Integrating Cognitive Neuroscience Insights into NLP:
                                A New Approach to Understanding Narrative
                                Processing (Abstract)
                                Avital Hahamy1,⇤ , Haim Dubossarsky2,3,4 and Timothy Behrens1,5
                                1
                                  Wellcome Trust Centre for Neuroimaging, UCL Queen Square Institute of Neurology, University College London, 12
                                Queen Square, London WC1N 3AR, UK
                                2
                                  School of Electronic Engineering and Computer Science, Queen Mary University of London, Mile End Rd, London E1 4NS
                                3
                                  Language Technology Lab, University of Cambridge, 9 West Road, Cambridge CB3 9DA, UK
                                4
                                  The Alan Turing Institute, 96 Euston Road, London NW1 2DB, UK
                                5
                                  Wellcome Centre for Integrative Neuroimaging, University of Oxford, John Radcliffe Hospital, Oxford OX3 9DU, UK


                                           Abstract
                                           This paper describes how a biological neural network comprehends narratives, with the goal of applying
                                           these insights to artificial neural networks. To this end, we present our findings, recently published
                                           in Nature Neuroscience [1], detailing a mechanism by which the human brain processes narratives.
                                           Our study utilized functional Magnetic Resonance Imaging (fMRI) to monitor brain activity in human
                                           participants as they were exposed to narratives. The human brain segments continuous narratives into
                                           discrete events that are represented by neural activity. Using a novel fMRI method and a Distributional
                                           Semantic Model, we revealed that whenever an event ends, the brain binds the representation of that event
                                           with the representations of contextually-relevant past event. This suggests that narrative comprehension
                                           is based on the continuous embedding of new events into the narrative context: newly-formed event
                                           representations are updated based on prior narrative events that are uploaded from memory. This paper
                                           not only summarizes our findings, but also advocates for interdisciplinary collaboration: we aim to
                                           inspire the incorporating of cognitive principles into NLP models, which has the potential to improve
                                           the way NLP models understand and process narratives.

                                           Keywords
                                           narrative representation, story evolution, shift detection, brain, movie, story, fMRI, cognitive, reactivation,
                                           events




                                References
                                [1] A. Hahamy, H. Dubossarsky, T. E. Behrens, The human brain reactivates context-specific
                                    past information at event boundaries of naturalistic experiences, Nature neuroscience 26
                                    (2023) 1080–1089.



                                In: R. Campos, A. Jorge, A. Jatowt, S. Bhatia, M. Litvak (eds.): Proceedings of the Text2Story’24 Workshop, Glasgow
                                (United Kingdom), 24-March-2024
                                ⇤
                                  Corresponding author.
                                � a.hahamy@ucl.ac.uk (A. Hahamy); h.dubossarsky@qmul.ac.uk (H. Dubossarsky);
                                timothy.behrens@ndcn.ox.ac.uk (T. Behrens)
                                � 0000-0001-5862-851X (A. Hahamy); 0000-0002-2818-6113 (H. Dubossarsky); 0000-0003-0048-1177 (T. Behrens)
                                         © 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).




CEUR
                  ceur-ws.org
Workshop      ISSN 1613-0073
Proceedings

                                                                                                          149