6000 lux of BLUE light is a _ton_ of photons. The light meter they used applies the human sensitivity curves to the photo frequencies. The sensitivity of the human retina to blue light is very very poor. Most of the time you see graphs of sensitivity vs wavelength the curves have been normalized. Here is one graph where the blue channel has only been inflated by 10x: http://www-i6.informatik.rwth-aachen.de/web/Misc/Coding/365/...

So 6000 lux of blue is approximately 30x as many green photons. For blue LED plus yellow phosphor that 30x "yellow" will be the signal that you will use to decide "that's enough light to perform work."

The rat's eyes were definitely cooked by all of the blue. Imagine if they had used near UV or near infrared LEDs and the same meter to calibrate the intensity. Cooked rats.