Direct measurement of memory for context

In this study, context was manipulated by presenting stimuli consisting of two words, a large grey word (target) appearing in back of a smaller black word (context). Participants were instructed at encoding to generate associations between two words but to remember only the grey word. At test, 4 classes of word pairs appeared; Targetold-Contextold, Targetold-contextnew, Targetnew-Contextold and Targetnew-Contextnew. Memory was measured by a recognition test, where participants were told to respond 'old' to targets. Applying signal detection to memory for context words (i.e. contextold = hit, contextnew = false alarm) reveals a significant effect only when target stimuli were old. The implications for these results are addressed