Since you are trying to run a controlled experiment, any time between two exposures that is not within your control are likely to introduce many confound variables that could render your results invalid.
For instance, what if the participant had little and terrible sleep over night. What if during the break they attempt using a similar product, or make conscious effort remember something?
I have done user testing in the past where during task 4 users could not repeat a procedure they have manage to carry in task 2 (about 5 minute gap). These sort of failures suggest some fundamental usability issues - but I guess that's not what you're after.
Anyhow, you are likely to gather more valid results if you reduce confound variables, which calls for a single session.
The two types of evaluation
It seems odd to me to carry out a single instance memorability evaluation.
UX evaluation is of two kinds:
- Formative - done during the design process in order to highlight issues and see how different designs affect various measures.
- Summative - done after the design has been delivered to ensure it meets a pre-defined fit criteria.
So the former assumes a repeated experiment after design changes; the latter requires a defined fit criteria. Your method seems to belong to neither.