Perceiving the positions of objects is a prerequisite for most other visual and visuomotor functions, but human perception of object position varies from one individual to the next. The source of these individual differences in perceived position and their perceptual consequences are unknown. Here, we tested whether idiosyncratic biases in the underlying representation of visual space propagate across different levels of visual processing. In Experiment 1, using a position matching task, we found stable, observer-specific compressions and expansions within local regions throughout the visual field. We then measured Vernier acuity (Experiment 2) and perceived size of objects (Experiment 3) across the visual field and found that individualized spatial distortions were closely associated with variations in both visual acuity and apparent object size. Our results reveal idiosyncratic biases in perceived position and size, originating from a heterogeneous spatial resolution that carries across the visual hierarchy.

1) Spatial localization biases dataset: Collected from nine observers on 6 separate sessions. Observers matched the location of the cursor to a previous briefly presented noise patch stimulus. There were 5 possible eccentricities on which the noise patch (i.e., the target) was presented. On each eccentricity, there were 48 possibble angular locations, resulting in a total of 240 possible locations. The x-y coordinates of the reported locations were recored and then each reported x-y coordinate was transformed into polar angle.

2) Vernier acuity dataset: 7 observers who participated in the spatial localization task participated in this experiment. Vernier acuity was tested on 8 equidistant locations on the eccentricity of 6 degrees of visual angles. On each trial, participants responded by either pressing left key or right key to indicate the spatial relationship of two lines that were spatially misalligned. To analyze the data, for every Vernier misalignment at a given location, we calculated the proportions in which observers reported that the outer line was shifted more clockwise than the inner line. Then we fitted the proportion of clockwise responses with a logistic function using a least-squares procedure. The just-noticeable difference (JND) was estimated by taking half of the distance between the Vernier misalignments that gave 25% and 75% clockwise responses on the best-fit logistic function.

3) Perceived size dataset: 3 observers who participated in the previous experiments participated in this experiment. On each trial, an arc was presented at one of the 20 locations separated by 18° at an eccentricity of 6 d.v.a.. Upon the offset of the arc, observers responded whether it was shorter or longer than the average. There were six different possible lengths and the mean length (never shown in the experiment) was 15°. To estimate the perceived size of the arc at each location, we calculated the proportion of trials in which the observer responded “longer than average” for each length of the arc. Then these proportions were fitted to a logistic function using a least-squares procedure. The point of subjective equality (PSE), which represents the perceived set mean, was defined as the arc length at which the proportion of “longer” responses was 50% on the best-fit logistic function.