What are the overconfidence bias and the belief bias, and why do they matter for reasoning?

Overconfidence: your certainty outruns your accuracy. Belief bias: you judge an argument's logic by whether you like its conclusion.

• Overconfidence bias — the systematic tendency to rate your own knowledge, judgement, and predictions as more accurate than they really are. Asked for a range you're "90% sure" contains the true answer, most people are right far less than 90% of the time. It's the direct target of the critical-thinking rule "know your limits," and the antidote to the arrogance that skill quietly breeds.
• Belief bias — accepting or rejecting an argument's validity based on whether its conclusion matches what you already believe, instead of on whether the reasoning actually holds. A logically valid argument with an unwelcome conclusion gets dismissed as "wrong"; a shaky argument with an agreeable conclusion gets waved through. It's what makes people worst at evaluating logic on the topics they care about most.

Both corrupt honest self-assessment: overconfidence inflates trust in your conclusions, while belief bias lets your existing beliefs override the logic sitting in front of you.

Tip: Belief bias is exactly why you should check the form of an argument separately from whether you agree with it — the whole reason validity and truth are kept apart.

Go deeper:

• Overconfidence effect — Wikipedia — when subjective certainty reliably outruns accuracy.
• Belief bias — Wikipedia — judging an argument by how plausible its conclusion feels.