Visual search within and across dimensions: A case for within-dimension grouping

We used a visual search task in which participants searched for the presence of either both members or just one member of pairs of spatially separated simple visual features. Pairs of target features could be drawn either from within the same dimension (two colours, or two orientations) or from across two different dimensions (one colour and one orientation). In Experiment 1, display items carrying target and non-target features (target and distractor items respectively) were presented at interspersed spatial locations. Reaction times (RTs) to find both of two feature targets were generally slower than to find one feature target for within- and across-dimension target pairs alike. This slowing was approximately constant across display set sizes, with one exception: for within-dimension targets, two features could be detected as fast as single features when there were just two items in displays. Experiment 2 replicated the finding across all display set sizes when target items were always spatially adjacent rather than interspersed with distractor items. We interpret these data in terms of grouping between target features within the same dimension, and argue against there being any qualitative limit on visual processing or selection when target features come from a single dimension relative to when they come from separate dimensions.