PUB - Publikationen an der Universität Bielefeld

In visual attention research, a current topic of debate is to what extent visual attention is biased by bottom-up factors (e.g., stimulus saliency), and by top-down factors (e.g., goal contingent orienting). The present work centers on the specific factor of expectation discrepancy, which likewise attracts attention, but does not fit clearly into either of the two categories. The effect of expectation discrepancy is usually tested by first familiarizing participants with a number of search displays containing color homogeneous stimuli, such that they expect the continuous presentation of the stimulus features also for the following trials. If then a single stimulus with a novel color (a “singleton”) is shown unannounced and for the first time in a surprise trial, it captures visual attention and the gaze. Over the course of three studies, the present work demonstrates that a novel feature must not necessarily be presented by means of a novel singleton to attract attention; that is, feature novelty alone is sufficient. The first study shows that a task irrelevant color singleton that was shown in every search trial strongly captured the gaze if it was presented unannounced with a novel color. Furthermore, the study tested an alternative explanation, being that surprise solely interrupts attentional control settings, which causes a reorientation towards perceptual saliency. However, results showed that such an effect does not contribute substantially to surprise capture. The second study yields evidence that surprise capture of the gaze by a novel color covaries with expectation narrowness of the familiar color. It was assumed that an expectation about a color becomes narrower with lower previously perceived color variability and with an increasing number of sampling occasions. Thus, expectation discrepancy of a novel singleton color should be high with a narrow color expectation and low with a broad color expectation. Experiments using a similar paradigm as the first study demonstrated that higher color variability of an irrelevant singleton and fewer familiarization trials reduced surprise capture of the gaze as an indicator for expectation discrepancy. An approach to mathematically model the emergence of an expectation was proposed. The third study shows that novelty can compete with saliency for attentional prioritization. More precisely, it was demonstrated that gaze capture of a novel color singleton in a surprise trial is attenuated if the remaining non-singletons likewise have a novel color, which in turn receive increased attention. The data pattern can be predicted by assuming novelty as an additional source of activation in a noisy priority map for visual attention. Together, the three studies contribute to a more precise specification of the mechanisms that link expectation discrepancy with visual attention.