Seeing What Matters: Framing, Reasoning, and Behaviour

Framing and Its Limits

People feel better about a medical treatment that has a 90% chance of survival than about one with 10% chance of death.  The way a choice is framed shapes decisions, which is intuitive, well documented, and also at odds with a basic sense of rationality. Yet, such findings are not in any substantial way integrated into most areas of economic activity. Why not?

Perhaps framing is not really that important in practice?  First, compared to standard factors such as information or incentives, the thinking goes, it may not have much impact on individual decisions, much less on firms or markets. Second, while the laboratory evidence is compelling, in daily life such errors do not  seem obvious or prevalent — especially in domains where you feel you have expertise or large stakes.  And finally, attempts to apply framing or choice architecture techniques in the field often produce muted or very heterogeneous results ^[1,2], suggesting to many that these effects are unreliable.

Why does framing sometimes work and sometimes not?  In fact, why does it work at all?  A growing cognitive approach to decision making suggests that the potential of framing but also its limits are two sides of the same coin.  Faced with many competing stimuli, we reason by deciding which features and information are relevant, attending to them, and neglecting others ^[3]. In this view, framing a choice is a way to influence what you pay attention to and what you remember – in a word, what you think is relevant.

Importantly, the same attention and memory mechanisms also explain why people can be risk-averse in some contexts and risk-seeking in others, and why their outlook can swing from overoptimism to pessimism. In turn, this helps account for volatility in financial markets ^[4], macroeconomic fluctuations ^[5], advertising ^[6], consumer competition ^[7], and so on. Framing is therefore a key window into the heart of decision making.

One Image, Two Readings: The Roles of Attention and Memory

Theories of decision-making, both rational and behavioural, are typically based on information and preferences, however quirky. The cognitive perspective starts elsewhere: decisions are guided by our mental representations, which select relevant features to attend to. The quality of those representations, more than raw processing limits, drives decision quality [1].

The classic T/-\E C/-\T illusion illustrates the point.  In T/-\E, you recognize an H; in C/-\T, you recognize an A.  How do we come to two completely different readings of the same image? In each case, the adjacent letters retrieve a familiar word from memory, and we attend to the features of the ambiguous letter that match that word.  Crucially, this crowds out thinking about the alternative word, or the other features of the ambiguous image. With both attention and memory aligned, the interpretation feels obvious – and therefore, to point 1 above, we do not feel we are committing an error.

Such ambiguous interpretations are ubiquitous. Is buying tech stocks an “investment” or a “gamble”? It depends on whether returns or volatility draw attention. Correct decisions would take both views into account; instead, like in the THE/CAT, we often use representations that are incorrect but familiar.  Portfolio managers, for instance, describe buying stocks as “new investment ideas” and selling stocks as “raising cash,” reflecting their representation of the job as stock-picking.  As a consequence, they devote effort to buying but less to selling — which hurts their returns ^[8].

Disagreement and instability: two sides of the same coin

This mechanism helps explain two pervasive features of human behaviour.  First, there is striking disagreement even on seemingly objective issues, driven by attention to different features.  When inflation hit in 2022, people with past experiences of high inflation were immediately concerned about current price growth, while others – with the same information, but different experiences – were instead concerned about unemployment [9], with profound implications for individual but also policy choices. Second, if context changes, we can shift to a new representation, often with equal confidence. A good idiosyncratic experience, such as receiving a raise, makes the same person more optimistic about future inflation, consider different scenarios for economic growth, and prefer different policies ^[10].  

The fact that people disagree so much, but also hold inconsistent views across similar situations, are major “puzzles” – a “large scale” version of the heterogeneity in framing effects.  This is however very natural from the cognitive perspective, because people represent the world through selective attention and memory.

Conclusion: measuring cognition as a step forward

The central claim is that errors arise less from information processing limits and more from the mental representation behind decisions. Recognizing this shifts the agenda and raises two opportunities. 

First, information and incentives – the twin levers of economics – shape decisions only to the extent they are attended to. The same information can lead to wildly diverging beliefs: some travellers focus on the possibility of a plane crash, while others think, if at all, of its minuscule likelihood.  Arguing with the former that the probability is small is not just ineffectual, it is missing the point.  Charging a fine to reduce an unwanted behaviour can be ineffectual if people are inattentive ^[11], or it can backfire if people represent it as a price for acceptable behaviour ^[12]. To interpret how people respond to information or incentives requires understanding how it fits with their representations. 

Second, and crucially, attention and memory have been studied extensively: we know they are shaped by cues and experiences following well-established regularities, which means we can predict the impact of interventions.  They can also be measured, offering a direct probe into the decision-making process. The framing experiments of the 1970’s were a beginning. We can now use this approach to improve the performance of organizations, the design of policy, and the efficiency of markets.  

Pedro Bordalo is Professor of Financial Economics at Saïd Business School, University of Oxford.  He co-leads BR-UK's work on Organisations, Markets & The Economy.