Abstract
Purpose::
The effect of eccentricity on orientation discrimination, one of the classical hyperacuity tasks studied extensively, depends on contrast. Hence, we investigated the simultaneous effect of contrast and eccentricity on orientation discrimination. The threshold of orientation discrimination indicates the smallest detectable change in the orientation of stimulus.
Methods::
Subjects VR and LC (aged 20 and 25 yrs) with normal or corrected to normal vision took part in the experiment. Viewing was monocular using the dominant eye. Three series of magnified versions of Gaussian-filtered line patterns were created at 100%, 30% and 10% contrast, using the software developed by Dr. Risto Näsänen. The ratio between the length and width of the line was 9:1. Stimuli were presented using Pentium 4 computer equipped with a 17 in. screen covered by a black board with a circular aperture of 20 cm in diameter. Two lines of the same size and contrast orientated either vertically or tilted were separately presented in successive intervals. Using a 2AFC-procedure, the subject responded via the keyboard to judge whether the tilted one was in the first or second interval. Auditory feedback indicated whether the response was right or wrong. Viewing distance ranged from 28 to 456 cm to obtain the stimulus sizes needed. Orientation thresholds were measured using a staircase method that estimated 84% of correct. Thresholds were measured at the fovea and four eccentricities, 2.5, 5, 7.5, and 10 deg in the nasal visual field.
Results::
• Equation (1) Th=Thmin (1+H/Hc)5 fitted threshold (Th) curves as a function of line length (H) accurately at all contrasts and eccentricities: goodness of fit (GoF) varied from 93% to 98%. Thmin is the minimum threshold obtained with the line much longer than critical line length Hc. GoF=100%(1-er.m.s.), where er.m.s. = (1/n(logYi(est)-logYi) 2)1/2 is relative r.m.s. error, n is the number of data points, Yi(est) are the estimates from the equation of least squares and Yi are the data.• Equation (2) Fi=1+E/E2+logC/k1+(logC)2/k2modelled both Thmin and Hc. Fi refers to Thmin and Hc divided by the foveal values at 100% contrast. E2 indicates the eccentricity where Fi doubles while k1 and k2 are contrast related constants. Using scaling factors Fi, threshold curves were double-scaled successfully: GoF was 92% for LC and 86% for VR.
Conclusions::
The success of Eq. (2) reveals that: (i) E2 is in fact independent of contrast and (ii) the determination of E2values separately for various contrast levels can, however, show apparent variations with contrast.
Keywords: discrimination • contrast sensitivity • computational modeling