Talk:Opponent process/Archive 1

This is an archive of past discussions about Opponent process. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 1

Long ago

I just re-discovered the opponent color page which I created long ago, based on the work of Koenderink and Jan-Mark Geusebroek on the "spatial color model" (opponent color model + Gaussian scale space).

The reason for creating the article it was the use of the "spatial color model" in the analysis of light microscopy images (http://ourworld.compuserve.com/homepages/pvosta/proc_rms.pdf). This is all in the past now.

Merge

I've redirected Opponent processes to here. The following is the only paragraph that originated at that location, but I'm not sure what, if any, of it fits in this article. Pasting below just-in-case (and striking out the obviously irrelevant parts). --Quiddity 20:59, 25 August 2006 (UTC)

Opponent processes are observable in neuro impulses (Garbor wavelets), similar to qualia in philosophy. These include color illusions of seeing an opposite color after habituation, physiological homeostasis reactions for temperature, oxygen, food, and stimuluation control. Drug addiction is best understood as being due to the modulation of a homeostatis into a positive and negative phase. Any stimulus, drug, or behavior will lose its strength due to learned habituation, but the opponent reaction to the original drug remains strong. In terms of drugs this means that eventually, people who started to take heroin to experience pleasure will be taking it to feel neutral.

I can see why you deleted the bit about drug addiction... it doesn't fit with the opponent process theory of color perception, but there is an opponent process theory of addiction. I'm considering starting a new article to cover that topic. As such, I wonder if the name of this article should be made more specific to better reflect its actual content? Jesserjames (talk) 17:30, 13 March 2012 (UTC)

Reference

Is their any reference to the claims of the last section. Cells receiving info fro M, L, S and ML?BartYgor 01:32, 17 July 2007 (UTC)

Which is the "last section," and which claims are you unsatisfied with? --jacobolus (t) 22:19, 18 July 2007 (UTC)

Sorry for the confusion: The article states: "Parvocellular cells, or P cells, handle the majority of information about color, and fall into two groups: one that processes information about differences between firing of L and M cones, and one that processes differences between S cones and a combined signal from both L and M cones. The first subtype of cells are responsible for processing red-green differences,and the second process blue-yellow differences. P cells also transmit information about intensity of light (how much of it there is) due to their receptive fields." That all kinds of biological processes take place or that some are linked to are vision doesn't supprise me. But that some were exactly mached (not some assumption) to some particular part of our experience (after all that is what color is) I would like to see referenced. Thanks --BartYgor 15:03, 20 July 2007 (UTC)

The words red, green, blue, and yellow sometimes refer literally to the perceived color. When this is the intended meaning, the color names are preceded by the words "unique" (for only for these four hues). For opponent hues (as indicated by hue-cancellation experiments) they are named "opponent-blue," etc. The opponent hues are off by quite a bit. In neuroscience, red, green, and blue are naming conventions referring to the L-, M- and S-cones, and the data channels into which they feed, with "yellow" representing combined L+M. I have a reference for this (with a separate link to the article itself), but am not sure where to put it. article

^[1]

Zyxwv99 (talk) 15:58, 13 January 2015 (UTC)

References

[1]
Conway, B. R.; Livingstone, M. S. (2006). "Spatial and Temporal Properties of Cone Signals in Alert Macaque Primary Visual Cortex". Journal of Neuroscience. 26 (42): 10826–10846. doi:10.1523/JNEUROSCI.2091-06.2006. ISSN 0270-6474.

Comparison to YUV?

Stable Reddish-green

Antagonistic colour responses - more citations needed

Image

"Other uses" section

S cones contributing to R-G opponency

The article currently says, and has a diagram showing, that the R-G opponent process only uses the outputs of L and M cones. I think that S cones also contribute on the R side, because:

Many diagrams show them contributing on the R side, such as Fig. 2 of Hurvich and Jameson (1957).
This article, which I have trouble understanding, appears to show R-G opponent midget cells getting input from S cones.
Input from the S cones seems necessary to explain the fact that light at the violet end of the spectrum is perceived as containing a significant amount of red.

Am I right? Can we change this? -- BenRG (talk) 20:02, 23 May 2015 (UTC)

I think this comes too close to original research. The reason violet looks like blue with some red added has already been figured out. Something to do with the fact that all three cone sensitivities decline in the lower wavelengths, but not at the same rate, so relative sensitivity comes into play. Also, it's not just cone fundamentals, but signal amplification and various adjustments. The point is, we can't just make up our own answers. Hurvich and Jameson (1957) is pretty old. It's a good theoretical model, but not to be taken as literal physiology. As for the third paper, it's interesting but in this case I think we should really stick with the standard model, which is pretty current. Zyxwv99 (talk) 20:20, 23 May 2015 (UTC)

Here's a nice open-access article from 2005 that covers this pretty thoroughly: "The cone inputs to the unique-hue mechanisms" by Wuerger, Atkinson and Cropper. It says that the opponent cells in the retina compute L−M and (L+M)−S, but that the unique hues are substantially inclined from those axes, meaning that there's another colorspace transformation happening in the brain. So I think the real problem with the Wikipedia article is that it wrongly conflates the L−M and (L+M)−S opponent calculations in the retina with the psychological opponent colors. The S cone does contribute to the sensation of unique red.

By the way, please don't accuse me of original research when I cited sources, you cited none, and the paragraph I disputed is marked "[citation needed]". -- BenRG (talk) 00:54, 24 May 2015 (UTC)

Your point number 3. about why we see violet as purple (blue-blue-red) looks like original research to me. The scientific community has already figured that one out, and it's something completely different than what you suggest.

As for point number 2., the source says, "The S-cone system is rather different from the M- and L-cone systems from the initial level of the cone itself to its circuitry though the retina and into the brain. We do not consider the S-cone pathway as a midget pathway (see later chapter on S-cone pathways). So this chapter is devoted to the M- and L-midget pathways and as we shall see, they carry both a resolution and a chromatic message through the retina to the brain." Furthermore, any basic text on how the eye works can tell you that the neurons that connect to cone cells can't tell M and L cones apart because they are the same except for a single molecule, the opsin, which is at the other end. The neurons can, however, distinguish between L/M and S cones. Midget cells connect to M/L cones.

Unique red is an impossible color that no one can ever see. The closest you can get is ~704nm, which stimulates the L cones 15 times as much as the M cones and thus looks like 15/16 red and 1/16 green. At higher wavelengths the sensitivity of the M cones decreases at a slower rate than the L cones, causing hue to shift back (slightly) towards orange. 15/16 red plus 1/16 green equals 7/8 red and 1/8 yellow, or 3/4 red and 1/4 orange. If you add just the right amount of blue, if will cancel out the 1/8 yellow, but at the cost of reducing saturation, since blue and yellow cancel out to white. Thus, 100% saturated unique red is an impossible color, while somewhat desaturated unique red is an extra-spectral color.

The unique hues come in three versions: retinal (at which point they are pretty screwy-looking, with lime, something like "hot pink" (desaturated megenta-red) etc. Some of this gets adjusted in the LGN, some further downstream. However, they're still trying to figure out how the adjustment process works.

I'm familiar with the work of Hurvich and Jameson. It's tremendously important, since it's the foundation of this kind of research.

As for the most recent paper you cite, "The cone inputs to the unique-hue mechanisms," I've only had time to read the abstract, but it seems consistent with what I've been reading, i.e., they had a feeling that the color correction wasn't all in the LGN.

Please feel free to make any edits you want. I was just expressing an opinion. Zyxwv99 (talk) 22:46, 24 May 2015 (UTC)