Academic journal article Perception and Psychophysics

Orientation Discrimination across the Visual Field: Size Estimates near Contrast Threshold

Academic journal article Perception and Psychophysics

Orientation Discrimination across the Visual Field: Size Estimates near Contrast Threshold

Article excerpt

Performance in detection and discrimination tasks can often be made equal across the visual field through appropriate stimulus scaling. The parameter E^sub 2^ is used to characterize the rate at which stimulus dimensions (e.g., size or contrast) must increase in order to achieve foveal levels of performance. We calculated both size and contrast E^sub 2^ values for orientation discrimination using a spatial scaling procedure that involves measuring combination size and contrast thresholds for stimuli with constant size-to-contrast ratios. E^sub 2^ values for size scaling were 5.77° and 5.92°. These values are three to four times larger than those recovered previously using similar stimuli at contrasts well above detection threshold (Sally & Gurnsey, 2003). E^sub 2^ values for contrast scaling were 324.2° and 44.3°, indicating that for large stimuli little contrast scaling (.3% to 2.3% increase) was required in order to equate performance in the fovea and the largest eccentricity (10°). A similar pattern of results was found using a spatial scaling method that involves measuring contrast thresholds for target identification as a function of size across eccentricities. We conclude that the size scaling for orientation discrimination at near-threshold stimulus contrasts is much larger than that required at suprathreshold contrasts. This may arise, at least in part, from contrast-dependent changes in mechanisms that subserve task performance.

The idea that performance can be equated across the visual field through an appropriate scaling factor was originally proposed by Rovamo et al. (1978) and Koenderink, Bouman, Bueno de Mesquita, and Slappendel (1978). In these studies, stimuli were scaled in inverse proportion to the size of the proposed cortical neural projection area, a procedure known as M-scaling. It was assumed that performance for all tasks could be made equal across eccentricities by the use of a single set of predetermined scaling factors, one for each principal meridian of the visual field. Later research, however, suggested that the amount of peripheral size scaling required to equate task performance depends on task demands (Klein & Levi, 1987; Levi et al., 1985; Westheimer, 1982). To overcome limitations associated with M-scaling, a procedure known as spatial scaling or S-scaling, was introduced (Johnston, 1987; Johnston & Wright, 1986; Saarinen, Rovamo, & Virsu, 1989; Watson, 1987; Wright, 1987). The technique makes no a priori assumptions concerning the size of peripheral magnification factors. Task performance (e.g., proportion correct responses or discrimination thresholds) is measured for a set of stimulus sizes at each eccentricity and is then plotted as a function of stimulus size. If size scaling alone is sufficient to overcome the peripheral sensitivity losses, performance-versus-size curves at all eccentricities will have a similar shape on logarithmic axes and will be laterally shifted versions of each other. The amount by which the peripheral curves must be shifted on the size axis in order to superimpose all data determines size scaling.

Implicit in spatial scaling theory is the assumption that achieving optimal performance at each eccentricity requires an appropriate match between stimulus dimensions and the mechanisms they engage. The major limitation on peripheral performance is assumed to be an eccentricity-dependent variation in the spatial scale of the underlying neural mechanisms. According to Watson (1987), this amounts to assuming that visual processing is homogeneous throughout the visual field apart from a change in the scale of the local mechanisms at each peripheral location; the parameter E^sub 2^ is thought to reflect this eccentricity-dependent scale change.

Mäkelä, Näsänen, Rovamo, and Melmoth (2001) argued that size-scaling estimates may be erroneous if one does not explicitly consider the role of stimulus contrast. They had subjects perform a face-discrimination task and measured the contrast required to perform the task as a function of stimulus size at a range of eccentricities. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.