Post-Kahneman Decision Theory: What Behavioral Science Missed

Behavioral economics has spent fifty years proving that humans are irrational, and in doing so, it has systematically overlooked something more important: the rationality of irrationality itself.

Daniel Kahneman's legacy reshaped how we understand choice. Heuristics and biases became the lingua franca of decision science. We learned that people anchor to irrelevant numbers, that they fear losses more than they value gains, that they mistake familiarity for safety. These insights were genuine. They held up. But they were also incomplete in a way that matters now, in 2026, when we have enough data to see what the original framework left out.

The thing everyone gets wrong is that behavioral science treated deviations from rational choice theory as errors—cognitive bugs to be catalogued and, implicitly, corrected. This framing was seductive because it positioned the researcher as the one who saw clearly, while the subject remained trapped in systematic misperception. But this inverted the actual problem. Most of what behavioral science called "bias" was not a failure of cognition. It was a failure of the model.

Consider how people actually make decisions about repeat purchases. The standard behavioral account says they're anchored to sunk costs, or they're exhibiting status quo bias, or they're falling prey to the mere exposure effect. All true, technically. But this misses the operative mechanism: people are reinforcing their own decisions through selective attention to confirming evidence. After buying something, they notice its benefits more acutely. They remember positive experiences and downweight negative ones. This looks like bias from outside. From inside, it's a coherence-seeking system that makes sense—it reduces cognitive dissonance and stabilizes identity.

The reason this matters more than people realize is that it changes what we should actually be measuring. If the goal is to understand decision-making, we need to stop asking whether people deviate from rational choice theory and start asking what problem their decision-making solves. Kahneman's framework was built for a world of isolated, one-off choices. It worked well there. But most decisions are embedded in sequences. People don't choose once; they choose, then live with the choice, then choose again. The decision-making system that evolved—or that people develop through experience—is optimized for consistency and narrative coherence, not for maximizing utility on each individual occasion.

This is where post-Kahneman thinking becomes necessary. We need a decision theory that accounts for how people reinforce their own choices over time. Not as a bias to be corrected, but as a feature of how humans maintain stable preferences and identities in an uncertain world. When someone buys a product and then notices its benefits more acutely, they're not being irrational. They're engaging in a form of active sense-making that keeps their behavior aligned with their self-concept.

The practical implication is that interventions designed around the old behavioral model often fail or backfire. If you try to "debias" someone by showing them they're anchored to a sunk cost, you're working against a system that's actually trying to maintain coherence. The person will resist, or they'll comply superficially while reverting to their original pattern. What actually changes behavior is providing new information that fits into their existing narrative, or creating conditions where a different narrative becomes more coherent.

This is not a return to rational choice theory. It's a move beyond the binary of rational versus irrational. It's an acknowledgment that human decision-making is rational at a different level of analysis—not at the level of individual choices, but at the level of maintaining a coherent self across time.

The next generation of decision science will need to grapple with this. The question is no longer whether people deviate from rational choice. It's whether we've been asking the right question at all.