Abstract
Purpose:
Spatial summation is important for understanding and modeling the early stages of vision. Optimal summation is stimulus-matched; however our previous work in the dark-adapted fovea suggests independent combination across multiple units of different sizes within a zone of spatial uncertainty. By developing detection models with different summation strategies and comparing predicted behavior with experimental data we investigate the number of different size units and whether detection is nearly optimal given both uncertainty and noise.
Methods:
Detection thresholds were previously measured in the dark-adapted fovea for spots from 0.3' to 25' in diameter (550nm, 10-30ms), viewed through either a 2 mm (conventional optics, 4 subjects) or 6 mm pupil (adaptive optics aberration correction, 5 subjects). Thresholds and psychometric function slopes were compared with predictions from analytic and Monte-Carlo models with different summation strategies given detection uncertainty, Poisson noise, and cone mosaic granularity.
Results:
Average change in log threshold with log stimulus area was 0.12±0.03 for stimuli up to 4-10' diameter, precluding both a single unit size and the optimal stimulus-matched strategy. Data were consistent with spatial uncertainty of at least 25'. Model comparison suggests suboptimal combination across multiple-sized units, where smaller units exhibit relatively more post-receptoral noise, with the smallest unit no larger than the smallest stimulus (~0.5' retinal full-width half maximum), and the largest unit 4-10' in diameter. At least two different unit sizes are required to describe the data. Although many different unit sizes are required for performance approaching optimal stimulus-matched summation, independent combination across two sizes reduces efficiency by only ~0.1 log unit. With a 10' largest unit (roughly parasol receptive field center size) and quantum efficiency of 0.11-0.35, dark noise is estimated at 9-435 events/cone/s.
Conclusions:
Dark-adapted foveal detection as a function of stimulus size is consistent with independent combination across at least two different sized units, the largest and smallest of which are consistent with midget and parasol ganglion cell receptive field sizes. An independent detection strategy across midget and parasol ganglion cells is expected to perform only slightly worse than the optimal stimulus-matched strategy, given noise and uncertainty constraints.
Keywords: 496 detection •
649 photoreceptors: visual performance •
641 perception