Re: Just How Blind is the Human Race?

From: Ian Stirling (root_at_mauve.demon.co.uk)
Date: 06/30/04 

Date: Wed, 30 Jun 2004 23:50:28 GMT

In sci.space.policy Dennis M. Hammes <scrawlmark@arvig.net> wrote:
> Ian Stirling wrote:
>>
>> In sci.space.policy Dennis M. Hammes <scrawlmark@arvig.net> wrote:
>> > Ian Stirling wrote:
>> >>
>> >> In sci.space.policy Christopher James Huff <cjameshuff@earthlink.net> wrote:
>> >> <snip>
>> >> > Well, we really are practically blind. Our eyes are trichromatic
>> >
>> > Monochromatic-with-filters; or "quadrichromatic." But not "tri-."
>> > 11-cis retinal is the /only/ photosensor we've got.
>> > The three oil filters (assorted among the "cones") reduce the
>> > incident light level considerably, not the sensitivity.
>> > Same happens putting color filters on a camera or litho
>> > separations.
>> >
>> >> > sensors, with limited resolution and only capable of giving very crude
>> >> > estimates of color, lightness, and size. We can not see spectral
>> >>
>> >> Looking at the spectral sensitivities, it's amazing there is
>> >> any vivid contrast between red and green at all.
>> >
>> > Heh. Some few people have none whatsoever.
>> > Monochromatic contrast is /all/ in the filters.
>> >
>> >> The two sensors are so similar that the difference in sensitivity
>> >> at any given wavelength between red and green is quite small.
>> >
>> > Actually not; we are rather more sensitive to green by at least
>> > e=hf.
>> > Why subs and planes are set "cockpit red" at night.
>>
>> That's to optimise the low-light sensitive cells, which are different
>> to the ones that are used to sense colour vision, and very insensitive
>> to red.
>
> Merely "less" by e=hf.

True, quite a lot less though, especially if you'r working down near
the tail end of red sensitivity, and comparing it to an equivalent
white or green light.

<snip>
> They're "most sensitive" to "green" because (see your chart,
> again) the "green" filter is the sloppiest, i.e., has the broadest
> passband. Neural response is proportional to number of successful
> (at 11-cis --> 11-trans-retinal) incident photons above the
> infrared, and the green cone is passing a lot of red and blue.
> Rods have no filter and a longer absorption path along which
> /more/ photons will be successful at converting retinal, making the
> /neuron/, not retinal, more sensitive with respect to incident
> levels.
> (N.B.: "Infrared" is /defined/ by retinal's e=hf threshold; it's
> the color whose energy is too low to succeed at the transition.

Which is in itself quite a fuzzy number, only hitting a millionth
way out at over 800nm.

> "Ultraviolet" is defined by the passband of the filters, humors,
> lens, and cornea; water and window glass, e.g., are opaque to it.)
>>
>> Normalising the sensitivity, where 1 is the sensitivity peak.
>> The differences are fairly small, compared to the differences between
>> them and blue.
>>
>> red green blue
>> 658nm .1 .085
>> 600nm .8 .3
>> 570nm 1 .8
>> 554nm .96 .96
>> 542nm .9 1
>> 513nm .5 .7
>> 503nm .32 .47 .1
>> 442nm .04 .07 1
>> 456nm .06 .1 .84
>
> This is a /filter/-sensitivity curve set. Be rather sharper and
> essentially exclusive were they dichroic rather than dye filters,
> too.

True.
Got a kit to upgrade the existing ones?

I was referring only to the dramatic difference between the sharpness of
the red/green filters, and the blue ones, which is really quite dramatic,
and emphasises how much of the red/green distinction is not made by the
filters, but by the processing of their output by the neurons and brain
behind them.

Obvious reasons can be seen for the tradeoff. Compared to having R/G
filters with similar sharpness to the B one, you gain significantly
in lower light visual acuity, though the colour drops out as the
signal-noise from the incoming photons becomes too poor to work out
if something is red or green.