Remembering pragmatic inferences

After hearing that 'the flimsy shelf weakened under the weight of the books,' people often remember having heard that the shelf broke-an inference that is not logically necessitated by the statement. Indeed, when asked to choose which word was heard - weakened or broke - people often mistakenly choose broke, a pragmatic inference, as having been heard (McDermott & Chan, 2006). On what basis are these decisions made? The present study explored the subjective experience accompanying memory for pragmatic inferences with remember/know/guess classifications (Gardiner & Java, 1991; Tulving, 1985). The phenomenological experiences accompanying such false memories were indistinguishable from those of true memories. The dual process signal detection account proposed by Wixted and Stretch (2004) provides a framework for understanding our results.