Xin’s Reviews > Thinking, Fast and Slow > Status Update

People tend to find coherence between things, constructing stories they believe. And boom, overconfidence emerges.

Seems like again, this chapter is a combination of what has been told.

Got it.
WYSIATI still works here—the halo effect and hindsight.
People are prone to simplify the reality, constructing "coherent" stories due to Sys 1, though the stories are usually just illusions.
They construct reasons for consequences, reshaping their mind to adapt new facts.
Basically, they are just illusions.

Feels like what the chapters just told are mainly combinations of what have been told earlier.

The regression effect, problems with correlation caused by WYSIATI and intuitive heuristics are pretty easy to understand once I've gotten the related basics.

So cool, let's proceed.

People are mostly unaware of what they'd do.
They've studied statistics and psychology, though rarely apply these to themselves. They automatically exempt themselves from such things—Sys 1 lol
People, presumably, only learn psychology when they are truly shocked.
Btw, this is also exactly why Think, Fast and Slow works. Its examples are highly relevant to us.

Seems like what I've just read are basically combinations of former biases and assumptions, and they work lol.

e.g. people usually substitute probability with plausibility, making the answer biased, even violating statistics.

People's opinions are also anchored by what is given. The so-called WYSIATI works perfectly here.

Again, people are prone to simplify things, exaggerate or underestimate certain risks of things based on what they know. That's the affect heuristics.

And the exaggeration or underestimation of certain risks is caused by the availability cascade.

Seems like the concepts mentioned are assembling now.

WYSIATI still plays a crucial role here.

Availability biases are caused mainly because people are prone to rely on their impressions to estimate certain things. They substitute abstract questions with simpler ones, then unconsciously use their Sys 1 to determine whether something frequently happens.

The entire thing, consequently, becomes an illusion lol.

Anchoring is really fun.

It includes some priming and certain laziness of thinking.

Sys 2 relies on Sys 1 to retrieve memories, priming plays a role here. Though Sys 2 can "deliberately" refuse to be affected, Sys 1 can't.

If Sys 2 is lazy, it might just use the result from Sys 1 directly without monitoring. Anchoring occurs here.

People aren't good at reasoning numbers. In fact, Sys 1 isn't sensitive to numbers and quantities and usually exaggerate or underestimate such factors.
I think this is mainly because Sys 1 isn't "designed" (or evolved) to be good at such tasks.
People are actually good at exaggerating patterns among small data. This could be both a flaw and a advantage—take a look at LLMs, small data training is useful sometimes lol

Question substitution is quite common for people who don't use their Sys 2 much when answering certain questions. Because clearly, Sys 1 requires less efforts compared to Sys 2.
Such a process of substitution is usually unconscious. People know exactly the distinction between concepts and things, yet mess them up when answering questions intuitively.