Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Neural Dynamics and Representational Content across Abstract and Concrete Concepts
Poster Session E, Sunday, September 14, 11:00 am - 12:30 pm, Field House
Martina Mancano1, Paolo Belardinelli1, Costanza Papagno1; 1University of Trento, Trento, Italy
While concrete concepts have been argued to rely on both a linguistic and a perceptual format, abstract concept representation has been considered purely linguistic for a long time (Paivio, 1991). However, more recent embodied cognition accounts claim that even abstract concepts rely on experience, namely the emotional one (Vigliocco et al., 2014). Moreover, concrete concepts have been divided into clear-cut categories derived from taxonomic information. While the literature highlighted the existence of different abstract categories (Conca et al., 2021), the distinction of these categories is fuzzy and often relies on experiment-specific measurements (Persichetti et al., 2024; Villani et al., 2019). In our study, we used MEG to record neural responses during written word comprehension, and used Representational Similarity Analysis (RSA) to correlate distributional, experiential, and taxonomic models separately on concrete and abstract word-to-word sensor and source-localised MEG responses. We investigated: (i) which information (i.e., which model) is encoded during concept comprehension, (ii) when, and (iii) where the information is encoded; in other words, the spatial localisation and the temporal unfolding in the brain of concrete and abstract concepts processing. Concrete and abstract words in the visual modality were presented to participants during MEG recording. MEG data was pre-processed, segmented around word triggers, and analysed through RSA. First, the signal elicited by each word, on both the sensor level and the source space, was used to compute word-to-word neural dissimilarity matrices (RDMs), separately for concrete and abstract words. Then, RSA was computed between the neural RDMs and model-derived word-to-word RDMs. The applied models were based on linguistic (distributional model), experiential (sensorimotor and emotional models), and taxonomic (continuous and categorical models) information. RSA results were compared between concrete and abstract concepts. We collected data from 6 participants, and data collection is ongoing. We expect significant correlations between the MEG signal and the distributional model for abstract and concrete concepts in frontotemporal linguistic regions (Hultén et al., 2021; Kaiser et al., 2022). This correlation is expected to precede correlations with the experiential models temporally (Vignali et al., 2023). Concrete concepts signal is expected to correlate with the sensorimotor experiential model in occipital and sensorimotor areas, while abstract concepts are expected to correlate with the emotional model (Vigliocco et al., 2014). Taxonomic models are expected to correlate with concrete, but we do not have strong predictions for a correlation with abstract concepts (Fernandino et al., 2022). This study, thanks to the MEG high temporal resolution, will fill the gap in the literature about the temporal unfolding of the semantic representation of abstract and concrete concepts, which has been scarcely studied until now, while providing new insights into their spatial localisation. By comparing the results between concrete and abstract concepts across multiple semantic models, we will be able to assess the significance of the different organisational principles for each of the two domains: we will observe whether sensorimotor, affective and taxonomic principles contribute to abstract concepts processing too, and how this contribution differs in time and space compared to concrete concepts.
Topic Areas: Meaning: Lexical Semantics, Multisensory or Sensorimotor Integration