There's some evidence that recalling a memory makes it labile, and that interrupting memory reconsolidation processes during this labile period can 'erase' them, much like not getting sleep after initially forming a memory can prevent it from becoming permanent.

Emotional/fear associated memories are a really easy model to work with, and are also the most clinically compelling application of memory extinction.

See Cristina alberini's work, and others, as well as reviews like http://www.nature.com/nrn/journal/v8/n4/abs/nrn2090.html