On March 24th 2015 Team Qualia Reverse Engineering (TQRE) went for a long walk within the Stanford campus and around Palo Alto. The purpose of this walk- the Pattern Walk -was to snap a picture of every interesting pattern (or texture) out there that got on our way. The following gallery contains 74 of these patterns. These display a wide range of texture properties: Natural/synthetic, regular/irregular, 2D/2.5D/3D, symmetric/asymmetric, structured/unstructured, etc.
Here are a few observations:
- Human languages do not have the necessary vocabulary (and conceptual primitives) to talk about visual textures adequately. When two images belong to the same category (say, “plants” vs. “rock tilings”), and have roughly similar first order statistics (mean, standard deviation, kurtosis, etc. of the RGB values) there is relatively little else to say about a texture in a way that a person would understand.
- Our visual system can recognize extraordinarily subtle properties that distinguish textures from one another. For instance, I bet you can recognize at an immediate experiential level the differences between picture 61 and 62. But can you verbalize such difference?
- Mathematics, and statistics in particular, may provide helpful semantic seeds for describing patterns. Indeed, having a basic handle on a few mathematical concepts can leverage one’s ability to talk about the differences between textures. For example, compare images 43 and 44. They are perceptually very different. But how long would it take you to convey the difference to a random person? If there was a person who could only hear you, how would you signal that you are not talking about 43 but 44? If both of you know of the concept of concavity you might only need a few words! Without it, you’d be fairly lost.
Fancifully, we may someday produce a good vocabulary that can effectively allow us to talk about visual textures without having to be currently sharing the same (similar) visual experience.
In practice, we already have some vocabulary that accomplishes this, but it is very obscure and sufficiently technical that its widespread adoption is unrealistic. In particular, I encourage anyone interested in the topic to read “A Parametric Texture Model Based on Joint Statistics of Complex Wavelet Coefficients” by Javier Portilla and Eero P. Simoncelli. They analyze (and synthesize) visual textures by computing a set of highly descriptive statistical properties characteristic of the pattern in question.
As we will see in future posts, their model can be used to point out perceptible statistical features that are perceived as regularities by the human visual system. It may not be sexy to say “Hey Ma’m I really dig the Cross-scale phase statistics of the pattern in your dress.” For now, that’s what we have.
If you want to help me figure out how psychedelics affect your visual experience:
Please browse through these images by clicking on the first one and exploring the slideshow. See which images you like, which produce “odd or interesting visual effects” and which “stand out” in however way you want to define that. Feel free to comment right below any of the images (there is a comment section beneath each image when you click through them as a slideshow) to point out the peculiarities that you notice.
Critically, also include your state of consciousness in the comment. If you took LSD (or any visually-affecting substance) two hours ago (or you are still high), it would be great if you could point that out. Please explain how you think that your visuals are affecting your experience of the various patterns. Everyone loves to talk about their LSD visuals. Now you can do it all you want! And your efforts may actually enable us to understand the way psychedelics affect the algorithms of human vision 🙂
The best case scenario:
You would make comments on these images while sober, and then add comments while high on a psychedelic (doesn’t have to by psychedelic – could be dissociative, though typing might be particularly hard in that condition). Point out the main differences between the textures as perceived on each of the states of consciousness you happen to be in. If you do decide to follow the above protocol, please provide information about the specific substance(s) you consumed and how long ago you did so.
That is, do this if you were planning on taking a hallucinogen to begin with. Independently of that, baseline data is still very valuable, so do add comments about these patterns even if you are sober and plan on staying sober 🙂
In the following post I will explain how this Pattern Walk, the statistical analysis of visual textures, psychophysics and psychedelics can ultimately fit into the larger project of reverse-engineering the computational properties of consciousness.