ERPs are voltage fluctuations in the Electroencephalogram (EEG) elicited by the presentation of a controlled stimulus. The latencies of different positive and negative components in an ERP reveal the time course of activation of the neuronal populations that are recruited during the processing of that stimulus. Given its exquisite temporal resolution (within milliseconds), the ERP technique is ideal to investigate fast and transient processes such as the ones that mediate word recognition. Although ERPs have a low spatial resolution and do not allow for a precise characterization of underlying electric generators, components characterized by different scalp distributions are assumed to reflect the activation of different populations of neurons and, therefore, non-identical brain regions.
The figure on the right shows how we can chart the time course of semantic analysis of words while we read single words presented on a computer monitor. The electrode shown in the figure is placed over the parietal areas. Here, we are comparing the brain activity elicited by target words preceded by related (e.g., cat-DOG, blue line) and unrelated words (e.g., sun-DOG, red line). The two lines overlap during the first 300 ms and then diverge: words that are not congruent with the prime word elicit a negative-going component that peaks around 400 ms, called the N400. The time course of this effect indicates that, in these reading conditions, the meaning of a target word is integrated with the previous context starting around 300 ms (this integration process can be faster in other conditions).