Abstract
Reality monitoring of verbal memories was compared with decisions about pictorial memories in this study. Experiment 1 showed an advantage in memory for imagined over perceived words and a bias to respond "perceived" on false alarms. Experiment 2 showed the opposite pattern: an advantage in memory for perceived pictures and a bias to respond "imagined" on false alarms. Participants attribute false alarms to whichever class of memories has the weakest trace strengths. The relative strength of memories of imagined and perceived objects was manipulated in Experiments 3 and 4, yielding changes in source attribution biases that were predicted by the strength heuristic. All 4 experiments generalize the mirror effect (an inverse relationship between patterns of hits and false alarms commonly found on recognition tests) to reality monitoring decisions. Results suggest that under some conditions differences between the strength of memories for perceived and imagined events, rather than differences in qualitative characteristics, are used to infer memory source.