Slide Slam N13
Flexible recruitment of sensorimotor information during language processing: An EEG Study
Emiko J. Muraki1,2, Alison Doyle1,2, Andrea B. Protzner1,2, Penny M. Pexman1,2; 1Department of Psychology, University of Calgary, 2Hotchkiss Brain Institute, University of Calgary
Grounded theories of semantic representation propose that simulations of sensorimotor experience are activated when processing the meaning of a word. These theories are supported by behavioural evidence that words high in body-object interaction ratings (BOI; the ease with which the human body can interact with a word's referent) are processed more quickly than those low in BOI ratings, the BOI effect. Further evidence suggests that this sensorimotor information may be flexibly recruited based on task demands. Tousignant and Pexman (2012) found that the BOI effect in behavioural responses was only present when participants were asked to decide if a word was an entity but not when asked to decide if a word was an action. In the present study we replicated Tousignant and Pexman (2012) with the addition of EEG recording, to further understand the impact of task demands on the BOI effect. We hypothesized that differences in event-related potentials (ERPs) linked to semantic processing (the N400 and P600) would reflect the influence of top-down task demands on bottom-up semantic activation processes. We recruited 90 participants (40 in Experiment 1, 50 in Experiment 2) who completed a semantic decision task during EEG recording. Participants were randomly assigned to one of two task conditions ("Is the word an entity?" or "Is the word an action?"). The same stimuli were used for both conditions; 100 entity words (50 high BOI entity and 50 low BOI) and 100 action words. We used a data-driven multivariate analysis technique to investigate differences in ERP activity while processing action words, high BOI entity words, and low BOI entity words in both task conditions. Our behavioural results replicated those reported in Tousignant and Pexman (2012), wherein a BOI effect was only observed in the entity task condition. Preliminary results from the ERP analysis identified significant latent variables in both experiments that differentiated between neural activity related to action word processing compared to high and low BOI entity words in the action task condition and compared to action words in the entity task condition. These differences were observed as sustained positivity for action words in the action task condition at frontocentral electrodes between 400 - 800ms post stimulus onset. This same difference was observed as a sustained negativity for action words in the action task condition at bilateral temporal and parietal electrodes between 400 - 800ms post stimulus onset. Finally, we observed a more negative-going waveform for action words in the action task condition at occipital electrodes 400 - 600ms post stimulus onset. Our preliminary results suggest that differences in task demands are evident in associated neural activity, however these differences do not mirror those observed in the behavioural data. Action-related words were related to differences in amplitude in the N400 time window relative to high and low BOI words in the action task condition and compared to action words in the entity task condition, suggesting that action-specific information was recruited preferentially in the action task condition.