A case of science correcting itself.

By The Learning and Implicit Processes Lab (From Psychology Today Website)

By Jan De Houwer and Yannick Boddez

Do you remember the movie “Eternal Sunshine of the Spotless Mind”? After a painful breakup, two lovers erase their memory of the relationship but end up falling in love again. Wouldn’t it be great if we could simply erase painful or even traumatic memories? Imagine the suffering that could be prevented. 

You might be surprised to learn that psychologists and neuroscientists are seriously considering the possibility that painful memories can be erased. The starting point is the idea that memories of emotional events in the brain must be “consolidated” (strengthened) before they become permanent. Furthermore, it is assumed that even old, already consolidated memories have to be consolidated again (re-consolidation) each time they are activated (e.g., every time the memory of a traumatic event comes up). Finally, it is assumed that memories are highly vulnerable to disruption in the specific time window during which this re-consolidation takes place. The idea is that this provides an opportunity to permanently change memories with specific laboratory techniques.

If these assumptions are indeed correct, then one could make memory traces unstable and even get rid of them permanently. Some claim that this is no longer science-fiction. In fact, one of the seminal studies on this topic was published in the scientific journal Nature by a group of world-renowned neuroscientists. A great idea in a great journal by a great team of researchers—what more proof do you want?

This is also what Belgian psychologist Tom Beckers thought when he initiated his research project on this topic. His plan was to start from the Nature paper and explore the mechanisms involved, as well as capitalize on some of the clinical treatment applications. However, to his surprise—and to the dismay of his collaborators—they did not observe the memory erasure effect when repeating the original study.

Furthermore, when they requested the data of that original study, they noticed that a statistical re-analysis of these data did not support the conclusions that had been published in Nature either. After four years of painstaking efforts to figure out what was going on, Beckers and his team have now published two papers: one reporting their failure to replicate the original results in a new study and one reporting their failure to reproduce the results using the original data

What can we learn from this story? First, it is always good to wait and see whether findings can be replicated independently in other labs, especially when the results have far-reaching implications. Second, science can correct itself—but that takes time and effort. Third, all of this could have been prevented if the original data had been available at the time of publication. Given the surprising results, scientists would certainly have scrutinized the original data, which would undoubtedly have revealed the problematic nature of those results. Good science is open science. 

Thankfully, we are now living in times where many scientists do practice open science. And thankfully, more and more scientific journals now reward researchers who make the effort to verify existing findings by allowing them to publish their verification and replication efforts. Science is a human activity and therefore potentially flawed, but these flaws can be overcome by making science open and collaborative. 

The authors wish to thank Tom Beckers for feedback on an earlier version.

Parts of the text are adapted from Box 2.7 in De Houwer, J., & Hughes, S., (2020). The psychology of learning: An introduction from a functional-cognitive perspective. Boston, MA: The MIT Press. (copyright of the authors)


McIntosh, R. D., & Chambers, C. D., The three R’s of scientific integrity: Replicability, reproducibility, and robustness, Cortex, https://doi.org/10.1016/j.cortex.2020.04.019