Context Dependency: The New Frontier in Decision Science

Most decision scientists still treat context as noise—a variable to control for rather than the central mechanism that determines how people actually choose.

This is the inheritance of Kahneman and Tversky's foundational work. Their heuristics and biases programme revealed systematic departures from rational choice theory, documenting how people rely on mental shortcuts and fall prey to predictable errors. The framework was revolutionary. It also became a trap. By positioning context as the source of bias—something that corrupts an otherwise rational process—we've spent fifty years trying to strip it away rather than understand it.

The uncomfortable truth is that context doesn't distort decisions. Context is the decision.

Consider a simple reframing: you're offered a choice between a certain gain of £800 or an 80% chance of £1,000. Most people take the sure thing. Flip the frame to losses—a certain loss of £800 versus an 80% chance of losing £1,000—and most people gamble. Kahneman called this loss aversion, a bias. But that interpretation assumes there's a "true" preference hiding beneath the frame, waiting to be discovered. There isn't. The preference emerges from the frame. The frame isn't obscuring the decision; it's constituting it.

This matters because the standard response to context dependency has been to try to eliminate it. Behavioural economists have spent decades designing choice architectures to "nudge" people toward supposedly better decisions. The assumption is that if we can just present information more neutrally, people will choose more rationally. But neutrality is a myth. Every presentation is a frame. Every frame activates different mental models, different reference points, different emotional associations. You cannot present a choice without context.

The real frontier is asking: what makes certain contexts more or less conducive to good decisions?

This requires abandoning the bias framework entirely. Instead of asking "how do people deviate from rationality," we should ask "what decision-making processes are activated by different contexts, and when are those processes well-matched to the problem at hand?"

Take the endowment effect—the tendency to value something more highly once you own it. Behavioural economists treat this as irrational. But in many real contexts, it's adaptive. If you've invested time, effort, or emotional labour into acquiring something, the object carries information about your preferences and commitments. Valuing it more highly isn't a bias; it's a reasonable inference. The effect becomes problematic only in specific contexts—like markets where rapid exchange is expected and ownership is purely transactional.

The same logic applies to anchoring, framing effects, and temporal discounting. These aren't universal biases. They're context-sensitive processes that sometimes serve us well and sometimes don't.

What changes when you see it this way? Everything.

First, you stop trying to design choice environments that eliminate context. Instead, you design them to activate the right processes for the specific decision at hand. If you're asking someone to make a one-time, high-stakes choice, you might want to activate deliberative, loss-averse thinking. If you're asking them to make repeated choices in a dynamic environment, you might want to activate faster, more adaptive heuristics.

Second, you recognise that "debiasing" is often futile and sometimes counterproductive. You're not fighting against human nature; you're working with it. The question becomes: how do we structure contexts so that people's natural decision-making processes align with good outcomes?

Third, you abandon the assumption that there's a single "rational" way to decide. Different contexts call for different cognitive strategies. The person who agonises over every decision isn't more rational than the person who trusts their gut—they're just activated by different contextual cues.

Kahneman gave us the tools to see that human decision-making is systematic and predictable. The next step is to stop treating that systematicity as a problem to solve and start treating it as the foundation for better decision design. Context isn't the enemy of good judgment. It's the only thing that makes judgment possible.