Carey Morewedge
Carey K. Morewedge is a behavioural scientist at Boston University and the first author of the 2015 paper reporting that a single training session reduced several decision-making biases, the sunk cost fallacy among them, with the improvement still measurable three months later. It is the strongest evidence for debiasing, meaning deliberate training that reduces a known error, that this site has found in this area.
Who they are
The debiasing literature is mostly a record of things that do not work. Telling people about a bias, warning them in advance, offering them money to be accurate: these have been tried repeatedly and the effects are small, short-lived or absent. That is the background against which the 2015 paper has to be read, and it is why a single-session intervention with a measurable effect three months later is worth attention.
The design is what makes it interesting rather than the headline. A single training intervention, several distinct biases tested rather than one, and a follow-up long enough that a transient effect would have decayed out of it. Sunk cost was among the biases reduced.
Two cautions belong with it. This site holds the abstract and the reported design rather than a full reading of the paper, which is recorded here rather than glossed over. And single-session interventions with durable effects are exactly the class of result that has often failed to hold up, so the honest framing is a promising and well-designed result rather than a settled one.
What it supports, read alongside Richard Larrick's work, is a narrow claim this site is willing to make: training changes the calculation a person runs, and information about the bias does not. Being told about an error is not an intervention. Practising a different procedure is.
Key papers
Morewedge, C. K., Yoon, H., Scopelliti, I., Symborski, C. W., Korris, J. H., & Kassam, K. S. (2015). "Debiasing decisions: improved decision making with a single training intervention", Policy Insights from the Behavioral and Brain Sciences. First author. One training session, several biases reduced, effects still present at a three-month follow-up.
Key ideas
Keep the three-month follow-up in the sentence. It is the part that makes the claim worth anything. A bias that reappears the following week is not a bias that has been trained away.
Training changes the calculation people run. Information about the bias does not. Read together with the work of Richard Larrick, that is a claim this site is willing to make: being told about an error is not an intervention, and practising a different procedure is.
The standard any training claim now has to clear is high. Single-session interventions with durable effects are exactly the kind of result that has often failed to hold up, so the honest framing is a promising and well-designed result rather than a settled one. See The Replication Crisis.
Concepts on this site
Sunk Cost Fallacy, one of the biases his training reduced
Narrow Framing, another of them
The Replication Crisis, the standard any training claim has to clear
Sources
- Morewedge, C. K., and colleagues (2015). Policy Insights from the Behavioral and Brain Sciences. Abstract and reported design only.
- Vault record, 2026-09-11, for Why You Finish the Series You Stopped Enjoying.