6 ideas
All thinking can be analyzed as a search for three objects—possibilities (candidate answers), evidence (information that bears on them), and goals (criteria for evaluation)—followed by inference that uses evidence to strengthen or weaken each possibility relative to the goals. Good thinking is defined by how thoroughly and fairly these searches are conducted, not by whether the conclusion turns out correct.
Rationality consists largely of searching for evidence and possibilities that oppose one's current favored conclusion and weighing that opposing evidence fairly rather than discounting it. The opposite—favoring a possibility and seeking only confirming evidence—is the central bias that makes thinking poor.
People systematically search for and weigh evidence in favor of conclusions they already hold, a defect of process rather than capability. This means a thinker can be intelligent and well-informed yet still reason badly because the bias operates at the level of how evidence is sought and counted, not how much is known.
The rational choice under uncertainty is the one that maximizes expected utility—the sum of each outcome's utility weighted by its probability—where utility measures how well an outcome satisfies one's goals. This serves as a normative benchmark against which actual human choices can be measured to identify specific deviations.
Any account of thinking should be separated into the normative (what ideal performance would be given goals), the descriptive (what people actually do and where they fall short), and the prescriptive (realistic guidance that helps real, limited people move closer to the normative ideal). Confusing these levels leads to either impossible standards or excusing genuine errors as merely human.
People judge harms caused by action more harshly than equivalent or worse harms caused by inaction, and treat the status quo as a privileged baseline. This asymmetry produces decisions—such as refusing a vaccine with small risks over accepting a disease with larger risks—that fail to minimize total harm because the mode of causation is weighted instead of the outcome.