Slide Slam B9
Symbolic Representation of Abstract Concepts using the N400 Mismatch Effect
Maurice Flurie1, Jamie Reilly; 1Temple University
Introduction: The representation of concepts using symbols is central to human communication and allows complex ideas and notions to be illustrated with simple letter strings, images, and animations. EEG/ERP has been used to illustrate neurological behaviors that underlie symbol-to-concept matching. Specifically, the N400 response has been shown to be sensitive to semantic expectancies in neurotypical individuals. Many past investigations have relied on pairing highly imageable, concrete stimuli with word or images pairs (i.e., word ‘dog’ with image of ‘dog’). We sought to explore the N400 across symbol modalities (words, images) as they relate to novel abstract concept animations. We examined the N400 mismatch potential in response to five-second animations depicted specific abstract concepts. We hypothesized that ‘mismatched’ words/images that are not associated, or related to target animations would elicit an N400 response, while potentials related to ‘matching’ words/images would remain largely unchanged. Methods: We recorded ERP responses using a Neuroelectrics Starstim-R32 amplifier as participants (N=20) observed 35 animations representing abstract concepts (e.g., competition, freedom) and words/images that were a either a symbolic ‘match’ (e.g., word compete, freedom), semantic ‘associate’ (e.g., word trophy, escape), or ‘mismatch’ (e.g., word explain, mug) of the animation. This resulted in 210 total trials per participant. We analyzed responses of 10 posterior-occipital electrode locations associated with semantic processing. EEG signals were filtered from 0.1-100 Hz. Data were epoched from -100-800 ms relative to word/image onset, and baseline corrected to 100 ms prestimulus. After eliminated artifacts (e.g., eyeblinks) using ERPlab software, this resulted in about 2,000 trials per condition across participants. We conducted a repeated-measures, two-tailed permutation test based on the t-max statistic using a family-wise alpha level of 0.05 to detect reliable differences between matching, associate, and mismatching conditions. Timepoints between 350-550 ms at 10 electrode locations were included in the test. This resulted in a total of 1212 comparisons. 2500 random within-participant permutations of the data were used to estimate the distribution of the null hypothesis. Thus, any differences in the original data that exceeded a t-score of +/-3.29 (df=20) were deemed reliable. Results: Overall, the ‘mismatch’ condition achieved the most negative deflection when compared to ‘match’ and ‘associate’ condition potentials. These differences were illustrated by a left lateralized, posterior topographical distribution of negativity. Specifically, the greatest mismatch effect was identified between ‘associate’ and ‘mismatch’ waveforms at CP6 (p=.04) and Pz (p=.04) electrode locations. Further these differences were found only in the word-to-animation pairings. This finding runs somewhat contrary to our hypothesis, where we predicted the greatest differences would arise between ‘match’ and ‘mismatch’ waveforms. Conclusion: We identified the potential for N400 mismatch responses to illustrate abstract concept processing across symbol modalities (words and images). Specifically, we illustrated a mismatch-effect when contrasting ‘associate’ waveforms with ‘mismatch’ waveforms. These results could suggest a preference for associated words/images (e.g., ‘trophy’ representing competition) over descriptive labels (e.g., ‘racing’ representing competition) in semantic processing of abstract concepts. In all, this work provides empirical support for unique electrophysiology associated with abstract concepts which warrants an expanded investigation.