Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz

What is so difficult about abstract words, anyways? A neuromechanistic explanation using brain-constrained neural network models

Poster Session D - Monday, March 9, 2026, 8:00 – 10:00 am PDT, Fairview/Kitsilano Ballroom

Fynn Dobler^1,2 (), Lorenzo Stroppa^1,3, Friedemann Pulvermüller^1,2,3,4; ¹Brain Language Laboratory, Department of Philosophy and Humanities, WE4 Freie Universität Berlin, 14195 Berlin, Germany, ²Charité – Universitätsmedizin Berlin, Einstein Center for Neurosciences, 10117 Berlin, Germany, ³Berlin School of Mind and Brain, Humboldt Universität zu Berlin 10117 Berlin, Germany

It is well known that young infants first learn words with concrete meaning, whereas abstract ones are typically acquired later. To obtain clues about the underlying causes and mechanisms, a brain-constrained neural network model with spiking neurons and unsupervised Hebbian learning was trained on instances of concrete and abstract concepts. Before word learning, concrete and abstract concept formation was simulated based on the similarity structure of objects, actions and scenes exemplifying instances of the concept. Subsequently, conceptual instances were co-presented with word forms to establish semantic links and direct grounding of symbols in the world. When instances of concrete concepts (e.g., different DOGs or HAMMERs) were learnt, the network built conceptual representations as neuronal circuits which consistently activated to each of these instances and remained reverberant, thus revealing a mechanism of working/active memory. Such concept formation was impossible for abstract concepts (e.g., FORCE, OBEY). The resultant neuronal circuits selectively responded to their grounding instances, but failed to generalize to the whole concept or to reverberate. When learning a word form for each concept, abstract symbols developed fully functional neural circuits with generalized activation and working memory, comparable with those of concrete symbols. However, we found a delay in forming representations for abstract relative to concrete symbols. Our results suggest that well-known differences in developmental trajectories of different symbol types are caused by structural dissimilarities between the real-world events these symbols are used to speak about. We discuss the underlying neural mechanisms and the psychological and linguistic differences they relate to.

Topic Area: LANGUAGE: Semantic