7 ideas
People evaluate outcomes as changes from a reference point rather than as final states, and the pain of losing a given amount is roughly twice as intense as the pleasure of gaining the same amount. This loss aversion makes people reject favorable bets and cling to losing positions to avoid realizing a loss.
When faced with a difficult question, System 1 quietly answers a simpler related question instead and maps that answer onto the original one without noticing the swap. For example, 'How happy are you with your life?' gets replaced by 'What is my mood right now?', producing confident answers to questions that were never actually addressed.
Cognition runs on two modes: System 1 operates automatically, quickly and effortlessly, generating impressions and intuitions, while System 2 allocates slow, effortful attention to computation and self-control. System 2 is lazy and mostly endorses System 1's suggestions without checking them, so errors arise not from the fast system alone but from the slow system's failure to intervene.
Planners take the inside view: they forecast from the specifics of their own project and imagine best-case scenarios, so they systematically underestimate time, cost and risk. The correction is reference-class forecasting: find the distribution of outcomes for similar projects, use it as a baseline, and adjust only modestly for case-specific information.
The self that lives through moments and the self that keeps score in memory disagree, and decisions are made by the remembering self. Memory ignores duration and weights the peak and the end of an episode, so in the cold-hand experiment people preferred a longer ordeal that ended less painfully, choosing more total pain.
Exposure to a number before estimating an unknown quantity pulls the estimate toward that number, even when the anchor is obviously random, such as a spun wheel of fortune. This works through insufficient adjustment away from the starting point and through priming of anchor-consistent information, and it affects experts pricing houses as well as students.
System 1 builds the most coherent story possible from whatever information is at hand and ignores information that is absent. Confidence tracks the coherence of that story, not the quantity or quality of the evidence, so less information can produce more certainty.