Abstract
Visual short-term memory (VSTM) performance decreases with set size, but the origins of this effect are disputed. Some attribute it to a limit on the number of items that can be memorized (the “magical number 4”, e.g. Cowan, 2001), others to internal noise that increases with set size (e.g. Wilken and Ma, 2004). We present new experiments and a neural model to distinguish these theories. Observers viewed widely spaced colored discs at fixed eccentricity for 100 ms. After a 1-second delay, one location was marked and the observer reported the color of the disc that had been at that location (the target) by either clicking on a color wheel or scrolling through all colors using arrow keys. A limited-capacity model predicts: 1) an observer's capacity, K, is independent of response modality; 2) when set size N satisfies N≤K, the target color is always reported; 3) any instance of not reporting the target color is due to random guessing; 4) when reporting the target color, response variance is independent of N. Instead, we find that: 1) observers' capacity is 36% higher in the scrolling than in the color wheel paradigm; 2) when N≤K, subjects do not always report the target color; 3) when subjects do not report the target color, they often report the color of another item, consistently with Bays and Husain (2009); 4) response variance increases continuously with N. We confirmed these findings in a two-alternative forced-choice experiment in which subjects indicated for a given test color, which of two marked locations contained that color. Our findings can be explained by a simple neural network characterized by spatial averaging and divisive normalization, without an item limit. We argue that VSTM must be reconceptualized in terms of noise and uncertainty, and that its limitations are likely tied to attentional ones.