Visual performance fields: frames of reference

Abstract

Performance in most visual discrimination tasks is better along the horizontal than the vertical meridian (Horizontal-Vertical Anisotropy, HVA), and along the lower than the upper vertical meridian (Vertical Meridian Asymmetry, VMA), with intermediate performance at intercardinal locations. As these inhomogeneities are prevalent throughout visual tasks, it is important to understand the perceptual consequences of dissociating spatial reference frames. In all studies of performance fields so far, allocentric environmental references and egocentric observer reference frames were aligned. Here we quantified the effects of manipulating head-centric and retinotopic coordinates on the shape of visual performance fields. When observers viewed briefly presented radial arrays of Gabors and discriminated the tilt of a target relative to homogeneously oriented distractors, performance fields shifted with head tilt (Experiment 1), and fixation (Experiment 2). These results show that performance fields shift in-line with egocentric referents, corresponding to the retinal location of the stimulus.

Figure 1. Top: A typical observer's performance fields (center = 0.5, chance performance) with an apparent: 1) Horizontal-Vertical Anisotropy (HVA): better performance at isoeccentric locations along the horizontal (East–E and West–W locations) than vertical (North–N and South–S locations) meridian of the visual field, and 2) Vertical Meridian Asymmetry (VMA), better performance in the location directly below fixation (S) than directly above (N).

Bottom: Hypothesized performance based on egocentric coordinates (tilting the observer's head should result in a corresponding shift in the associated performance fields), versus an allocentric frame of reference (tilting the stimuli should shift performance fields).

Figure 2. Observers viewed briefly presented radial arrays of eight suprathreshold Gabors at four cardinal and four 45° intercardinal locations, equidistant from fixation, and determined the CW versus CCW tilt of a target Gabor (in this example, the target is tilted CCW in the NW position).

Each trial began with the fixation dot presented alone in the center of the display for 1000 ms. Next, the stimulus display of eight Gabors was also presented centered around the fixation dot for 100 ms, followed by a 400 Hz tone response prompt and the fixation dot for 500 ms, and then only the fixation dot for another 500 ms. To dissociate egocentric and allocentric coordinates, each participant viewed vertical (0°) distractors with CW and CCW tilted targets (±60°) and −45° CCW tilted distractors with tilted targets (−105°, +15°) in both upright (0°) and −45° CCW tilted head postures.

Figure 3. Experiment 1 results: Performance fields shifted with the position of the head, not the distractors.

Observers' (n = 4) average performance (in 2arcsin(sqrt(accuracy)) units) for upright (graph on left) and tilted (graph on right) head postures, and upright (solid lines) and tilted (dashed lines) distractors at each of the 8 possible target locations. Note that the SEMs are not shown, as they were too small across all data points to be visually useful.

Figure 4. In Experiment 2, performance in the Fixation Shifts condition was well-predicted by performance in corresponding target locations with respect to retinotopic versus head-centric reference frames.