We designed and executed an experiment where the participants had to read (distractor words) or memorize (target words) a list of words while lying inside an fMRI scanner. All words were drawn from a comprehensive curated corpus of the Greek language to match several statistical features of the words included in the auditory verbal learning task (VLT). We selected words to match the words in VLT for their number of syllables, length, frequency of occurrence in the corpus, frequency of their letter pairs, and syllable frequency. The words selected included commonly used words with little distinguishing characteristics.
We recorded brain activation while the subjects were instructed to read distractor words or to memorize the target list of words. We compared the changes in activation of different brain areas estimated using generative statistical models between the two conditions to uncover the set of brain areas associated with learning. Early stages of learning were associated with activation changes in visual and temporal cortical areas of the left hemisphere and the right dorsomedial prefrontal cortex. As the transition to the late phase was taking place, the areas with affected activation included additionally part of the left somatomotor cortex and the right amygdala. The later stages of learning were associated with changes in the activation of the right hemisphere’s precuneus/posterior cingulate cortex.
The changes in activations were accompanied by widespread changes in connectivity. Brain areas playing a key role in connectivity changes throughout the process of learning included both hemispheres pHIP and GP, parts of the right hemisphere’s thalamus and amygdala, and the right hemisphere’s OFC and dmPFC.
The most predictive areas for learning through their changes in activation included frontal areas (left lPFC and PFC, and right frontal operculum/insula), right pHIP and temporal cortex, as well as left post-central and parietal cortical areas. The most predictive areas for learning through their changes in connectivity included frontal cortical areas (left precentral ventral, frontal operculum/insula, and PFC and right dmPFC), left temporal and post-central cortical areas, right somatomotor cortex, both hemispheres precuneus/posterior cingulate cortex as well as both hemispheres’ pHIP, the left GP, the left NAc (core) and the left medial amygdala.
The areas participating in the presumed memory trace following the learning task include early visual and temporal pole cortical areas of both hemispheres, as well as the left inferior parietal lobule and angular gyrus.
Our results are largely consistent with the existing literature although they extend previous results on verbal learning in healthy adults in temporal precision of changes in activation during the learning processes and the maximal inclusion of brain areas in our dynamic connectivity models.
Considered jointly with existing literature on the functional role of the affected brain areas, this evidence suggests that the memory trace of the verbal learning task consists of the engagement of a set of areas related to a process of internal visual narration.
We participated extensively in dissemination activities with the participation of the general public including high-school students, scientific writing in a personal blog and the local media, teaching and student supervision. The project’s software code and data will be publicly released before the publication of our results. The project’s results will be submitted as a pre-print and an open-access peer-reviewed publication.
