Abstract
Many models of visual search assume that the visual system uses a salience map during difficult search. This salience map encodes the conspicuity of items and is used both to guide attention to potential targets and to keep track of items that have already been visited and rejected. It has been argued that the latter is achieved by placing an inhibitory tag at the item's location.
Without this inhibitory tagging, difficult visual search should become less efficient and slower, because previously inspected items would be revisited, increasing the time it takes to find the target. Inhibitory tagging of locations, as implemented in models of visual search, leads therefore to a clear prediction: if items change position during difficult visual search, search performance should suffer and search slopes should become steeper.
However, I will report the results from several visual search experiments (using up to 18 items) which show that: (1) Participants do not find it harder to search for a T amongst L's when the items smoothly moved around at velocities up to 10.8°/s than when the items remained static. (2) Information about the moving items accrues over time (3) There is no tagging of the moving items
These results provide a challenge to any model of visual search that uses a fixed location as the index during accumulation and storage of information about search items.