What Determines the Absolute Threshold of Cone Vision?

Abstract

Purpose: : Accurate modeling of the factors underlying cone detection is essential now that adaptive optics allows psychophysics to be conducted on the scale of individual cones. Unfortunately this knowledge is rather limited. For example, we know that rod thresholds are noise-limited in the sense that subjects can access information from each quantal event by adopting ever more lenient criteria for seeing¹. However, we do not know whether cone detection is similarly noise-limited, nor even what the absolute threshold of cone vision actually is, since previous studies have always utilized very strict seeing criteria.

Methods: : Foveal cone thresholds were measured in 4 subjects with a graded response scheme using a 550nm, 34ms point source (~2.2' full width). Stimuli were presented at 4 intensities (including blanks) and thresholds were determined from psychometric curves constructed for each seeing criterion. Three subjects underwent a similar rod threshold experiment. Both rod and cone thresholds were modeled in a signal detection framework and also compared with forced-choice measurements in a subset of subjects.

Results: : Relaxing the criteria for seeing resulted in a more limited threshold reduction for cones than for rods. The average number of photons at the cornea at rod and cone absolute threshold was ~46 and ~360 respectively. Rod threshold behavior confirmed previous reports¹ but cone thresholds were lower than previous estimates, as expected since previous studies used stricter criteria.

Conclusions: : If cone thresholds, like rod thresholds, are limited by isomerization-like noise they should respond similarly to changes in the criterion for seeing, despite the differences in sensitivity and noise of the two systems. Instead the different response in cone vision suggests an internal limit below which existing sensory information cannot be accessed. In the receiver-operator framework this implies freedom to move along the operator curve in rod vision, but a limit to movement in cone vision. It is this limit, for our data 45-55% higher than the average intrinsic noise, that determines the absolute threshold of cone vision. These results will be discussed in the context of previous physiological and psychophysical studies.1) Sakitt B (1972). J Physiol, 223, 131.

Keywords: photoreceptors: visual performance • detection