Research on false recognition of items and events, often accompanied by
strong subjective feelings of confidence, has informed the study of the
structure and processes of episodic memory (Gallo, 2006). Double
dissociations between true and false recognition (e.g., Stahl & Klauer, 2008) have been difficult to account for by single- process models of
memory and have been invoked as argument for the existence of two
separable mnemonic processes (e.g., Brainerd & Reyna, 2005). In two
false recognition experiments, we varied target repetitions and the
number of lures in study lists and demonstrate the two-dimensionality of
old-new responses through state-trace analysis. We then fit the
Generalized Context Model (GCM, Nosofsky, 1989)—a member of the larger
class of single-process global matching models that predict performance
based on inter-item similarities (e.g., REM, Shiffrin & Steyvers,
1997)—to these data. The model was able to reproduce the two-dimensional
pattern of old-new responses. GCM and global matching models, thus,
provide a single-process account of false recognition: mnemonic
activation of targets and lures differs as a function of inter-item
similarities. We discuss criterion shifts as alternative explanation and
the implications of our findings for dual-process accounts of false memory.
