Substrate Independence
I explore below the reductionist stories of life and consciousness, the origin of living organisms, fitness and natural selection, substrate-independent phenomena, and what all this means for us as a species.
Contents:
II. Understanding Substrate-Independence
III. Building A General Framework
IV. Why Substrate Independence is a Big Deal For Us
I. The Story of Life
It is one thing to exist, and another thing to philosophise about it.
To start from the basics, living matter arose from non-living matter between 3.5 and 4.3 billion years ago, soon-after the formation of oceans on planet Earth. We have no consensus over how, but the strongest hypothesis suggests a gradual process involving (1) autocatalytic reactions, (2) molecular self-replication, (3) molecular self-assembly, and (4) the emergence of cell membranes.
Broadly speaking, life has two properties: ‘survival’ (metabolism and self-repair) and replication. All living organisms regenerate in accordance to their suitability to these parameters, thus possessing a ‘fitness function’ by nature, which was originally observed by Darwin as natural selection. To go full circle, this function is what NASA uses to define life itself: “a self-sustaining chemical system capable of Darwinian evolution”.
The power of this theory is that it encompasses so many other explanations. It states that for all species of organism, the reproductive process generates small variations in offspring, causing them to develop into ‘fitter’ species, more capable of surviving and reproducing in their environment. An organism’s fitness determines its ability to produce offspring, creating a feedback cycle; survival is granted only to those most fit.
In 3 billion years, life advanced incrementally from single-celled (bacteria) to multi-cellular (plants, fungi) to protostomes and echinoderms, then fish, amphibians, reptiles, birds and eventually mammals. This endless process of speciation is caused by a never-ended process of change on the Earth’s surface; be it tetonic, climate-driven, or social. Individuals are constantly exposed to different conditions that impact their chance of survival, hence the diversity in the lifecone we see today.
The other powerful result of this theory is its universality. Fitness scores are applicable to any genomic process, and as almost anything can be framed as an evolutionary problem, genetic algorithms can approximate anything from travelling salesmen to the solutions in your Gold Olympiad paper.
When running evolution on static problems like these, the majority of progress is made in the first few epochs — but in the wild, progress comes in bursts; freak mutations (like the first nervous system or cortical column) new vectors of growth create a series of cabrianesque explosions, consisting of predatory arms races, new environments to explore and adapt to, and an influx of diversity in the gene pool.
On a micro scale, the observation that inspired Darwin’s theory was the shape of finches’ beaks in the Galapagos. For a given beak size and shape, the original species of immigrant finch landing on the islands had only been able to exploit a share of a fixed food pool. Therefore, the fittest individuals would find creative sources of nutrients by probing into bark crevices, or by navigating the spines of a cactus. At a catatonic pace, the finches then tailored their dances, songs, and biology, to the diet that suited them best until their offspring were distinctly unique to eachother.
Figure 1. Darwin’s Finches: One ancestral species radiated into 14 branches.
It wasn’t until 100 years after the publishing of Darwin’s theory that the biological evidence was discovered by Watson, Crick and Franklin, in 1953. ‘DNA’ gives us the reductionist story of life itself. In Crick’s work, he states that “All approaches at a higher level are suspect until confirmed at the molecular level”.
But the plot thickens. As animals evolve, and their ability to generate and process information from their environments become more complex, a degree of ‘wakingness’ is produced. Some organisms become so complex that they not only see their environment, but feel it, hear it, smell it, and taste it. They may develop the ability to recognise complex patterns, make predictions, and plan many steps ahead. Given sufficient generations, some mammals learn to cooperate in large groups, build tools, and ‘solve’ their environment to increase their quality of life, by growing food on-command for more efficient survival.
Ultimately, hydrogen, in the right conditions, becomes sociable and self-aware.
Figure 2. Life from its origin to today.
II. Understanding Substrate-Independence
DNA gave us a lens to view all life through; the axiom of reproduction and metabolism. But it failed to square the circle for those who had been paying attention. If any stable, well optimised polymer could propagate biological life, then the real mystery becomes how consciousness emerges. The key to natural selection lies not solely in DNA, but rather in the behaviours of organisms, which are influenced far beyond their double helical structure.
This shift in understanding created the perfect segue for Francis Crick, who after his discoveries in molecular biology, switched to what was, in his words, the most vital study of all: consciousness.
Both evolution and consciousness are by-products of physical machinery that permit powerful emergent behaviours. The mind is a meshing of neuron frequencies, akin to the tones from individual instruments producing the rich and complex sounds of an orchastra. For each musician, the sounds of their instrument are distinct and analogue, but to those sitting beyond the conductor, a sympohony of transcendent emotion emerges.
This experience can be attributed to the idea of substrate independence, which applies not only to the brain, but to other naturally occuring phenomena. A conscious character in a future computer game will have no inherent way of knowing whether it is running on a desktop, tablet of phone, in exactly the same way we cannot detect our own synapses, and only know about their exist by conducting scientific experiments on our environment. [4]
What other substrate independence exists?
For one, waves are substrate-independent; their speed, length, frequency, intensity, and all behaviour related to their identity can be calculated without the knowledge of what matter the wave is passing through – by use of Schrodinger’s wave function.
The basic tenets of computation are also medium agnostic; the capacity to read and write to memory, perform conditional branching, basic arithmetic, and recursion, are all that is needed for a system to be Turing complete (Turing, 1950). This can be distilled into a minimal set of commands, used by the Brainfuck programming language (>, <, +, -, [, ], ., and ,). These operations allow the system to perform universal computation, meaning it can execute any computation given sufficient time and resources, regardless of the physical substrate. [2]
While these phenomena satisfy the basic concept of substrate independence, the matter from which they emerge still govern their nature. In other words, waves made in sand will never behave exactly like waves made in water or oil, despite being comparable along certain dimensions like frequency or intensity. [3]
III. Building A General Framework
With this in mind, we ought to make a framework for what defines valuable human-like sentience, because at some point we’ll be able to iterate over cognitive systems where ‘IQ’, ‘emotionality’, ‘open-mindedness’, or ‘curiosity’ are preserved, or emulated through other matter.
For humans, inputs are received through a set of sensors that detect certain wavelengths of light, vibrations in the air, odorants in the air, dissolved chemicals on the tongue, and mechanical deformation of the skin.
Cranial nerves then transmit these via electrical impulse to the brain, where a transformation process occurs and applies existing action potential to the inputs to interpret them. This transformation process produces characteristics (qualia) emerge like self-awareness, intentionality and agency, temporal continuity and emotionality. All qualia are produced by reasonably well-documented processes (sets of neurons firing in parallel), but every brain is pretty different, hence qualia are ultimately unique to us.
A larger and more complex mind would be able to transform far more computations per second across a larger subspace of meaning. We can imagine ourselves as a rung on a ladder of all possible minds; those below us have more constrained brains or cognitive systems, perhaps driven more by instinct and reaction, and less by learned experience (even chimps that know sign language don’t ask questions). Those above us can conceive of more complete and articulate thought experiments (and humorously observe that we don’t do things that they find trivial).
The purpose of building such a framework would be to meet the burden of proof when (1) growing and (2) transmitting our conscious experience into more capable mediums. By adding neurons and growing the mass of our virgin brains we can gain new superpowers, even today, cochlear implants and eye surgery enable new senses to form in the brain. But without the ability to transmit our minds, we will be forever fighting the onset of cell death and pathological disease in the mind — not to mention general vulnerability to the environment.
The key is transmitting our experience without sacrifing the ways in which we update. We currently use delta in long term action potentials to weaken and harden synaptic connections, and like DNA, this is an axiom of the mind. Without it, our minds would regenerate differently.
IV. Why Substrate-Independence is a Big Deal For Us
Natural selection is almost always self-induced; members of a species act on instincts that increase their odds of survival and replace those who can’t compete. But never has a species replaced other memebers of their species by building more intelligent machines, which is on the cards for humanity. Granting fire into non-human agents has the potential to lead to self-improvement at some far greater rate than humans are capable of adapting to, as all of our coping mechanisms are behavioural — we are yet to start work on updating hyperparameters at a rate beyond that of mutation.
With the ability to modify cognitive architectures and optimize thought processes, synthetic minds could evolve at a pace that dwarfs the rate of any biological evolution. An upshot of super advanced AGI would be the rapid progress on problems that have baffled humans for centuries. The downside is the instability and speed of regeneration; things that would put human civilisation at risk.
If we can nestle into a soft minima in intelligence, where the AI exhibits diminishing returns, and is intelligent but not profoundly dangerous, we can expect great accelerated returns in this space of transhumanism. Imagine minds free from the risk of degenerative diseases, minds that can be duplicated, backed up, or enhanced with updates that increase processing speed, akin to software patches that add bug fixes. This isn’t just about superseding our physical limitations —– it’s about crafting a new paradigm for what naturally occuring intelligence can achieve.
Figure 10. The ‘jump-off’ point for humanity.
The human brain receives 11 million input bits per second. 10 million of those are visual. What would it be like to have new types of inputs from our eyes? Implants could enable us to read x-rays, radio waves, and microwaves. Implants could enhance the resolution of our vision until we could see any object’s molecular arrangement or disturbances in gas particles. By understanding the parameters of our conscious experience, we can change them like dials in the settings menu, enhancing our reality.
The human brain is the most complex object in the universe. Finally, we are in a race to turn it back on itself before our inventions take the top spot.
References
[1] - 8 different types of eye? Let me check that out!
[2] - Alan Turing’s ground-breaking paper on computational logic is accessible here.
[3] - If a conscious human mind is merely a complex system of electrochemical reactions and a conscious algorithm is merely a complex system of electrical current, then a complex system of kinetically charged particles, such as a thundercloud, could also be conscious.
On a larger scale, the biosphere is a complex, regulating system of energy so complex that it is teeming with conscious agents. Who is to say that the biosphere itself is not conscious? Arguably, Earth has a metabolism, a biosphere that self-repairs, and an ability to yield human astronauts, Teslas, and city light. Those three components fit the definition of life we used so convincingly before. Lovelock’s Gaia hypothesis is still of contention 50 years later. [7]
Pushing scale to the limit, stars, solar systems, and galaxy clusters are all vastly more complex systems arbitrarily, and thus too could theoretically be waking. To prove this theory wrong, we will need to determine what it is exactly that guarantees consciousness and what prevents it.
[4] - Max Tegmark talks more about this in this article.
Imagine two programmers are jointly hunting a bug in their code, they are probably not conceptualising the computational architecture, but the software, variables, and the outcome of their button-inputs. The computations behind the code could be done with integrated circuits, vacuum tubes, relays, or electromechanical boards and the programmers wouldn’t even know without opening it up. [5]
[5] - For the journey to a bizarre, transcendent state of mind, start here: https://tim.blog/2022/04/13/donald-hoffman/ and continue here: https://www.youtube.com/watch?v=oYp5XuGYqqY.
[6] - ‘Suitcase word’ - coined by Marvin Minsky, one of the ‘Big 4’ Artificial Intelligence pioneers of the 60s.
[7] - Lovelock’s ‘Gaia Hypothesis’ is frequently cited as a prophetic view of the world and the world we find ourselves making. The new revision of the theory also focuses on humans developing hyperintelligence. It’s available here.