta-skinned passenger next to you informs you that cyanskins are nasty and deceitful. In reality, skin color doesn't correlate with honesty or good intentions, but you can't know that.
Or perhaps there's no magenta passenger, just the first cyan skinned person you happen to engage with turns out to be nasty because their space-tricycle was t-boned at the holo-brothel that morning but you fall for the fundamental attribution error.
You meet cyan skinned people. If they're blunt, you perceive that as nastiness. If they're tactful, you perceive that as dishonesty. You literally see facial twitches and hear notes that aren't there, PP making confirmation bias propagate all the way down to your basic senses.
If they're actually nice, your brain gets a prediction error signal and tries to correct it with action. You taunt to provoke nastiness, or become intimidating to provoke dishonesty. You grow ever more confident in your excellent intuition with regards to those cyan bastards.
This is why confirmation bias is the mother of all bias. CB doesn't just conveniently ignore conflicting data. It reinforces itself in your explicit beliefs, in unconscious intuition, in raw perception, AND in action. It can grow from nothing and become impossible to dislodge.
How to cure one of their anticyanism?
The least useful is to shout "no, it's magentas who are bad" and punish anyone who disagrees. Explicitly stated beliefs are the least-important part of their anti-cyan bias, and humans will say out loud whatever is politically expedient.
The least useful is to shout "no, it's magentas who are bad" and punish anyone who disagrees. Explicitly stated beliefs are the least-important part of their anti-cyan bias, and humans will say out loud whatever is politically expedient.
The model "magenta-skinned people are bad and cyans are good" will make bad predictions in a world where skin color is uncorrelated and in a brain that has anti-cyan perception bias. Repeating this will only reinforce the bias and disconnect it from any explicit reasoning.
I think the most useful thing would be to have the person ACT as kindly as possible towards cyan people, regardless of what they say or believe. For example, this can be achieved by cyan people joining communities where the norm is being really nice and supportive to each other.
Acting nice works in two ways:
1. To avoid dissonance, the brain updates its models to explain its own behavior (I'm kind to them, so they must be good).
2. People you are nice to usually reciprocate, reinforcing again the idea that they are good.
1. To avoid dissonance, the brain updates its models to explain its own behavior (I'm kind to them, so they must be good).
2. People you are nice to usually reciprocate, reinforcing again the idea that they are good.
To bring this back to Earth: if you think you're fighting racism (or any other bias) by yelling, punishing, breaking trust, enforcing conformity in what people say publicly, and creating environments of hostility and suspicion β you're almost certainly making it worse.
Reading people is incredibly hard. Given huge interpersonal variety and multicausality, judging individuals based on surface attributes is almost always confirmation bias and not wisdom. If you're not sure, try just being extra kind to everyone for a while and see how that goes.
