Abstract:
The general approach of visual information acquisition into cognition favors a coarse-to-fine order such that lower spatial frequency (LSF) information is extracted earlier than higher spatial frequency (HSF) information. An alternative approach suggests some flexibility in information acquisition: For example, top down processes may modulate the initial usage of LSF (coarse) or HSF (fine). The aim of this study was to compare these two approaches focusing on the construction of visual working memory (VWM) representations. For this purpose, we utilized a change detection paradigm. Intact faces were presented for 100 ms/500 ms durations and participants were asked to memorize these images. Then (after a brief delay), featurally or configurally manipulated distractors (together with identical matches) were presented in broadband (BB), in LSF, or in HSF. The task was to decide which of the probe faces matched the previous target. The results revealed a flexible order in VWM construction. In the first experiment, when upright faces were presented, an initial Configural-LSF and a later Featural-HSF association were observed. Further, both LSF and HSF information were available at a very initial stage, after encoding the stimuli for 100 ms. In the second experiment, when face images were presented in inverted orientations (to reduce the effect of face specific configural information), the performance of observers were reduced to chance level in the 100 ms exposure duration, for all conditions. Further, in the 500 ms condition, they had a tendency to represent inverted faces in HSF and featurally rather than in LSF and configurally. Thus, there was not a fixed coarse-to-fine order in VWM construction in face processing.
