ABSTRACT
Two models of memory structure—schema-copy-plus-tag (Graesser & Nakamura, 1982; Schmidt & Sherman, 1984) and associative-network/depth-of-processing (Craik & Lockhart, 1972; Hastie & Kumar, 1979)—were tested in a 2 × 2 between-subjects design. Type of argument (typical vs. atypical) and measurement interval (immediate vs. 2-day delay in recognition and recall) were manipulated in a print-advertising context. Results indicated that atypical arguments (unusual information) prompt deeper processing of the entire message (implying an associative-network memory structure) rather than some part of the message (as would be hypothesized by the schema-copy-plus-tag formulation) and that this effect prevails under both immediate-and delayed-measurement conditions.