Categories
Questions

Question #32: Does the F*cking “Colour” of the Light Change When We Increase or Decrease the Quantity of Light?

As we look back to the past two questions, it is hopefully more evident that no simple path as forwarded by “stimulus-as-colour” can aid in charting us successfully through the perils of the colour iceberg laden waters. The worst part is that the USS Colourimetry vessel is chocker full of flesh eating obsessed Colourimetric Stimulus Zombies on a hunt for brains to eat. In order to survive these perilous waters, we are going to need to finally jump off the ship, and swim to the USS Computational Neuroscience.

This post marks that moment where we plunge into the waters. It is with hope that the cold waters don’t claim the reader, but after all, anything is better than having one’s brain eaten alive in the Colourimetry Zombie Apocalypse horde.

Grab onto the life jacket…

So let’s revisit Question #3, and reframe it…

Question #32: Does changing the intensity of a single RGB light change the cognition of colour?

This question has many colourimetric-oriented (Hint: Stimulus) non-answers that are common “orthodox” knowledge, and in turn totally bunko nonsense from the lens of actual colour cognition. Even if a Colourimetry Zombie acknowledges the complexity, it tends to cease with a short non-answer grunt being “Changing the intensity of the stimulus changes the colour!” and the zombie lumbers on their way, clamouring to eat your brain. They might cite numerous “effects” or “illusions” such as the Bezold-Brücke shift, the Purkinje effect, or any of another dozen or so peculiarities around “intensity” and the cognition of colour.

Let’s smack open some of those Zombie heads, and in doing so perhaps also learn some Computational Neuroscience self-defence to even manage to repel the Brand Identity Stimulus obsessed animated corporate corpses…

First… we would do well to contextualize the prior questions and discussions around “intensity” as a unit of stimulus. Under the lens of what we discussed in Question #31. A few pieces of relevant information are worth reframing our discussion to one of colour, or the cognitive interpretation of colour qualia:

  • The bulk of the processing in the retina appears to be differential in nature12. That is, we can’t discuss “scalar” values as “magnitudes” from “zero”.
  • Only an extremely limited, somewhere near 1% of cells appear to provide a local mean scalar value of luminance2.

This understanding means that we can’t specifically say that “intensity” is this common and misleading notion of “scalar” value in our colour cognition. Rather, it is more grounded to suggest that “intensity” is something else, and is relational in nature.

An emergent facet of this slippery “intensity” means we must take care to understand that “intensity” cannot possibly exist along some fictional polemical ruler if the ruler anchor point keeps moving! Rather, this paints a more fluid dynamic that emerges out of the spatiotemporal articulation. As organisms, we are literally ill equipped to “measure” “intensity” along a scalar ruler the way stimulus is all too frequently quantified, and conflated with colour! To this end, the specific intensity we are referring to is a notion around lighter and darker. Emphasis on the suffix.

What does this mean in a practical sense? Usually when the subject of a relational / differential foundation around visual cognition pops up, the common response is to nod one’s head and carry on. To really affix this mind blowing concept in a visceral manner, I typically refer folks to one specific effect that is so jaw dropping it needs to be experienced. The effect is known as Induced Contrast Asynchrony, and is best experienced in real-time.

With that nightmare demonstration experienced and in the memory bank, we can explore the idea of relational differentials in some very simple still pictures, which echo the same brain pop as the Induced Contrast Asynchrony effect4. So go grab a couple of litres of LSD and hop on board the train…

First… let’s provide some stimulus. Here, we have “a disc”.

At the moment we have some spatiotemporal articulation, our meatware goes into overdrive of meaning-making. Even though we could say “This is a disc”, we should at least realize that identifying the boundary conditions has to happen at some mechanical level in our neurophysiology. From the web page assembly, we are identifying the “This is a boundary of the picture” and then a meta decomposition into “I cognize5 further boundaries, and now I think I cognize a single form as a disc.”

I can already hear the screeching Colourimetry Zombies jumble mutter “Why are you bothering to ramble on about something so trivially obviousszzzz!!11!??” Someone, quick… grab the chainsaw!

What is interesting is that with even a small degree of questioning, we could make a case that our disc is fundamentally unstable. That is, our differential inference meat engine is always engaged, and in a constant state of flux in terms of the underlying interpretations of meaning. Is the disc figure6 “in front of” the ground? Is the disc figure “behind” some form, such that the surround is “on top of” the disc? None of these inferences as information units are present in the stale bread stimuli presented in front of us; we must form these information units.

Take a moment to look back at the disc and try to cognize the disc in these sorts of orientations…

Did you notice anything peculiar happen?

If one successfully focuses on the picture for a moment, it is quite possible that a Necker’s Cube-like modulation of state will occur, by which the disc might flow into and out of “lighter” or “darker” depending on the phase of our cognitive modulations. And of course this might not appear to happen immediately for someone, as we all approach our inference formation by way of learned experiential biases.

The following picture is an attempt to describe the possible topmost level of a scission decomposition processing that our meatware inference engine might be able to dream up probabilistically.

Note that if our probability starts at two “highly probable” configurations, we end up with very different ideas of the colour of the “disc” form.

In the left hand case, the disc could be considered “on top” of the “ground”, and therefore in one configuration, it could be occluding the “ground” in some way. This leads to an interception where we might consider the disc “as is” in terms of “lightness”.

In the right hand viable inference, the disc is “under” the “ground”. In this inferential context, the “disc” is “occluded” by the “ground” in some way, and as a result of the interception of probability, the disc must be “lighter than” the inference of the prior “on top” configuration. I tried to match the disc tristimulus in both examples to the resulting picture.

It is reasonable to suggest that these implicit instabilities are omnipresent however, as hopefully the following demonstrations will show.

With this under our experience belts, let’s try to examine the picture again and see if we can discern a peculiar sense of “lightness” that may be incredibly elusive and unstable.

And with that, let’s return to Dr. Tse’s completely unstable discs. These discs add a little more statistical probability ingredients that our meatware inference engine can form information from. Are we able to discern the instability of which disc is “lighter” versus which disc is “darker”? Remember, the “forms” of the “discs” are identical tristimulus with respect to the “other disc” form.

And here is a slightly different variation, with the polarity flipped on the “discs”. Can we modulate the “lightness” of the “separate” discs?

Already our discussion of “intensity” has at least hinted that all is not well in terms of potential “matching”. Within the prior demonstration, the disc “exists” as the cognitively inferred state of the form, not as the stimulus as measured scalar value. That is, we cognize the qualia of the disc, “outside” of the context of the direct stimuli.

Let’s make that situation worse…

Here, we have “two” discs now, on a slightly graduated “ground”.

Notice how even though this is a rather low spatiotemporal articulation, the “lightness” complexity of instability of each disc can cascade down the probability inference chain. As such, we end up with two unstable disc lightnesses, each modulating based on cognitive inference state.

And what might we expect to happen if we try to “match” one disc to the other?

What is absolutely incredible about this configuration is that if one makes this in their content creation application of choice, and varies the magnitude of the stimulus of one of the discs to “match” the other disc, no magnitude will likely be satisfactory78910. This is an incredibly little known phenomena, so please… share it around when someone discusses the idea of “colour matching”. More importantly, try it yourself!

This is arguably the first and most challenging entry session at the Visual Cognition Dojo. I know personally it was one of the most gargantuan leaps I had to make from the sinking cruise liner littered with Colourimetry Zombies hunting for more brains to eat. I hope that the basic premise, that of visual cognition instability, has been presented here in a way that is digestible. With a little luck, it will take hold, and more potential victims of the Colourimetric Zombie Apocalypse can be rescued by way of a thrown life line.

So where do we go from here?

Well first up, we have to answer the question. Note that we are going to answer the question with respect to “lightness” and “darkness” as a subset of the broader idea of “colour” in terms of qualia.

If one experiments with the above articulation, and concedes the point that the disc lightness or darkness is impossible to match to the other disc configuration, we will immediately have way more questions than what we started with.

A key question here is why does the articulation possibly lead to this completely unstable set of interpretations? Why does the cognitive processing of such a “simple” spatiotemporal articulation lead to such a confounding result? No one knows for certain, although the research around these things has been going on for at least two centuries. This research falls squarely in the field of computational neuroscience, which likely saw a sizeable chunk of the field begin in earnest with folks like Horn111213.

Some things that are worth identifying patterns of might be:

  • The “interiors” of the discs are propagated from the boundary conditions14, likely resulting in something akin to a “cognitive buffer”.
  • The top-down processing works relentlessly to scise or fission the “cognitive buffer” into forms, and their cascading respective underlying “meaning”1516171819.

If we do engage in a cognitive scission / fission mechanic, decomposing the forms into their cognized qualia “outside” of the contextual information of the stimuli, there is a dilemma! There are multiple potential interpretations of the low articulation of “disc” and “ground”, and that depending on the inferential frame, there are no absolute “solutions”. Anyone who tries to offer us Cartesian “model” implying a stimulus to colour qualia “appearance” or “uniformity” is outright lying to us. In their defence, they were sadly probably infected through casual contact, but equally sadly, will soon likely evolve into full blown brain eaters.

These models are not “approximate”. They are not “good enough”. They don’t “work”. They are nothing more than flat out lies and nonsense.

The following answer is going to appear false in that it relates “colour” to “without colour”. How can an answer follow a series of demonstrations that don’t have “colour” in the classic “ZOMFG RED?!?!!” It will hopefully become clear that we can’t discuss “without colour” without, ironically, discussing “colour”. And in turn, we cannot discuss “lightness” or “darkness” without also simultaneously discussing “colour”. That is a long winded way of saying that we have indeed been discussing colour, even by way of demonstrations that appear to be “without colour”. More on that later, but for now, here’s an answer…

Answer #32: Changing the intensity of a single sample of RGB light will have echoing changes across the entire cognitive inferential field. As such, even a slight shift of a stimulus energy intensity will indeed change the inference of the colour qualia.

Phew. A helluva daunting point to get into the books, but one which will allow us to make far more intuitive steps further, as well as all sorts of conjectures.

  1. Ichinose T, Habib S. On and off signaling pathways in the retina and the visual system. Front Ophthalmol. 2022;2:989002. doi:10.3389/fopht.2022.989002 ↩︎
  2. Schiller PH. The ON and OFF channels of the visual system. Trends in Neurosciences. 1992;15(3):86-92. doi:10.1016/0166-2236(92)90017-3 ↩︎
  3. Barlow HB, Levick WR. Changes in the maintained discharge with adaptation level in the cat retina. The Journal of Physiology. 1969;202(3):699-718. doi:10.1113/jphysiol.1969.sp008836  ↩︎
  4. Shapiro AG, D’Antona AD, Charles JP, Belano LA, Smith JB, Shear-Heyman M. Induced contrast asynchronies. Journal of Vision. 2004;4(6):5-5. doi:10.1167/4.6.5 ↩︎
  5. I have been chastised by an unnamed computational neuroscience PhD for using the term “cognition”, who shall remain nameless to avoid any incrimination with the imbecile author. I have chosen this term to impress the incredible importance of treating visual perception as incredibly active, as opposed to the normative passive intake of stimulus. The critique stems from a redundancy among neuroscience types who would suggest that perception is entirely active, therefore using cognition is not normative. I hope any neuroscience PhDs who happen to stumble across this pile of nonsense written here forgive the author for abusing the term cognition in the name of smashing the casually normative notions around perception as a passive activity. ↩︎
  6. The astute reader would note that it is also a computation to determine the figure from the ground, and that indeed is an incredibly complex computational problem in and of itself! ↩︎
  7. Vladusich T. Simultaneous contrast and gamut relativity in achromatic color perception. Vision Research. 2012;69:49-63. doi:10.1016/j.visres.2012.07.022 ↩︎
  8. Heggelund P. Achromatic color vision-I: Perceptive variables of achromatic colors. Vision Research. 1974;14(11):1071-1079. doi:10.1016/0042-6989(74)90203-X ↩︎
  9. Heggelund P. Achromatic color vision-II: Measurement of simultaneous achromatic contrast within a bidimensional system. Vision Research. 1974;14(11):1081-1088. doi:10.1016/0042-6989(74)90204-1 ↩︎
  10. Heggelund P. A bidimensional theory of achromatic color vision. Vision Research. 1992;32(11):2107-2119. doi:10.1016/0042-6989(92)90072-Q ↩︎
  11. Horn BKP. Image Intensity Understanding. Published online August 1, 1975. Accessed July 29, 2023. https://dspace.mit.edu/handle/1721.1/6236 ↩︎
  12. Horn BKP. Understanding image intensities. Artificial Intelligence. 1977;8(2):201-231. doi:10.1016/0004-3702(77)90020-0 ↩︎
  13. Horn BKP. Determining lightness from an image. Computer Graphics and Image Processing. 1974;3(4):277-299. doi:10.1016/0146-664X(74)90022-7 ↩︎
  14. Grossberg S, Todorović D. Neural dynamics of 1-D and 2-D brightness perception: A unified model of classical and recent phenomena. Perception & Psychophysics. 1988;43(3):241-277. doi:10.3758/BF03207869 ↩︎
  15. Anderson BL, Winawer J. Layered image representations and the computation of surface lightness. Journal of Vision. 2008;8(7):18. doi:10.1167/8.7.18 ↩︎
  16. Wollschläger D, Anderson BL. The role of layered scene representations in color appearance. Curr Biol. 2009;19(5):430-435. doi:10.1016/j.cub.2009.01.053 ↩︎
  17. Faul F, Ekroll V. Transparent layer constancy. Journal of Vision. 2012;12(12):7-7. doi:10.1167/12.12.7 ↩︎
  18. Kozaki A, Noguchi K. The relationship between perceived surface-lightness and perceived illumination: A manifestation of perceptual scission. Psychol Res. 1976;39(1):1-16. doi:10.1007/BF00308942 ↩︎
  19. Anderson BL, Khang BG. The role of scission in the perception of color and opacity. Journal of Vision. 2010;10(5):26-26. doi:10.1167/10.5.26 ↩︎

2 replies on “Question #32: Does the F*cking “Colour” of the Light Change When We Increase or Decrease the Quantity of Light?”

Leave a comment