Abstract
Traditional accounts of working memory are divided into two irreconcilable camps: memory is either thought to be capacity-limited to 3 – 4 items, or to be resource-limited as the number of items grows. Here, we show that certain computations – namely identifying the maximal and minimal element along a dimension such as length – is neither capacity- nor resource-limited: i.e., finding the maximal or minimal element in a scene is automatic, efficient, and effortless. In three separate experiments, observers are shown 5 – 7 colored lines on the screen (e.g., a blue, a yellow, a green, a purple, a red, a black, a white) for 1200 milliseconds. In Experiment 1 (N = 40), participants are asked to either perform a pairwise comparison (e.g., "Is the blue line longer than the yellow line?") or a maximal comparison (e.g., "Is the blue line the longest?"), thereby requiring them to attend, remember, and compare all seven lines. We find that the maximal comparison is faster and more accurate that the pairwise comparison, even though the computation requires the representation of all lines (Fig1). In Experiment 2 (N = 80), participants identify the color of a particular line in the sequence (e.g., "Which line is the second longest?"). We find a pronounced advantage in accuracy and RT for identifying the longest and shortest lines, with an increasing, serial cost to each successive line in the sequence (Fig2). Finally, we replicate these effects developmentally and find that the identification of the maximal element is easier compared to pairwise from at least age 2 onward. These results suggest that the computations supporting the identification of the maximal element are distinct and more efficient than those supporting the identification and comparison of only two items, providing a challenge to traditional views of working memory capacity limits.
Meeting abstract presented at VSS 2017