As per cognitive science, all learning are modified brain cells. What we generally call understanding is considered as deep and wide neural network resulting in deep (long tern) memory. What we call as rote learning results on a shallow network.
Note that as per working memory model, both types of memory as useful in solving problems and learning.
deep memory is necessary for critical problem solving (working memory model).
Shallow memory is also useful for developing automaticity and reducing cognitive load which as per researchers helps new learning(cognitive load theory). Think of it as knowing the times tables by heart frees the learners limited working memory resources to help create new knowledge of factorization or simplifying fractions.
References: Cambridge Handbook of Expertise and Expert Performance (Cambridge Handbooks in Psychology)
https://en.m.wikipedia.org/wiki/Working_memory
https://en.m.wikipedia.org/wiki/Cognitive_load
https://t.co/FfIVMcoNd0?amp=1
https://t.co/FuX5iwxEe0?amp=1