LEARNING TO CODE, WITH AN OCTOPUS AND TURING
Artist collective 0rphan Drift, founded by Maggie Roberts and Ranu Mukherjee, and technology agency Etic Lab are coming together for ISCRI: a unique research project that seeks to create an AI programmed by the alien mind that is an octopus .
ISCRI is an exploratory artistic, scientific and technological project that will create an AI programmed by an octopus. The AI will be generated from octopus responses to its environment; In an iterative process, the AI will mediate between the octopus and its environment, reflexively re-programming the video stream based on octopus responses.
My role so far has been to experiment with code that reproduces effects and patterns produced on the skin of octopus vulgaris, the common octopus. I have made versions in Processing, p5.js and C++ (in openFrameworks).
Learn more about the ISCRI project >
Why an octopus?
As philosopher Peter Godfrey-Smith explains in Other Minds [2016], the octopus is “the closest we will come to meeting an intelligent alien.” That’s because, our nearest common ancestor being a worm-like creature over 700 million years ago, cephalopod and human intelligence evolved in entirely different ways. Whereas we have a highly centralised nervous system, that of an octopus is distributed throughout its protean body. Their intelligence and other-worldliness has been a subject of human mythology for thousands of years, yet our understanding of them remains sparse.
Their formless, flowing shape, so differently structured to us, is at one with its watery substrate, sensing chemicals and currents, becoming its surroundings and dazzling with its mysterious displays. It’s the nature of one element of these displays that I’m addressing here.
HOW PATTERNS ARE FORMED ON AN OCTOPUS SKIN
Cephalopods have what marine biologist Roger Hanlon calls “magical skin”, using three types layers of cells to produce patterns:
Chromatophores: millions of pigment cells (producing long wavelengths, reds yellows and browns) - controlled by motor-neurons.
Iridophores: for shorter wavelengths and whites - iridescent, reflecting cells, proteins controlled by neuro-chemicals.
Leucophores: white cells, assemblages of spheres, light diffusers.
After analysing thousands of photographs of the patterns formed on an octopus’ skin, Hanlon and his team grouped them together into three principal categories, uniform, mottled and disruptive. Of these three, the second, mottled, seemed to closely resemble a Turing Pattern - as described in his 1952 paper “The Chemical Basis of Morphogenesis” - a foundation text of theoretical biology which describes how naturally occurring patterns in nature, such as camouflage, can arise “autonomously from a homogeneous, uniform state.”
More specifically, these mottled patterns resemble those of the Gray-Scott model, a (relatively) simple articulation of Turing’s idea.
So how do we explain the emergence of Turing patterns in Cephalopod skin with it’s dynamic, ever-changing nature and many colours?
This is a question that has preoccupied me for some time (it’s quite common to read that these patterns are encountered throughout nature, without asking why). I was first working on the assumption that the Gray-Scott equation merely coincidentally produced a pattern similar to certain formations in cuttlefish and octopuses. That’s because previous research has principally focused on the role of the motor-controlled Chromatophores in pattern formation, whereas as we’ve seen, there are three layers of cells (including the Iridophores and Leucophores).
However, in the new edition of Cephalopod Behaviour [2018], Hanlon et al’s definitive work in the field, they find that the Iridophores are not controlled by motor-neurons, but they are in fact platelets (of a yet unstudied protein) who’s form (and thus the wavelength of light reflected) is controlled by a neurotransmitter, Acetylcholine (ACh), the relative concentration of which is used to control the cell. So perhaps here is our answer to the question: “Why does an octopus skin sometimes display a Turing pattern?”. Because patterns produced with Iridophores are in fact a classic Turing pattern, as they’re governed by the relative concentrations/gradients of two chemicals - the neurotransmitter and the substrate plasma. It’s exactly the type of chemical diffusion reaction described in Turing’s “Chemical basis of Morphogenesis in Nature”.
With this in mind, I was more confident that manipulating visuals based on the Gray-Scott model would be a legitimate road to travel.
Making a translation device
In his Children of Ruin (2019), Adrian Tchaikovsky fictions a world where interspecies communication is possible, mediated by technology. Specifically in this case, handheld devices replicating the syntax of patterns/textures used by octopuses. I noticed that this was a speculative manifestation of areas that I was already working on, so have made a process diagram to imagine how this could in fact be expressed as a prototype.
