Linguistic and attentional factors – not statistical regularities – contribute to word-selective neural responses with FPVS-oddball paradigms

Abstract

In recent years, a fast periodic oddball-like paradigm has proved to be highly sensitive to measure category-selective visual word representation and characterize its development and neural basis. In this approach, deviant words are inserted in rapid streams of base stimuli every nth occurrence (e.g., Lochy et al., 2015). To understand the nature of word-selective representation and improve its measurement, we tested 22 adults with EEG, assessing the impact of discrimination coarseness (deviant words among nonwords or pseudowords), the relative frequency of item repetition (set size or item repetition controlled for deviant vs. base stimuli), and the nature of the orthogonal attentional task (focused or deployed spatial attention). In all stimulation sequences, base stimuli were presented at 10 Hz, with words inserted every 5 stimuli generating word-selective responses in the EEG spectra at 2 Hz and harmonics. Word-selective occipito-temporal responses were robust at the individual level, left-lateralized and sensitive to wordlikeness of base stimuli, being stronger in the coarser categorical contrast (among nonwords). Amplitudes were not affected by item repetition, showing that implicit statistical learning about a relative token frequency difference for deviant stimuli does not contribute to the word-selective neural activity, at least with relatively large stimulus set sizes (n=30). Finally, the broad attentional deployment task produced stronger responses than a focused task, an important finding for future studies in the field. Taken together, these results confirm the linguistic nature of word-selective responses, strengthen the validity and increase the sensitivity of the FPVS-EEG oddball paradigm to measure visual word recognition.