Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Neural correlates of word retrieval under visual noise: Interactivity between visual input and lexical processes
Poster Session B, Wednesday, September 30, 4:30 - 6:30 pm, Wangari Maathai
Raphael Fargier1, Giulio Massari1, Fanny Meunier1; 1Université Côte d'Azur, CNRS, BCL, Nice, France
How the processes underlying word retrieval for speech are organized in the brain is still debated. Most recent evidence points to a globally modular system - semantic-lexical processes precede phonological ones - yet with some degree of interactivity and cascading between levels [Pinet & Nozari, 2023]. The precise nature and extent of these dynamics, however, has not been clarified. In particular, it remains unclear whether an input deficit— an alteration in the access to stored representations—extends to all subsequent stages of naming or selectively affects some of them. To shed light on this issue, we manipulated task demands of word retrieval by visually degrading stimuli in a picture naming task. Two previous studies attempted to determine how diminished access to visual information affects picture naming. Results showed that participants were slower to retrieve words if the prompt image featured some form of visual masking, e.g. superimposed circles hampering object recognition [Catling et al., 2008; Laws et al., 2002]. As those studies only considered behavioral results, the neural correlates of these effects are currently unknown. In the present study, we recorded brain activity with HD-EEG while French-speaking participants (N=30) named pictures with or without visual noise. For that, two different levels of a “salt-and-pepper” visual filter were applied to our pictures to create two experimental lists (N=100 each). Stimuli lists were matched on a set of psycholinguistic variables known to affect accuracy and speed (e.g. name agreement, image complexity and variability, familiarity, age of acquisition; lexical frequency, word length, phonological neighbors) and assignment to experimental conditions was counterbalanced across participants. Behavioral results were analyzed in terms of accuracy and response times on correct trials. EEG recordings were cleaned from artifacts and epoched into baseline-corrected stimulus-aligned event-related potentials (ERPs) (from 200 ms prior to picture onset to 600 ms post-picture onset). Results indicated that participants were significantly less accurate (Mnoise=0.76, SD=0.11 vs. Mwithout-noise=0.84, SD=0.09; p<.01) and slower (Mnoise=1058 ms, SD=125 vs. Mwithout-noise=984 ms, SD=125; p<.05) to name pictures with visual noise than pictures without. First analyses on ERPs with cluster-based permutation tests revealed three time-periods of significant differences. Early differences were seen around 105-140 ms after picture onset on occipital channels and corresponded to a reduced amplitude of the P1 component for pictures with visual noise compared to pictures without visual noise. The two other time-periods of significant differences were observed between 380-440 ms on occipital channels and between 460-600 ms on right-lateralized temporal electrodes. Further analyses, including spatio-temporal segmentation, are required to precisely determine whether these later processes have similar durations but delayed in time, or different durations. Our results replicate previous work showing that degrading visual input impacts the accuracy and speed of word retrieval. ERPs suggest that visual noise alters immediate recognition but likely affects later stages of word retrieval as well, including lexical-phonological processes. While visual noise disrupts object recognition, lexical processes may also receive weaker or incomplete activations, thus delaying word retrieval. These findings highlight that interactivity already operates between visual input and later lexical stages.
Topic Areas: Language Production,