[Proposal] Can we really control our memory?

Can we really control our memory?

More than 50% of people believe they can (Otgaar et al., 2021). This year marks the 20th anniversary of the Think/No-Think task (TNT; Anderson & Green, 2001); an experimental paradigm that was developed to provide an empirical answer to this highly debated question, thus, studying suppression-induced forgetting – a phenomenon whereby an individual’s “concerted attempts at preventing […] memories from entering awareness can subsequently make it more difficult to voluntarily retrieve these suppressed memories and eventually cause forgetting” (p.3; Stramaccia et al., 2020). To date, the concept of memory suppression still remains highly controversial, and has led to decades of fervent debate within the field of psychology, often referred to as the “memory wars” (e.g., Patihis et al., 2014).

Examining suppression-induced forgetting is not only theoretically relevant, but may also affect clinical and legal outcomes immensely. In clinical settings, the belief in suppressed memory and memory recovering techniques persists (Dodier et al., 2021), but their utility is dependent on whether suppression-induced forgetting is indeed possible. As such, if suppression is not possible, recovering suppressed memories may also not be possible, rendering the therapy itself counterproductive (Dodier et al., 2021). Further, in legal settings, allegedly suppressed or recovered memories previously suppressed could lead to false accusations (Otgaar & Howe, 2018), and potentially wrong convictions (Howe & Knott, 2015). So, there is a need to end this war!

Multiple studies found statistically significant effects of suppression-induced forgetting (mostly in the same-probe test). However, several studies also report mixed results, null findings, and small effect sizes. Reviews and meta-analyses have been conducted (e.g., Anderson & Huddleston, 2012; Stramaccia et al., 2020). However, they may be biased by publication bias, or low power of the individual studies. This is why we argue that a well-powered replication study would be of high interest, not only from a theoretical perspective but also from an applied perspective due to its immense clinical and forensic relevance.



The TNT task consists of three critical phases. First, participants are required to learn stimulus pairs (i.e., cue-target pairs; WAFFLE-MAPLE). Second, participants are presented with the cue item (WAFFLE), and instructed to either consistently think or not think about the target item (MAPLE) of the pair (i.e., think/no-think phase). Third, individuals are tested on their memory in two separate tests (same-probe, independent-probe). In the same-probe test, the cue word is presented, and participants must recall the target item of the pair. In the independent-probe test, individuals are presented with an overarching category item of the target and the first letter of the target (TREE-M), and have to recall the target item of the pair. In both tests, the presence of suppression-induced forgetting is demonstrated by a reduced probability to recall no-think items, in comparison to baseline items that were studied, but not presented in the second phase.


Given that we will solely conduct Bayesian analyses, we will test N=100 and assess the evidential value of the same- and independent-probe effect. We will stop data collection if we have substantial evidence that speaks for or against the effects (BF>5 or BF<0.2). If not, we will add batches of N=50 until we reach substantial evidence or N=200 of included participants.

Data will be collected through the online portal Prolific . Participants will complete one session (approximately 75 minutes), and will be compensated monetarily. The inclusion criteria for study participation are: current student status, highly proficient English speaker, neurological and psychological health (i.e., as this can confound the TNT effect; Stramaccia et al., 2020), and aged between 18 and 60 (i.e., no developmental TNT differences; Paz-Alonso et al., 2009). Adherence to all inclusion criteria will be secured by using pre-screening measures.

Proposed Analysis

Exclusions. We will exclude all participants that have failed to learn all word pairs within the time limit, have not completed necessary parts of the experiment, have failed to accurately react to half of the trials during the think/no-think phase, or have scored above a 4 in the compliance screening. Further, if technical issues arise during the session, the experimenter will note this in the logbook, and these participants will be excluded.

Confirmatory Analysis. The study represents an experimental within-subject design assessing whether directly suppressing a target word leads to a significant lower rate of recall in the no-think condition as compared to the baseline condition in the same- and independent-probe test (two Bayesian student one-tailed paired samples t-tests). Confirmatory analyses will be conducted on the final total sample after listwise exclusions in RStudio (Version 1.2.5019) and JASP (Version 0.14.1).

Pre-Registration & Open Access

This manuscript will be submitted as a registered report to the journal Memory in June 2021; there, the specific procedure will be evaluated in the Stage 1 RR stage. Thus, small alterations to the procedure may be possible. Anonymized raw data, study materials and the analysis code will be made publicly available on the OSF repository upon publication. Pilot data has been conducted and the procedure and analysis plan verified.

Description of Study Costs

For the project, we will potentially test a maximum 200 inclusions; each participant will devote a total of 75 minutes to the study. Assuming an exclusion rate of about 15%, approximately 230 participants will be needed. With an average of £6.60 per hour (£9.50 per participant), the total costs of the test subject funding will be £3059 (£2185 + service fee £728.33 + VAT £145.67).