State-trace analysis of medial temporal lobe activity during recognition memory: A test of single and dual process accounts

The ability to recognise previous events is essential to everyday life. Two rival theoretical accounts have characterised recognition memory in terms of a single strength-of-evidence dimension or two qualitatively different processes, termed familiarity and recollection. Despite robust debate concerning the evidence for and against single and dual-process (DP) theories in the psychological literature, the neuroimaging literature has tended to interpret recognition memory related brain activity as exclusively supporting a DP account. The aim of the current study was to test single and DP explanations of recognition memory related brain activity via the application of state-trace (ST) analysis, a method for determining the dimensionality of intervening variables that mediate the effect of IVs on two or more DVs (Bamber, 1979). A levels-of-processing (LOP) manipulation (Craik & Lockhart, 1972) was employed using functional magnetic resonance imaging (fMRI) to contrast proposed familiarity and recollection related brain activity in the perirhinal cortex and the hippocampus, respectively. Seventeen healthy participants performed a "deep" semantic or "shallow" orthographic judgement on single words presented one, two, or four times at study. Subsequently, studied and unstudied items were presented for an old or new decision in conjunction with a high or low confidence rating. Analyses of hit rates (HRs) revealed significant effects of study condition and presentation with greater proportions of high confidence ratings for "deeply" encoded and repeated items. ST analysis provided support for a one-dimensional account of the behavioural data. fMRI data analyses revealed significant changes in perirhinal cortex and hippocampal activity for DP theory dependent contrasts at study and test. However, separate ST analyses at study and test revealed little support for any relationship between the perirhinal cortex and hippocampal regions that showed significant activity. Failure to observe any consistent pattern indicates fMRI signals perhaps reflected vi additional sources of variability. The results are not consistent with either single-process (SP) or DP model accounts, questioning the interpretation of DP theory dependent contrasts as relevant measures of recognition memory related brain activity