Hacker News Reader

Comment by jacobolus | Hacker News Reader

We can’t identify "colors" in isolation either. Color is all relative. If a person had to identify the luminous intensity of a visual stimulus to within a factor of 2^(1/12) [the interval of one semitone], they wouldn't be able to do it. (With significant training and in standardized surroundings it could probably be learned by some people.) And precisely identifying hue/chroma in isolation is just as difficult.

(Note: there is no way to make a perfect analogy about sound vs. color identification, because the physical mechanisms and resulting perceptual spaces are completely different.)

View on HN

Of course we can. Everybody that isn't somehow colorblind can reliably distinguish between a basic number of colors, say Red, Green, Blue, Yellow, Orange, Brown, Green, Purple, Pink, Teal, and to add Black and White allows for all the grays. It's when you start mixing these that naming them is harder because there are many more variations than there are notes on our 'regular' Western scales, from A0 to G#9 if you want to stay within a practical range, and from A0 to C8 if you want to stick to a standard piano, and the way pitches repeat every 12 semitones has no real equivalent in color.

You can also distinguish between a soprano singing vs. a baritone, or a flute playing vs. a saxophone.

It’s when you start trying to distinguish the same type of sound to the nearest semitone that it gets hard (unless trained, ideally from a young age).

I have a trick for that. I search up or down whistling from middle-C and count, that number modulo 12 is the pitch. Of course that only works for the range that I can whistle.

Even on the same instrument, skilled people can often pick out what range it is being played in due to timbre changes. However, this is much more difficult if not nearly impossible when using pure tones like a tuner. These timbre changes can even differ between two semitones depending on the physical properties of the thing producing sound. A B on a trombone is going to sound different than a Bb since B is played in 7th position while Bb is played in 1st position.

Yes, this is very clear on wind instruments where the timbre can change substantially from one note to the next. The saxophone is notorious for this, it is technically a woodwind and it is absolutely unplayable if you don't tune the individual notes as you play them, you have to use your embouchure to get the notes to match pitch. Especially noticeable when playing with other instruments.

https://www.sarahlynnroberts.com/beyond-the-staff/2020/1/30/...

For all we know, there may be an equivalent light octave to the sound octave (mathematically it would make sense). The catch is that the frequency range of visible light falls entirely within a single “octave,” but then if you think about the color wheel which puts red next to violet which are at opposite ends of the color spectrum and it suddenly makes sense.

In fact it would be only about half an octave.

Unless I've done my math wrong, it's roughly a doubling of frequency between the two ends of the spectrum, that makes an octave. From Wikipedia: “A typical human eye will respond to wavelengths from about 380 to about 750 nanometers.[1] In terms of frequency, this corresponds to a band in the vicinity of 400–790 THz.”

How do you explain optical phenomena such as the viral dress phenomenon then?

https://en.wikipedia.org/wiki/The_dress#Real_colours_of_dres...

That is based on the color of the illumination and this of course affects the perceived color. It's the difference between emitted and reflected light, but in the case of a comparison with musical notes it would be fair to only use emitted light.

There is no exact equivalent to reflected light with its own color illuminating a colored drawing. Though it would be interesting to see if such a thing could be constructed somehow artificially using a device that receives sounds and then somehow frequency shifts them before emitting them again. That would be a fun experiment!

> It's the difference between emitted and reflected light

Wat?

Everyone who observed that optical illusion did so by observing it on an emitted light panel.

Color perception is entirely relative. There are countless images that demonstrate this. For example: http://www.optical-illusionist.com/illusions/same-color-illu...

When we perceive emitted light color we’re also perceiving RGB emissions that are blended. I love giant LED panels that when you get close you can clearly see the individual colors. It’s a trip.

Humans are horrible (incapable?) at evaluating absolute color. It’s entirely relative.

Ah ok, that's simply an optical illusion. The brain is full of pre-processing that you can mess with in order to trick it to see things that aren't there and to shift colors around as well as to play with figure-background. But that is a case of 'bad faith', you could do the same for audio illusions, it wouldn't help to draw any further equivalence between the visual and the auditory system.

Both work on the perception of waves with a certain periodic repetition but there the equivalence ends, there is no such thing as 'timbre' in vision, we simply don't work with harmonics there and the shape of the wave in sound is very important and non existent in vision (you can see a single photon in sufficiently dark adapted conditions, your eye as a fundamental particle detector!).

You may want to amend your original statement "Of course we can. Everybody that isn't somehow colorblind can reliably distinguish between a basic number of colors [...]" then, because clearly that's not the case as you state yourself.

I think a Ring Modulator might have some equivalence. Depending on the frequency you set it to the ability to accurately detect the frequency of the input notes can diminish quite drastically.

But then still everyone would have the identical misconception. Clearly, that was not the case.

Part of that was variations in displays. Cheap laptop LCDs especially crush white levels and black levels if not viewed from exactly the right angle.

The weird aspect about our perception of colors is how our brains interpret them relative to nearby colors (e.g. https://en.wikipedia.org/wiki/Optical_illusion#/media/File:C...)

True, but that's optical illusion territory, similar things will happen with music, there are all kinds of acoustical illusions:

https://en.wikipedia.org/wiki/Auditory_illusion

As well as the 'missing fundamental' I linked to in another comment in this thread.

If you pick some Jazz piece apart it isn't rare at all to come across a chord that sounds absolutely awful. But then you play the piece as intended and it all makes sense within the larger context of the notes/chords/intervals around that chord. This never ceases to surprise me.

“Everyone can do it, aside from those who can’t”

There are 88 notes on a piano, but there are at least 2000 pantone colors... I'd be surprised if there are more than 10 people total who can correctly match them all.

The difficulty of both problems is understated.

I'd be surprised if anybody would be able to do the color test.

At the same time: even if people can't tune a piano they can usually fairly reliably tell when one is out of tune.

But more impressive than absolute pitch to me is to be able to identify a four-note chord at once.

A piano is particularly easy to tell if it is out of tune because most notes have multiple strings. They beat against each other horrendously when one of the strings is at a different frequency than the other. If all the strings for a single note were out of tune by the same amount most people would think the piano was fine.

Can most people tell when a guitar is out of tune? A guitar so badly out of tune it plays different notes is recognizable by almost everyone, but a guitar only a little out of tune would not be noticed by most, IME.

While the farnsworth-munsell 100 hue test is definitely doable. I'd love to see a bigger test just for the hell of it.

Even if we can reliably identify 12 unique divisions of spectral color, that's still very different from the 12 semitones in Western music, because the 12 spectral colors would span the entire range of human spectral color perception, whereas the 12 semitones repeat every octave, and humans can hear up to 10 octaves.

> Everybody that isn't somehow colorblind can reliably distinguish between a basic number of colors, say Red, Green, Blue, Yellow, Orange, Brown, Green, Purple, Pink, Teal, and to add Black and White allows for all the grays.

If you're not deaf you should be able to distinguish between wide swaths of the scale too. Especially once you give the swaths names and get used to that.

> It's when you start mixing these that naming them is harder because there are many more variations than there are notes on our 'regular' Western scales

Not if you stick to pure single wavelengths. You could divide that into a mere 50 colors and I doubt people would have a chance at naming them reliably.

Brightness is not related to colours, and while colours are "relative" in the sense that colour perception is influenced by context, it's still the case that you can accurately identify lots of colours in an absolute sense, given a specific context.

You can also accurately identify a lot of sounds, given a specific context.

But if you had to identify colors with the same precision that you expect someone to identify pitches to be considered to have "perfect pitch", it would be very difficult for almost everyone. If you took random chips from the Farnsworth–Munsell 100 hue test, one at a time, and had to give the correct numerical code for each hue, you would not be able to do it. (Which is why the test itself only requires that people put the hues in order when comparing them side by side, not identify each one absolutely.)

I mean you can easily identify 5-10 absolute colour frequencies, but how many can identify even a single absolute tone?

Ordering things in relative order isn't very relevant to this discussion I think, since the point was about the difficulty in detecting absolute sound frequencies. And of course, anyone can put all the notes on a piano in relative order.

Identifying 5-10 absolute colors is similar in difficulty to being able to identify whether a note is bass, baritone, tenor or soprano. Which anybody who knows what those 4 words mean would be able to do.

Except for border or overlapping notes of course. Giving them one of those would be like expecting cyan to be consistently labelled as blue or green.

"bass, baritone, tenor or soprano", that's only four categories, that's much less than 10. And I would guess that timbre will play in here as well, since the human voice is pretty restricted. A bass is not just a frequency shifted soprano.

Play a note and ask someone to reproduce it after ten seconds or so. How close would people come? Then show them a card with a certain pure colour and then ask them to reproduce that with a hue slider.

Don't you think people would come much closer with the colours?

Don't confuse language and perception, we are not talking about labels such as "cyan" here, that's not directly relevant.

Heh. Personal anecdote: at 10 seconds I would do as you expect, but if the gap were longer I would do much better with sound.

I discovered relatively recently that I simply cannot remember colors. My vision is fine, and my short-term memory for colors is also fine. If I'm in a room and you ask me to close my eyes and say what color the walls are, I can do it. But if you ask me the color of the wall of my bedroom, where I go every night, then I will only be able to tell you if we recently painted it and verbally discussed paint colors. (I think it's a shade of blue? Or maybe green. Possibly gray. My family likes to pester me with this question, so you'd think I would memorize the answer at some point, but I haven't.)

I'm kind of curious how common this is. But since I lived several decades without noticing it in myself, I wouldn't be surprised if it wasn't very well recognized.

People will fail the pure color test and that is all there is to it. They will know it was light blue, but won't be able to choose correct light blue. Nad they will sux even more with mixed non-primary colors.

Reproducing visual properties is much harder then you think.

There is no "fail", it's just a matter of how many colours one can distinguish, vs how many absolute tone frequencies.

Infinite amount of both. But there is also issue of remembering which is not as easy as people assume and of comparing.

And of light conditions and context affecting your perception. And so on.

Infinite? Not at all, we’re talking about identifying absolute frequencies here, not just telling them apart. For sound I’d say it’s less than five, for colour perhaps a dozen.

The difference between red and blue is half an octave in frequency. So 4 tones is about right.

I'm pretty sure I'd do a lot better at reproducing the note.

> If a person had to identify the luminous intensity of a visual stimulus to within a factor of 2^(1/12)

Without a unit this affirmation makes no sense.

The visual light spectrum starts at around 380nm, if we arbitrarily assign that to be "C", and we ascend(ascending wavelength, descending frequency) from that with the same 12-tone "equal temperament" used in music we get:

C 380

B 402.595975856532 ~violet

- 426.535578357562

A 451.898703701034 ~blue

- 478.769998960052

G 507.239144584613 ~green

- 537.401153701776

F 569.356689213139 ~yellow

E 603.212399747916

- 639.081275592823

D 677.083025786658 ~red

- 717.344477638087

C 760 infrared

With 760 being one "octave" below 380, though the visual spectrum ends at around 740, which means the visual light spectrum is a bit less than one octave.

But yellow isn't necessarily just a spike at 569.356... There are plenty of other combinations of frequencies that together will stimulate the green- and red-cones enough to create a perception of (nearly?) the exact same yellow.

Right. I put the tildes before the color names to indicate that they are "around here".

If your ear had only 3 types of detectors which only detected 3 specific frequency distributions within about half an octave but could locate stimuli within your field of hearing with pinpoint accuracy, after a lifetime of using that equipment you would probably be able to make relatively fine distinctions in pitch in that very limited range.

Instead, the human cochlea contains thousands of little pitch detectors spread over 10 octaves, and the perceptual architecture and typical training built around it is designed to detect relative pitches (e.g. noticing the difference between two different people’s voices more strongly than the absolute frequency of the fundamental pitch of either voice).

Eyes and ears just have fundamentally different physical mechanisms and we make sense of visual and auditory stimuli in fundamentally different ways. They are not really directly comparable.

In both cases, however, our perception is strongly context-relative.

Reply on news.ycombinator.com