The method feels problematic. The first trials are basically learning that one can associate celebrities with animals. Repeating that same type of task to the same group feels like latent learnings behind the first trial could account for a causal amount of the improvement.
If it feels problematic, it's probably because you are missing the fact that there is a control and treatment group of this experiment to control for what you are saying.
It was stated that the treatment group did the assigmment twice, the first evening without, the second evening with electrical stimulation. Without evidence to the contrary, the memory improvement could be due to consolidation of the stimulus received instead of 'new' memories being stored better.
Although it could be much clearer, the article doesn't actually specify an order. Half the participants did the stimulation night first, per the paper:
>participants were tested during two experimental nights (order counterbalanced): an intervention night and an undisturbed night
They should have controlled for it with a control group taking the two tests in reverse. Or maybe just increase the delay before repeating the second test.
Interesting overall still.
[edit] Full paper here. Issues persist, still cool. https://www.nature.com/articles/s41593-023-01324-5.pdf