Recent research extending Broadbent's work on selective auditory attention is described through the medium of the irrelevant sound effect (the loss of efficiency when irrelevant sound is played during a serial short-term memory task). This breakdown of selective attention cannot be explained by interference at encoding, but rather as disruption following some obligatory entry of sound into memory. Within memory, interference does not arise as a result of the conflict between the contents of memory and the irrelevant sound, since neither phonological nor semantic similarity predicts its extent. Instead, the interference seems to be a product of the similarity in process - the degree of seriation - between rehearsal in short-term memory and t...