Most people know the feeling of choosing a safe option and then wondering about the other one. Regret theory takes that feeling seriously. It says people often think ahead to how bad a wrong call might feel, and that this expectation can change what they choose today.
Regret is the difference in value between the decision actually made and what would have been optimal in hindsight.
What Regret Theory Claims
The core claim is simple. Regret is not only a feeling after the fact. People anticipate it, and that anticipation can shape decisions under uncertainty. Sometimes it leads them to choices that are not the best on expected value alone.
Regret here has a plain definition. It is the difference in value between the choice you made and the choice that would have been best in hindsight. The bigger that gap could be, the more weight it may carry when you decide.
How the Model Works
In the formal version, regret becomes a term in the utility function. That term depends negatively on the outcome you actually get. It depends positively on the outcome of the option you passed up. So a good result on the path you took softens regret, and a better result on the path you skipped sharpens it.
This matters because it lets regret be analysed as part of rational choice, not dismissed as noise. Per Wikipedia's overview, the theory was developed at the same time, in 1982, by Graham Loomes and Robert Sugden, by David E. Bell, and by Peter C. Fishburn.
A worked example
The sources give no worked everyday example, so this one is illustrative only. Suppose you can take Option A, a sure 50 dollars. Or you can take Option B, a coin flip for 100 dollars or nothing. Both have an expected value of 50 dollars.
An expected-value reasoner is indifferent. A regret-averse chooser may not be. If you take B and land on nothing, you know you gave up a sure 50. If you take A and the coin would have paid 100, you know you missed 50 more. Which sting feels worse depends on the person. That difference in anticipated regret is what can tip the choice.
What the Theory Helps Explain
Standard expected utility theory does not always match how people choose. Regret theory helps explain some of the gaps. The sources list three.
- Status quo bias, where people stay put even when change looks better.
- Inaction inertia, where a missed opportunity makes later action less likely.
- Avoiding decisions that could produce an easily imagined counterfactual regret.
The common thread is the foregone alternative. When it is easy to picture, it may weigh more heavily on the choice.
Using the Idea Without Overreaching
You do not need the equations to use the idea. Two questions come straight from the framework. Does the option I am leaning toward mainly minimize imaginable regret, and what does that cost me in expected value? And would my regret be about a foregone alternative that is easy to picture?
There is a real warning here. Avoiding a decision mainly to escape possible future regret can produce inaction inertia or status quo bias. The theory itself flags these as departures from expected utility. Whether they are mistakes in a given case is a separate judgment.
The limits are practical. The sources do not explain how to measure anticipated regret, or how heavily the regret term should be weighted. So any number you attach to it is a guess. Treat it as a prompt to examine your own reasoning, not as a calculation that settles the matter.
The idea has also reached machine learning. A 2026 arXiv preprint proposes Regret-based Preference Optimization, which reframes learning from human feedback around regret minimization instead of reward maximization. The abstract reports consistent gains on mathematical reasoning benchmarks and human preference datasets, with no specific figures given. A related but separate approach, minimax regret, is covered in the Wikipedia article. The portion available did not explain it in detail, so it is not described here.
The takeaway is modest. Regret is a real input to many decisions, and naming it can make a choice clearer. It does not tell you which option is right.
Sources
- Regret (decision theory) - Wikipedia
- [2606.09124] A Regret Minimization Framework on Preference Learning in Large Language Models
- https://vertexaisearch.cloud.google.com/grounding-api-redirect/AUZIYQHec7rE_sQQJDxvEUM-gfGhPKzmHwtZdjU_7YCLEZdUtysAgCETUXtG66JyFdNQdNDiCCe05nSDR8fDpDo1TzrjFjG4qub436z47Va3K77rK7pfrmYaxs1fOOPVQsc2ZSQUQcZkXrbqCXJmWw==
- https://vertexaisearch.cloud.google.com/grounding-api-redirect/AUZIYQEAblyvY1MBrvPDm6xNYhnlZ7M5zyfvXbogiuHfdWqnlarvOC1DCN8flzz7yvBbHQFaE7xsjjlxPQwDc4WpysnojBOEjIh8-ZPlfcBWcfmIvLVLcXx_j3cfSyalg6ISURa-ZuUgWaZAAzPMLhuWnIguf5Ab1WUbb9iWrt9EtC7Kat-k7FWS7roo
Siobhan O'Reilly investigates the friction between urban decay and digital expansion for Deeperpectives. She spent a decade reporting from transit hubs and temporary settlements across the rust belt. Her work prioritizes the raw data of lived experience over polished corporate narratives.
