A meta-analysis of event-related potential correlates of recognition memory

Abstract

A longstanding question in memory research is whether recognition is supported by more than one mnemonic process. Dual-process models distinguish recollection of episodic detail from familiarity, while single-process models explain recognition in terms of one process that varies in strength. Dual process models have drawn support from findings that recollection and familiarity elicit distinct electroencephalographic event-related potentials (ERPs): a mid-frontal ERP effect that occurs at around 300-500 ms post-stimulus onset and is often larger for familiarity than recollection contrasts, and a parietal ERP effect that occurs at around 500-800 ms and is larger for recollection than familiarity contrasts. We sought to adjudicate between dual- and single-process models by investigating whether the dissociation between these two ERP effects is reliable over studies. We extracted effect sizes from 41 experiments that had used Remember-Know, source memory, and associative memory paradigms (1,000 participants). Meta-analysis revealed a strong interaction between ERP effect and mnemonic process of the form predicted by dual-process models. Although neither ERP effect was significantly process-selective taken alone, a moderator analysis revealed a larger mid-frontal effect for familiarity than recollection contrasts in studies using the Remember-Know paradigm. Mega-analysis of raw data from six studies further showed significant process-selectivity for both mid-frontal and parietal ERPs in the predicted time windows. On balance, the findings favor dual- over single-process theories of recognition memory, but point to a need to promote sharing of raw data.

Keywords: Event-related potentials; Familiarity; Meta-analysis; Recognition memory; Recollection.

Fig. 1

Diagrammatic representation of the stages of the systematic literature search. The full protocol, data, and code are available in the Online Supplemental Material and via the Open Science Framework at https://osf.io/acwtv/. The literature search yielded 6,817 articles (see Fig. 1 for details). We (SK, AM) first screened article titles and abstracts for relevance. If those fields were irrelevant to recognition memory and/or the event-related potentials of interest, we excluded the corresponding articles. We then read the full text of the remaining 291 articles to assess if they did indeed meet the inclusion criteria (see Inclusion criteria section). This inspection yielded 41 articles that could be included in the current meta-analysis

Fig. 2

Forest plots for mid-frontal and parietal event-related potential (ERP) effects. Individual study values (filled squares) represent standardized mean differences (effect sizes) for ERP effects. The overall meta-analytic effect size across all contrasts is given below (filled black diamond) with model statistics. Subgroup model statistics and meta-analytic effect sizes are plotted separately for familiarity and recollection contrast subgroups. Error bars represent 95% confidence intervals. For definitions of quantities see Methods – Meta-analysis

Fig. 2

Forest plots for mid-frontal and parietal event-related potential (ERP) effects. Individual study values (filled squares) represent standardized mean differences (effect sizes) for ERP effects. The overall meta-analytic effect size across all contrasts is given below (filled black diamond) with model statistics. Subgroup model statistics and meta-analytic effect sizes are plotted separately for familiarity and recollection contrast subgroups. Error bars represent 95% confidence intervals. For definitions of quantities see Methods – Meta-analysis

Fig. 3

Interaction plot of effect sizes by event-related potential (ERP) effect and contrast. (a) Mid-frontal ERP effect. (b) Parietal effect. Filled circles show effect sizes (Hedge’s g) for each study, shaded areas show the probability density function of the data, and boxplots show the median and interquartile range. Insets illustrate the electrode inclusion criteria for each effect (note that mid-frontal effects had to have been measured symmetrically, whereas electrode groupings for parietal effects could have been left-sided or symmetrical; see Methods – Inclusion criteria)

Fig. 4

Interaction plot of effect sizes by experimental paradigm and process for each event-related potential (ERP) effect. (a) Mid-frontal ERP effect. (b) Parietal effect. Filled circles show effect sizes (Hedge’s g) for each study, shaded areas show the probability density function of the data, and boxplots show the median and interquartile range. Insets illustrate the electrode inclusion criteria for each effect (see Methods – Inclusion criteria)

Fig. 5

Funnel plots of event-related potential (ERP) effect sizes. The group of four plots on the left shows the relationships between effect size (Hedge’s g) and its standard error for mid-frontal and parietal effects for recollection and familiarity processes. The pair of plots on the right shows this relationship for the differences in each ERP effect according to process

Fig. 6

Interaction plots of effect sizes in the mega-analysis by time window and process for each event-related potential (ERP) effect. (a) Mid-frontal effect. (b) Parietal effect. Filled circles show marginal means for each experiment, and bars show the median and interquartile range across experiments. Insets illustrate the electrode inclusion criteria for each effect (see Methods – Inclusion criteria)