Processing and encoding of Latin letters and kanji characters in the brain: a visual ERPs study

Multichannel EEG was recorded from human subjects during the performance of a 1-back comparison short-term memory task with Latin alphabet single letters and Japanese single kanji words presented visually. Event-related potentials (ERPs) were analyzed from EEG recordings during two conditions (memory and control). Timing and source localization of different generators taking part in visual information processing and short-term memory encoding were estimated. The main generators of P1, N1, and P2 event-related potentials were estimated in the primary visual, lateral occipital and posterior temporal, and medial parietal regions of the cortex. Furthermore, from later ERPs related to encoding of stimuli in memory, subsequent activations were estimated in anterior regions of the left temporal and frontal cortex between 300 and 450 ms. Definition of few most activated areas in these simple tasks (source localization was obtained from distributed and dipolar fitting inverse solutions) could be used in construction of simple, relatively reliable brain source models. Based on these multiple source models, applying weighted linear inverse filters to the continuous data, useful source signals could be extracted for time-frequency, event-related (de)synchronization and coherence analysis. The observed event-related theta frequency synchronizations and alpha frequency desynchronizations were related to stimulus encoding.