Litmus test of rich episodic representations: Context-induced false recognition

Context-dependent episodic memory is typically investigated using tasks in which retrieval occurs either in the reinstated context of encoding or in a completely new context. A fundamental question for episodic memory models is how detailed those episodic memory representations are that contain contextual information about the encoded event. The present study examined whether memory is affected when the contexts of encoding and retrieval are highly similar but not exactly the same. At encoding, participants saw unique object images presented on the background of unique context scene images. On a surprise recognition test, the objects were either old or visually similar to ones seen at encoding (lure stimuli). The objects were presented on either the old or a lure context image; the lure context image was visually similar to the corresponding object’s encoding context. Context reinstatement increased the hit rate for the old objects. Additionally, the false alarm rate for the lure objects was higher in the old context condition indicating that the presence of the encoding context at test does not always aid recognition memory decisions. These results suggest that slight visual differences between the contexts of encoding and retrieval matter and that the unique association formed at incidental encoding between a focal stimulus and its own visually rich context affects subsequent recognition memory performance.