Forgetting as a consequence of retrieval suppression: a meta-analytic review of the think/no-think paradigm

Abstract

The present thesis used meta-analytic methods to define the typical magnitude of the suppression induced forgetting (SIF) effect produced by the Think/No-Think paradigm as well as assess several moderators. A literature search was conducted using seven online databases, review papers and discussion with experts in the field to identify 82 relevant studies. Using Bayesian multi-level modelling techniques to provide a synthesis of the existing literature it was determined that the typical Think/No-Think study produced a small SIF effect. This means that the Think/No-Think paradigm decreases recall accuracy for no-think items relative to baseline items, with broad heterogeneity in the observed effect across studies depending on the methods. Additionally, it was determined that the SIF effect was larger (a) in studies using same probe (SP) tests than independent probe (IP) tests; (b) when the data were conditionalized on successful learning of the stimuli pairs during the test-feedback phase; (c) when participants were given more specific instructions on how to avoid thinking of the no-think items, and; (d) at higher numbers of repetitions (for unconditionalized data). Neither stimulus emotional content nor stimuli type impacted the magnitude of SIF. These findings provide strong evidence of the viability of the SIF effect while also informing future theoretical discussions concerning its mechanisms. Several recommendations are made for future Think/No-Think studies concerning sample size and methodological decisions in an effort to improve the precision and reliability of findings.