Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Cortical reinstatement during learning supports later holistic retrieval of newly learned lexical memories
Poster Session E, Sunday, September 14, 11:00 am - 12:30 pm, Field House
Rebecca Crowley1, Amir-Homayoun Javadi2, Jakke Tamminen1; 1Royal Holloway, University of London, 2University of Kent, Canterbury
Theoretical accounts of lexical representation suggest that newly learned words are stored as multi-element units that bind form and different aspects of meaning together in memory (e.g., Perfetti, 2017). We recently provided the first behavioural evidence for these elements being retrieved holistically, showing that recalling a newly learned word form automatically evokes all discrete elements of its meaning, and recalling one element of the meaning automatically evokes the other elements and the word form (i.e., retrieval dependency; Crowley et al., preprint). A likely neural mechanism for such holistic retrieval is hippocampal pattern completion whereby all elements of a memory are reinstated in response to a partial cue (Marr, 1971). While this mechanism is widely researched in episodic memory, no evidence to date has linked it to the integration of lexical elements during word learning. Moreover, it remains unknown whether reinstatement requires active recall or can also emerge during passive restudy. To address these issues, we recorded EEG from 64 electrodes whilst 38 native English-speaking participants learned 20 novel written word forms (e.g., flimir) paired with a picture (e.g., balloon) and environmental sound (e.g., baby crying), representing the word’s multi-element meaning. These word triplets were designed to reflect how naturalistic word learning integrates information across different modalities. Participants first learned each word triplet through repeated simultaneous exposure to all three elements, and then passively restudied pairwise associations from the word triplets (e.g., balloon-baby crying). One week later, we tested memory for the pairwise associations. EEG data were analysed using general linear models in SPM v12. First, we identified cluster- and peak-level activations specific to each modality. We then assessed whether modality-specific regions were reactivated during restudy trials that excluded the relevant modality (e.g., whether auditory areas activate when participants restudy a word form and its associated picture). Activity specific to word forms was primarily found in the left superior temporal gyrus (z = 5.65, puncorr < .001) with this region also being reactivated during picture-sound restudy. Similarly, activity specific to sounds emerged in the left superior frontal gyrus (z = 9.20, puncorr < .001) and reappeared during picture-word restudy trials. Hence, EEG revealed reliable evidence of cortical reinstatement whereby non-presented elements of a lexical memory were spontaneously reactivated during restudy. Critically, the degree of reinstatement for word forms during picture-sound restudy trials predicted participants’ likelihood of holistic retrieval one week later, indicating that reinstatement during restudy supports long-term lexical binding. The implications of these findings are two-fold. First, they identify a neural mechanism for how word forms and multi-element meanings become integrated in long-term memory. Second, they challenge the view that reinstatement mechanisms are exclusive to intentional retrieval and suggest that educational approaches emphasising restudy may enhance vocabulary acquisition. This work extends theories of word learning and deepens our understanding of how hippocampal pattern completion contributes to lexical knowledge being formed over time.
Topic Areas: Meaning: Lexical Semantics,