But what did this test actually consist of? A memory recognition task during which both the words to remember and the words to forget were presented one after the other, in a disorganised and random manner, based on a set that also included 100 control words – words that hadn’t appeared on the computer screen during the learning phase. The participants were asked to push a button when they recognised a word, regardless of the instruction (“To be remembered” or “To be forgotten”) accompanying it during the encoding phase. They had to push another button if they considered that the word presented hadn’t been shown before. Their brain activity was recorded twice by functional magnetic resonance imaging (fMRI): during the encoding phase and during the recall phase (recognition). “The aim was to determine whether different brain regions were activated during these two phases depending on whether the words had to be remembered or forgotten”, Fabienne Collette tells us.

A filter

What were the results of the study? First of all, the subjects recognised 83% of the items to be remembered and just over 50% of the items to be forgotten. The difference was revealing and showed the existence of a "directed forgetting effect". But the core of the experiment, as we have underlined, related to the mapping of brain activations. Let us first look at the encoding aspect. Here we found two interesting results. First of all, the comparison between “encoding” words to forget, which were indeed not recognised, and those to remember, which the subjects didn't remember seeing: in the first case, a higher level of activation in the right middle frontal gyrus, a region associated with the function of selecting. Fabienne Collette: “This region intervenes to sort items depending on whether they have to be stored in memory or not; it turns out to be more active with items to be forgotten, as though its mission was to prevent their in-depth processing insofar as they don’t theoretically need to be remembered.” Furthermore, the activity of the right middle frontal gyrus is associated with that of another region, the right posterior parietal cortex, involved in storing information as such. What did we observe? That this region is very active when we present subjects with words to be remembered and that they will remember them well. On the other hand, activity is very low for words associated with the instruction to forget them. “This means that the filter operated by the right middle frontal gyrus prevents the storage of information that the subjects mustn’t remember”, Fabienne Collette says.