Semantic, repetition and morphological contexts in visual word recognition: Evidence for multiple and independent feedback loops.

摘要

By hypothesis, awareness is involved in the modulation of feedback from semantics to the lexical level in the visual word recognition system. Knowledge garnered from being aware of the prime [and, by extension, being aware of relatedness proportion (RP)] in a lexical decision experiment in which semantic context is manipulated and many of the prime-target pairs are related (i.e., high RP) is used to configure the system to feed activation back from semantics to the lexical level so as to facilitate processing. When subjects are unaware of the prime (and, by extension, unaware of RP), the default set is maintained in which activation is not fed back from semantics to the lexical level so as to conserve limited resources. Qualitative differences in the pattern of data from two lexical decision experiments that employ masked priming are consistent with this hypothesis. Semantic context and stimulus quality interact when the prime is processed with awareness whereas these same two factors produce additive effects on RT when the prime is unlikely to have been processed with awareness.; Experiment 3 produces a repetition context by stimulus quality interaction, despite masking of the prime so as to prevent awareness and despite a low RP. This finding discounts the possibility that masking, per se, eliminated the semantic context by stimulus quality interaction in Experiment 2. More importantly, this finding also shows that the underlying components of the visual word recognition system that produce the repetition context by stimulus quality interaction are either separate from or operate differently for this type of context than they do for semantic context.; In Experiment 4, semantic context and morphological context are manipulated concurrently with stimulus quality under low RP conditions. Morphology is thought to be represented at the lexical level whereas the repetition context effect reported in Experiment 3 could be attributed to it influencing the letter level, the lexical level, or both. Morphological context therefore provides a more precise measure of how the visual word recognition system operates under low RP conditions than repetition context does. Morphological context and stimulus quality interacted whereas semantic context and stimulus quality produced additive effects on performance. This result is consistent with the claim that repetition context and morphological context interact with stimulus quality under conditions where semantic context does not because they are driven by sub-components of the visual word recognition system that operate independently from the sub-components that produce the semantic context by stimulus quality interaction.; The finding that two sub-components of the same functional system can operate independently from one another also speaks, in concert with other findings, to the distinction between thresholded and cascaded processing. When processing is cascaded, activation at an initial stage activates the next stage prior to processing being completed at the initial stage. This makes it difficult to explain how two sub-components of the same system can operate independently of one another. These experiments (a) illustrate one way that awareness (or lack thereof) affects the dynamics of the visual word recognition system, (b) highlight the functional separability of its underlying sub-components, and (c) provide further evidence consistent with the claim that at least some of its sub-components involve thresholded processing.