Abstract: This paper models how regret and rejoice arise in sequential search and shows that anticipation of regret can largely explain why people are found to "search too little" compared to the theoretical benchmarks. Due to the fundamental asymmetry in feedback structure in search, the model predicts: First, "search too little" is optimal as long as people are more sensitive towards regret than towards rejoicing. Second, if additional feedback is offered so that people expect to see the next price after search is stopped, behaviors become observationally indistinguishable from the benchmarks. The present model also explains why people occasionally recall past prices.

Key words: sequential search, optimal stopping, reservation price, search too little, recall, regret, rejoicing, feedback