Classical economics assumed that people behave rationally — that they calculate expected value, update their beliefs in response to evidence, and act in their own interest as they understand it. Behavioral economics is the systematic documentation of how wrong that assumption is. It turns out humans are not primarily rational agents who occasionally make emotional errors; they are pattern-matching systems that produce rational-seeming outputs in some domains and spectacularly irrational ones in others, in ways that are consistent, predictable, and largely invisible to the person making the decisions.
The foundational text is Thinking, Fast and Slow by Daniel Kahneman. Kahneman spent 40 years with Amos Tversky documenting the structure of human cognitive error, and this book synthesizes that work for a general audience. The two-system framework — System 1 (fast, automatic, associative) and System 2 (slow, deliberate, effortful) — gives you a vocabulary for why your intuitions are sometimes right and sometimes badly wrong in ways you couldn't have predicted. The specific biases — anchoring, availability, the planning fallacy, loss aversion, the focusing illusion — are not personality quirks but features of cognition that emerge from the structure of System 1. Knowing their names doesn't eliminate them, but it creates the faint possibility of catching them. This is the most practically useful book I know on how human minds actually work.
For the evolutionary frame — why these "biases" exist at all, what they were optimizing for — The Selfish Gene by Richard Dawkins is indispensable. Dawkins's argument is that natural selection operates at the level of the gene, not the organism, and that behavior that appears irrational from the organism's perspective may be perfectly rational from the gene's. Loss aversion — the finding that losses feel twice as painful as equivalent gains feel pleasant — makes sense once you consider that losing resources was twice as dangerous as gaining them in the environment where this system evolved. Dawkins gives you the selection pressure that produced the machinery Kahneman describes. Together, they explain not just what the biases are but why they're there.
Oliver Sacks's The Man Who Mistook His Wife for a Hat contributes something the others don't: detailed case studies of what decision-making looks like when the machinery is damaged. His patient who processes faces as objects — who sees the features of a face but cannot assemble them into a recognized person — is doing something structurally similar to what we all do with information: pattern-matching against a model of the world. When the model fails, the failure mode is often invisible to the person experiencing it. We don't know when our pattern-matching is off because the subjective experience of confident wrong inference is identical to the subjective experience of confident correct inference. Sacks makes this visible in extreme cases, which illuminates the subtler version in normal cognition.
The question of how to make better decisions — given all these structural limitations — is addressed most directly in Flow by Mihaly Csikszentmihalyi, though not in the way you might expect. Csikszentmihalyi's finding that absorbed engagement produces clearer thinking than anxious effort is relevant to decision-making because the conditions that produce good decisions — calm attention, clear feedback, appropriate challenge level — are the same conditions that produce flow. High-stakes decisions made under stress, with unclear feedback and overwhelming complexity, produce the worst outcomes. Csikszentmihalyi helps you understand why, and what kinds of conditions allow better thinking to emerge.
Finally, The Myth of Sisyphus by Albert Camus is an unusual addition to a behavioral economics reading list, but it addresses something the empirical literature doesn't: the question of what to do once you've accepted that you're irrational. Camus's response to the absurd — the recognition that the universe doesn't offer the meaning you're looking for — is not despair but revolt: you act anyway, fully, without false comfort. The behavioral economics equivalent is: you know your decisions are systematically biased, you know you can't eliminate the biases, and you act anyway, with appropriate humility about what you can and cannot know. That's not paralysis. It's the most honest available approach to choice.
Read Kahneman first, Dawkins second for the evolutionary context, Sacks third for the case studies, Csikszentmihalyi fourth for the practical implications, Camus last for the philosophical frame. The result is not a decision-making system — those are part of the problem — but a more accurate model of what you're working with when you decide anything.