Thanks for writing about vision. I like to think of vision as a creative process. As you point out, we perceive a linear spectrum that ranges from red to violet. However, we imagine a circular spectrum that links violet to red. This does not exist in the physical world. Likewise, we imagine a visual field without holes. Yet, one exists where our optic nerve forms at our "blind spot". I have an image of a strawberry pie in which the strawberries are themselves gray (equal contributing values red, blue, green), yet the strawberries appear to be red because their "whitish" background is teal colored - it is an example of color constancy.
I recently wrote on our subconscious ability to track 1-5 objects. It is a level of visual processing that comes after contrast and edges are detected. Find it at https://tomrearick.substack.com/p/visual-indexing.
Really nice piece! Part of it reminds me of Dennett’s "windowless control room".
The most important part is, I think, the comparison to science. This is the relevant way in which "even science must be experienced in consciousness" as the idealists persevere about. Although I am hesitant to say that "I" am locked in my skull - I there's no line to draw between the instruments and "me". I am the instruments too. The inference begins already in the retina. As I see it.
I understand this interpretation process is not arbitrary due to evolutionary selection, but in what sense can you say it's not subjective? Some interpretations are in-built (like your example of the circles), but those influenced by personal experiences and thought constructs will change from person to person. Even if the differences are subtle.
I guess it depends on how hard a line you want to draw on what's subjective. There's certainly going to be differences in opinion on how we see some things (not sure if you remember the dress example from many years ago: https://en.wikipedia.org/wiki/The_dress)
But what we see isn't all just a guess informed by our experiences, there is real structure being picked up by our visual system. We might differ in situations dominated by ambiguity where our top-down assumptions play a large role (like the dress) but we're all going to agree on simpler less ambiguous visual scenes
The top down modulation of what we see can be hilarious. One time, I was working on a computing project with black and white calibration targets, and I walked next to a Target store, and the big red logo turned black and white and looked the same as my targets. This was a full blown hallucination, but a straightforward example of what the brain can do from the front that makes us see what we see.
I did see it! It's cool! But I saw someone (maybe Pete Mandik?) point out that we've already been able to do this with hyperbolic colors using much less sophisticated technology (staring at stuff) https://en.wikipedia.org/wiki/Impossible_color#Chimerical_colors
But even if it isn't totally new (and maybe there's some twist on it that makes it new other than the cool technology, I don't know, I haven't read the paper), it's a super cool effect
I always feel the use of words like construction, inference, and computation to confuse our understanding of what’s really going on.
The cortex seems to function by extracting and separating features into different specialized areas, not to put them together in an inference or construction. That’s really the mystery of how a distributed cortex functions to provide us He subjective illusion of a consistent stable world “out there“
Construct or infer always seems to me to imply a homonculus that’s putting together the visual scene. But if you look at what cortical neurons respond to they are specifc to edges, movement, maybe forms such as faces, but there’s no neuron infering that the person in front of you is grandma. You don’t infer an edge, you extract an edge. As I said, the remarkable thing is that subjectively there’s a scene in front of you and those can be sequentially focused on.
Visual illusions are informative in that you can’t chose to see the Müller-Lyer illusion as being two lines the same length, but where is that judgement made in the brain? The V1 is probably extracting the “wrong” length because of top down input based on context cues from the arrows. It doesn’t seem to me that V1 is infering. The idea of inference is a higher level, logical description of the whole thing.
But that’s me. I just find it confusion when someone says “the brain” has calculated or inferred. At least for me, thinking about how these metaphors don’t describe mechanism helps me think about what the brain is actually doing.
Thanks for writing about vision. I like to think of vision as a creative process. As you point out, we perceive a linear spectrum that ranges from red to violet. However, we imagine a circular spectrum that links violet to red. This does not exist in the physical world. Likewise, we imagine a visual field without holes. Yet, one exists where our optic nerve forms at our "blind spot". I have an image of a strawberry pie in which the strawberries are themselves gray (equal contributing values red, blue, green), yet the strawberries appear to be red because their "whitish" background is teal colored - it is an example of color constancy.
I recently wrote on our subconscious ability to track 1-5 objects. It is a level of visual processing that comes after contrast and edges are detected. Find it at https://tomrearick.substack.com/p/visual-indexing.
Really nice piece! Part of it reminds me of Dennett’s "windowless control room".
The most important part is, I think, the comparison to science. This is the relevant way in which "even science must be experienced in consciousness" as the idealists persevere about. Although I am hesitant to say that "I" am locked in my skull - I there's no line to draw between the instruments and "me". I am the instruments too. The inference begins already in the retina. As I see it.
Anyway, thanks!
I understand this interpretation process is not arbitrary due to evolutionary selection, but in what sense can you say it's not subjective? Some interpretations are in-built (like your example of the circles), but those influenced by personal experiences and thought constructs will change from person to person. Even if the differences are subtle.
I guess it depends on how hard a line you want to draw on what's subjective. There's certainly going to be differences in opinion on how we see some things (not sure if you remember the dress example from many years ago: https://en.wikipedia.org/wiki/The_dress)
But what we see isn't all just a guess informed by our experiences, there is real structure being picked up by our visual system. We might differ in situations dominated by ambiguity where our top-down assumptions play a large role (like the dress) but we're all going to agree on simpler less ambiguous visual scenes
The top down modulation of what we see can be hilarious. One time, I was working on a computing project with black and white calibration targets, and I walked next to a Target store, and the big red logo turned black and white and looked the same as my targets. This was a full blown hallucination, but a straightforward example of what the brain can do from the front that makes us see what we see.
Since you decided to write about vision, I thought I'd ask if saw this. It popped up in my feed yesterday: https://www.iflscience.com/a-new-color-scientists-claim-olo-is-like-nothing-youve-ever-seen-before-7887
I did see it! It's cool! But I saw someone (maybe Pete Mandik?) point out that we've already been able to do this with hyperbolic colors using much less sophisticated technology (staring at stuff) https://en.wikipedia.org/wiki/Impossible_color#Chimerical_colors
But even if it isn't totally new (and maybe there's some twist on it that makes it new other than the cool technology, I don't know, I haven't read the paper), it's a super cool effect
I had seen the criticisms of the claim as well. It was a cool effect, but I found the claims to be a bit overhyped as well. It just seemed timely.
I always feel the use of words like construction, inference, and computation to confuse our understanding of what’s really going on.
The cortex seems to function by extracting and separating features into different specialized areas, not to put them together in an inference or construction. That’s really the mystery of how a distributed cortex functions to provide us He subjective illusion of a consistent stable world “out there“
"Extract" versus "infer" or "construct" seem pretty similar to me in this context, why do you feel so strongly that "extract" is more accurate?
Construct or infer always seems to me to imply a homonculus that’s putting together the visual scene. But if you look at what cortical neurons respond to they are specifc to edges, movement, maybe forms such as faces, but there’s no neuron infering that the person in front of you is grandma. You don’t infer an edge, you extract an edge. As I said, the remarkable thing is that subjectively there’s a scene in front of you and those can be sequentially focused on.
Visual illusions are informative in that you can’t chose to see the Müller-Lyer illusion as being two lines the same length, but where is that judgement made in the brain? The V1 is probably extracting the “wrong” length because of top down input based on context cues from the arrows. It doesn’t seem to me that V1 is infering. The idea of inference is a higher level, logical description of the whole thing.
But that’s me. I just find it confusion when someone says “the brain” has calculated or inferred. At least for me, thinking about how these metaphors don’t describe mechanism helps me think about what the brain is actually doing.
To me "extract" has the same issue--it could be interpreted as implying a homunculus doing the extracting. All language is imperfect metaphor