Here we reassess the finding, originally reported by Wheeler and Treisman (2002), that change detection accuracy for bindings of features is particularly impaired by a whole display probe, in which multiple test items are presented, relative to a single probe. Importantly, the different methods of probing visual short-term memory place different constraints on how information regarding features and their combination can be used to respond. In Experiments 1 and 2 of the present work we use simple processing models of change detection to account for the different task constraints and find no evidence for binding specific whole display interference. Experiment 3 then makes a simple alteration to the single probe task which allows for a more direct comparison of accuracy between the two probing methods and also fails to demonstrate binding specific whole display interference. We argue that the original finding is more reflective of how different change detection tasks allow for different uses of information in memory, rather than reflecting a property of visual short-term memory itself.
Dissociations of the number and precision of visual short-term memory representations in change detection
