| |
Abstract:
Upon seeing a familiar person retrieval of semantic knowledge
is usually faster and less error prone than recalling the name.
This is accounted for by two competing classes of models. According
to sequential models access to names is dependent on and follows
access to semantic information whereas parallel models propose
proper names to be accessed in parallel with semantic knowledge but
to be disadvantaged by their specific representations or
connectivities. Experiment 1 investigated the time course of naming
and categorizing familiar faces (politicians) by recording response
times (RT) and the lateralized readiness potential (LRP) as a
measure of central response activation. A 2-choice go/no-go
procedure required easy or difficult semantic (foreign/domestic
vs.government/opposition) and easy phonological (inititial phoneme)
classifications.When semantics specified response hand and name
phonology determined whether to respond or not, the semantic
difficulty affected RT and LRP onset latency. However, at variance
with serial models, there was also an effect of semantic difficulty
on the interval between LRP onset and the response. In a second
experiment, we also manipulated phonological difficulty (initial
phoneme vs. name length) which, according to serial models, should
be additive with semantic difficulty in RTs. However, the effects
of semantic difficulty for the easy phonological task were not
maintained but abolished by increasing phonological difficulty. The
present findings are at variance with rigidly sequential models and
provide support for parallel processing.
|
