Investigating melanopsin contribution to scene brightness perception

Abstract

Experiment 1 used a "Green" stimulus composed of LED illumination peaking at 525 nm and 625 nm, and a "Fuchsia" stimulus composed of LEDs peaking at 465 nm and 625 nm. S-cone stimulation from the Green stimulus was only 10% of that from the Fuchsia stimulus, melanopsin stimulation however was substantial (for both stimuli). The hypothesis was therefore that Green would be perceived as brighter than predicted by a provisional brightness model developed by Rea et al. (2011) that only takes into account cone responses, and not melanopsin.

The ability to quantify and predict scene brightness perception evoked by a particular light source opens up the possibility of optimizing lighting solutions for maximum brightness and minimum energy usage. Brightness appears to be related to pedestrians' perceived security in night-time outdoor spaces; thus brightness as a metric becomes a meaningful criterion for outdoor lighting.

Post hoc analysis using data from Experiment 1 and 2, and utilizing previously published (in 2011 and 2013) provisional brightness models as a foundation, showed that the data could be fitted well by modifying the relative S-cone contribution in the equations and including a melanopsin coefficient that increased relative to an increase in light level. This refined model also predicted results from prior experiments. The improvements between this new model and the 2011 model were relatively large, while they were not as large in comparison with the previous 2013 model.

Brightness judgments for the stimuli with higher potential for melanopsin stimulation were indeed higher in both experiments than would be predicted if S-cones were the only short-wavelength-sensitive mechanism contributing to brightness perception, thus supporting the possibility of melanopsin contribution.

Experiment 2 used two "Amber" stimuli, one composed of an LED with peak output at 599 nm (Amber 1), and another composed of LEDs peaking at 520 nm and 635 nm (Amber 2). While both stimuli had negligible S-cone stimulation, Amber 2 did stimulate melanopsin 8 times more than Amber 1 for the same photopic light level. The hypothesis was therefore that if melanopsin were to contribute to brightness perception, Amber 2 would be perceived as brighter than Amber 1.

Human brightness perception is the result of at least three visual channels: The achromatic luminance channel, the red-green (R-G) color opponent channel and the blue-yellow (B-Y) color opponent channel. Short-wavelength output of light sources enhances brightness perception for stimuli subtending large visual angles in the low-to-moderate photopic range (~2 lux - ~150 lux). This is partially explained by a contribution of short-wavelength (S) cones. Recent evidence from experiments conducted with mice as well as humans suggests that intrinsically-photosensitive retinal ganglion cells (ipRGCs) containing the photopigment melanopsin might also underlie increased short-wavelength spectral sensitivity of human brightness perception.

The objective of the present thesis was to investigate whether melanopsin could be contributing directly to human scene brightness perception. Two experimental investigations of the relative potential contribution of S-cones (B-Y opponent color channel) and ipRGCs to scene brightness at 6 light levels between 2 lux and 150 lux were conducted. Subjects provided forced-choice brightness judgments and relative brightness magnitude judgments at two light level ranges, with two different stimuli presented sequentially in counterbalanced order.