A 2% probability of winning $10,000 receives a decision weight of 8.1% — four times its actual probability. A 98% probability receives a weight of only 87.1%. Humans don't process probabilities as numbers — they process them as emotions, and the emotions don't respect the math.

The Framework

Decision weights are the psychological values people assign to probabilities — distinct from the probabilities themselves. The decision weight function is not linear: it overweights low probabilities (the possibility effect) and underweights high probabilities (the certainty effect), with compressed sensitivity in the middle range. A 2% probability gets 4× its fair weight. A 98% probability gets only 89% of its fair weight. The middle range (30-70%) is relatively flat — people don't distinguish well between 40% and 60%.

The decision weight function, combined with the value function, produces prospect theory's complete model of choice under uncertainty. The value function determines how gains and losses are experienced. The decision weight function determines how probabilities are experienced. Together they explain lotteries, insurance, risk-seeking for losses, risk-aversion for gains, and the fourfold pattern.

Where It Comes From

Chapter 29 of Thinking, Fast and Slow presents decision weights alongside the fourfold pattern. The numbers come from Tversky and Kahneman's cumulative prospect theory (1992), which refined the original 1979 prospect theory. The key data point: a decision weight of 0 for probability 0, then a sharp jump to ~6% at probability 1%, then gradual increase through the middle range, then a sharp jump to 100 at probability 100%.

> "People overweight small probabilities and underweight moderate and high probabilities." — Thinking, Fast and Slow, Ch 29

The Implementation Playbook

Offer Design: A '2% chance of winning a free year' (overweighted to ~8%) creates more excitement than its expected value warrants. Conversely, '98% satisfaction rate' (underweighted to ~87%) sounds less reassuring than it should. Design promotions around low-probability high-excitement events and communicate safety through certainty (100%), not near-certainty (98%).

Risk Communication: In the middle range (30-70%), people are relatively insensitive to probability differences. 'A 40% chance' and 'a 60% chance' feel more similar than they are. For important risk communications, anchor to the extremes: frame as 'nearly impossible,' 'coin flip,' or 'nearly certain' rather than precise percentages.

Insurance Pricing: People will overpay for insurance against low-probability events (terrorism insurance, extended warranties) because the decision weight overweights the small probability. The overweighting is the insurance company's profit margin.

Key Takeaway

Decision weights mean that humans experience probability on a warped scale. Small chances feel bigger than they are; near-certainties feel less certain than they are; and the middle range is compressed. Every pricing, risk, and guarantee decision should account for this warping — because your customers, counterparts, and stakeholders are making decisions based on decision weights, not on probabilities.

Continue Exploring

[[Possibility Effect]] — The specific overweighting at low probabilities

[[Certainty Effect]] — The specific underweighting at high probabilities

[[Fourfold Pattern]] — Decision weights + value function = four behavioral zones

📚 From Thinking, Fast and Slow by Daniel Kahneman — Get the book