Many studies have found a gap between willingness-to-pay and willingness-to-accept that is inconsistent with standard theory. There is also evidence that the gap is eroded by experience gained in the laboratory and naturally occurring markets. This paper argues that the gap and the effects of experience are explained by a caution heuristic. This conjecture is tested in a repeated market experiment with symmetric and asymmetric information. The results support the conjecture: people do seem to use heuristics rather than reacting optimally and their behavior adjusts slowly when the environment changes.