Abstract
The task of detection requires that at least one target component (i.e. “feature”) be detected, while the task of identification requires the detection and integration of multiple features into a recognizable object. Enumeration seems to be an intermediate stage between feature detection and object identification. As in identification, it requires the detection of multiple features, but as in detection, it does not require the encoding of relative feature positions. As we count 1,2, 3…10, enumeration inherently seems to be a serial detection of items. However, when we enumerate briefly presented objects, we have a capacity to accurately report 4–5 items. Enumeration within this range is called subitizing.
In subitizing, do we automatically spatially integrate features? Feature detection is dependent on contrast. As the visibility of the features is decreased the number of activated feature detectors is less. If subitizing only requires detection, then as we decrease the contrast, the subitizing capacity (4–5 items) should also decrease. However if the subitizing capacity is independent of contrast, it must be processed by a second stage of integration that is independent of the first stage of feature detection.
In order to determine whether subitizing requires feature integration, we measured accuracy as a number of Gabor patches presented at different contrasts. The Gabors were presented for 50 ms along an imaginary circle at 5 deg viewing eccentricity. We found that the maximum accuracy changed with contrast, but the subitizing capacity (4 items) is immune to the effects of contrast. This means that subitizing requires feature integration.