| |
Abstract:
Anatomical source modeling (BESA) was used to study the
componential structure of the N400 effect to words that violate
semantic expectancies. Some studies (e.g., Curran et al., 1993)
have proposed that the sentence-final N400 may be generated by
delay of the P300 to the incongruous word. This proposal has been
difficult to address, since ERP topographies are plagued by
ambiguities due to superposition of field potentials. Conventional
analyses have used the subtraction method to isolate effects due to
experimental manipulation. However, this approach confounds effects
due to latency-shifting of components across conditions. Source
analyses have likewise tended to focus on "difference waves" for an
isolated range of samples, ignoring residual effects due to prior
activations, such as the P1r, and condition differences in peak
latency of components, such as the P300. The present study used an
alternative method. We modeled the location and time-dependent
activations of components prior to (and following) the N400 effect
for each condition, to determine whether a latency-shifted P300
could account for the difference between conditions. Although
analyses confirmed latency-shifting of the P300, results suggest
that even in this context, a distinct source (the N400) was needed
to account for the difference between congruous and incongruous
words. This source appears to be located in medial temporal lobe,
although cortically constrained methods are needed to confirm this
result.
|
