Nice to see more evidence for the thesis that spatial relations are encoded by shifting attention between objects! My colleagues and I found evidence for this also (Holcombe, Linares, & Vaziri-Pashkam, 2011, Perceiving spatial relations via attentional tracking and shifting, "Current Biology"). In the present paper our findings are instead cited as pertaining to the ability to extract the location of an individual object (not its spatial relationship to others), by “filter[ing] processing for one location at a time”.

In our experiments, observers viewed an array of moving objects and we manipulated which object of the relation was attended to first, in a similar spirit to that of the present experiments. The attentional shift theory predicted that attention might sometimes not reach its target before that target had moved on, with attention then occasionally ending up on the trailing disc, causing the participant to mistakenly report the trailing disc as adjacent. The attentional shift theory thus predicted a characteristic pattern of errors for spatial relation judgments - the pattern of errors that we found.