Abstract
We often need to encode relative visual differences between objects. Which of the last two pieces of pie is bigger? Which tomato is redder? Which football team brought more fans? Is Mary taller than Jill? Surprisingly, there is almost no literature exploring this fundamental visual ability. The mechanism that allows us to encode relative differences must solve a difficult problem - the compared features must be bound to the correct objects. Although we feel that we judge relative differences between objects by selecting both simultaneously, this strategy would likely result in binding problems. Instead, we argue that across a range of relative judgment abilities in vision, we select a first object, load its relevant values into memory, select a second object, and then compare the currently selected values with the values in memory. This account makes a strong prediction: that even when making the most simple relation judgment about two objects, selection must shift from one object to another. We tested this idea by seeking evidence of a likely shifting pattern: moving from left to right, as in reading order. When students judged the relative height of two rectangles, they were 52 ms faster when the objects appeared over time in left to right order (200 ms SOA), compared to the when they appeared right to left, suggesting that that the left to right animation was congruent with their natural inspection order. Critically, this advantage disappeared in a control experiment where the relation aspect of the task was removed (a same-different judgment). A second experiment replicates these results, and the control results, using circle diameter instead of bar height.. These results suggest that the binding demands of relation judgments require that objects be sequentially selected.