I'm starting my thesis with a little section about the sudden flashes of self-awareness in children and it makes me tear up with beauty.
There probably was a single moment in evolution when intelligence first grasped qualia and reason first stood witness to the fabric of reality.

Public service announcement:
(illustration courtesy of ChatGPT)

A common exchange I see:
Journalist: "How likely is a positive outcome?"
Commentator: "I firmly hope it's high!"

That is, people confuse their hopes and expectations - or even say we're morally obligated to be optimistic (expect a good outcome)

My new post offers an alternative approach

The problem may be that this info depends on emotional emphasis and it's hard to say "use emotion only when appropriate", given LLMs ' immunity to negative prompts. E.g. I learnt about this doctrine but GPT can't tell me "how much it 𝘳𝘦𝘢𝘭𝘭𝘺 manifests in people's thinking"

People were briefly shown the image and asked to say where they saw the lines. The higher the contrast, the better they did -but the worse they did at estimating how good they were.
That's because crisper noise bears both more positive and negative evidence.
Why are we like this?

People underrate the extent to which they are bad at meta-cognition because they are bad at meta-cognition (compared to cognition).
One smart study showed that when people judge how certain they are, they only consider positive evidence (X looks like Y), and neglect dissimilarities (X looks like Z)