Abstract
We hypothesize that distance judgments between two items in the frontal plane are made on the basis of information recorded in a salience map that registers only salience and not the features that produced the salience. In terms of the number of neurons required for distance judgments, this would be an extremely economical solution. To test this hypothesis, we presented subjects with stimuli containing two target disks for 200 msec followed immediately by an effective random masking field. The target disks were separated by center-to-center distances varying from 1.52 to 20.83 cm (1.49 to 20.21 deg). Three subjects estimated target separation in tenths of inches. Subjects were trained until there was no further improvement and then given complete feedback after each of 1,453-1,463 test trials. The critical aspects were the composition of the background (either filled with 142 distracter disks equal in size to target disks but of various shades of gray to conceal targets, or plain gray backgrounds to maximally expose targets). Target disks were either maximally black or white, purple, gratings of different orientations, or outline circles. The gratings and color could not be discriminated from the background by luminance, only by being different from their neighbors in some important dimension, i.e.., salience. Five pairs of identical targets and 7 pairs of different targets were tested on the 142-distracters background, one different-targets pair and two identical-targets pairs were tested on the plain gray background (easiest possible conditions). Simple result: For all subjects, judged distance accuracy was statistically equivalent for all 15 pairs. Once the two target disks were identifiable, being of identical composition or being easily discriminable from the background offered zero advantage for judged distance accuracy. Conclusion: Frontal plane distance judgments are made on the basis of a salience map.