A common serial structure causes offline excitability changes linked to generalization between different memory types

Different human memories are predominantly processed within different systems.^1,2 Overcoming this segregation can allow common information to be shared, performance to generalize, and learning to be enhanced.^2,3,4 Yet, how complex serial information is shared between different memories (actions vs. words) is not well understood. Action and word sequences have a common structure when the serial relationship between categories-either of actions or words-is preserved between the sequences. Network excitability increases may link together these different memories, allowing the sharing of information.^5,6,7,8,9 We tested for changes caused by serial structure in motor network excitability following learning of a motor skill during subsequent word-list learning. When the tasks had different structures, there was no excitability change and no enhancement of word-list learning. By contrast, we found that a common structure caused a substantial motor network excitability increase (∼20% from baseline), which was correlated with an enhancement in word-list learning. Excitability within the motor network created a framework for the rapid learning of the word list. It increased specifically when there was sufficient information to form a mapping between motor elements and words. An associative mapping explains how the learned motor sequence was translated into word recall, the magnitude of word-list learning enhancement, when the greatest improvements develop during learning, and how motor excitability became correlated to word-list learning. Thus, serial information generalizes by creating an associative mapping between the different memory types, which enables a sequence of actions to be rapidly translated into and so enhance word-list learning.