Different sources of predictions during natural reading: a co-registration study

B. BIANCHI; J. KAMIENKOWSKI; D. E. SHALOM

Congreso; Society for Neuroscience; 2019

SfN

During reading our brain predicts upcoming words. This predictions, if correct, could speed up theprocessing of the word when it is finally read. It have been amply shown that Predictability (the variablethat estimates the probability of guessing the next word) have an impact on how we move our eyes acrossthe text: more predictable words are fixated for shorter periods of time than less predictable words.Additionally, EEG experiments showed that the amplitude of the N400 potential is modulated by thisvariable: more predictable words correspond to less N400 amplitude. Classically, EEG experiments wereperformed avoiding eye-movements, because the muscular contraction interfere with the brain electricalsignal. During the last years, helped by the advances in computational power, co-registration experiments,in which eye movements are allow, generated more natural contexts to study such a complex process.In the present work we analyzed brain potentials during natural reading to address for the sources ofpredictions. Previously, in independent studies, it was shown that mnemonic predictions (i.e. predictionsperformed purely on long term memory, like when reading a proverb or a son lyric) and predictions donepurely on the linguistic context have different impact, both on gaze duration and on the N400.Categorical analyses of fixated-word-Predictability showed a N400 effect. A late effect on fixated word isalso seen in correlation with the next word Predictability (parafoveal-on-foveal effect). The categoricalanalysis of the interaction between Predictability and Sentences Type also showed slight differences.Finally, analyses performed using nonparametric cluster based statistics and Linear Mixed Modelsreplicate effects seen in the previous study on fixed gaze.