Retrieval from memory of dietary information

Abstract Two experiments investigated the accuracy with which subjects report their dietary intake for specified periods. In Experiment 1, subjects recorded their diets for 2 or 4 weeks, and returned 0, 2, 4, or 6 weeks after the end of the recording period for a memory test which required that they attempt to report all items that they had eaten or drunk during the recording period. The reported items were scored against the recorded ones. Overall memory performance deteriorated as retention interval increased. In addition, the intrusion rates of subjects who reported for 2 weeks increased steeply as a function of retention interval length, whereas those of subjects who reported for 4 weeks were roughly constant across retention interval length, whereas those of subjects who reported for 4 weeks were roughly constant across retention intervals. Encouraging subjects to report by food groups elevated intrusion rates, relative to other instructions, without increasing the proportion of targets that was reported. In Experiment 2, subjects recorded their dietary intake for two separate periods, and reported for both periods at the end of the second. The items reported for the second period better matched the record for that period than the record for the first period, but there was no such difference for the items reported for the first period. Taken together, the results of the two experiments suggest that subjects rely on generic memory about their own diets when they attempt to report their intake for extended or remote periods.