Abstract:
The early versus late selection issue of attention was addressed via the visual interference paradigm. In Experiment 1 facilitation was found between related picture distractors and word targets. Discrimination of LVF and RVF distractors was tested to determine spatial (attentional) biases. Interference between words appeared to be more difficult to demonstrate in that and the subsequent experiment. Interference effects between attended and unattended words were found in a replication of Lambert et al.'s (1988) third experiment, but only when a post hoc analysis was performed on the data from spatially-biased subjects. The effects occured in the visual field contralateral to the subjects' bias. No relationship between bias and intereference was established in Experiments 4 and 5, although there was some suggestion of an effect for unbiased subjects in Experiment 5. Questions were raised over the validity of bias measures, and the techniques used to determine them. Psychophysical techniques such as ROC analysis and the rating method avoid these hazards and were therefore applied as measures of bias were refined. It was concluded that the subjects' bias was unlikely to have been attributable to a single factor because bias varied with word length for some individuals. The psychophysical same-different task was adapted for words in the final experiment. ROC analysis indicated a RVF superiority for words, and that subjects adopted an optimal decision strategy rather than a differencing one when indicating whether the distractor and target belonged to the same category. This implied a high degree of processing of the distractors, yet no interference effects were found except for the unbiased (no visual field superiority) subjects. This result is contrary to the findings of Lambert et al. and the replication experiment. It was concluded that interference effects are unlikely to be found when distractors are indiscriminable and that discerning whether distractors were unattended or not hinges on the use of accurate measures of distractor discrimination. The results of the present study were considered to support the early selection view.