What is a TALKING OCTOPUS?
The materials on this website reflect an attempt to clarify the concept of embodied, causally prior biological learning, i.e. “epistolution,” such that it can be studied in detail in laboratories, and also to imagine what some of the societal consequences will be when we decipher this unknown mechanism. The arguments put forward in my essays suggest that epistolution is the source not only of morphology and organismic behavior but of intelligence and morality. The “talking octopus” is what I imagine the first really impressive implementations of this principle to be; unlike todays algorithmic mimicry of intelligence, it will be the first creative, intelligent, moral machine on earth. It is a talking octopus because such machines will be embodied, with moving arms and legs, eyes, ears, etc. and will require training like children.
Can we refute the existing genes-first paradigm in biology? If it isn’t in a form that could be refuted, it isn’t science. Neo-Darwinism imagines that individuals are not real causal entities; they are only the visible effects of the causal action of genes. If you scratch for the fundamental Neo-Darwinist statement, it reads something like: 1. coding regions of the DNA specify the RNA configurations, 2. both coding and non-coding regions specify how many RNAs to make of each kind in all possible survivable conditions, and 3. control over the supply of particular RNAs is sufficient to regulate all the functions of life.
The first statement stands up; it is observable and has been supported by empirical evidence. But the second is an assumption to which I can imagine an alternative. It is also possible that another physical process exists in every living cell, and it is that algorithm that selects which RNAs to transcribe from the genome at which times. In this case, the third statement would also be falsified. This alternative would not replace evolution by natural selection, but would bolster it by showing that heritable changes are not blind but rather the result of cognitive intelligence.
I am not the only researcher to feel that current biology has already falsified this second Neo-Darwinist statement. There are nine or so types of result that have shown that the motivations of cells are not tied to their heritable material but rather use that material for cognitive ends. I suppose that if the theory is in a form that can be falsified, any behavior that shows individuals working on their own behalf beyond the narrow interest of their genes is such a refutation. This is a very tricky proof because of course any such examples must still be consistent with the existence of the genome or the lineage would go extinct. In other words, the very strongest refuting examples (like celibate monks) may leave no descendants, but that does not prove they didn’t exist.
1. During meiosis in sexual species there is a stage called crossing-over, when sister chromatids exchange segments. The division of these segments respects no boundaries between regions of the DNA, meaning that no particular gene can be a unit of Darwinian selection. Instead the fundamental unit of selection must (usually) be the individual organism.
2. Barbara McClintock, working in the 1950s, discovered that under stressful conditions, some organisms (her model was maize) can rearrange or duplicate their genomes to better cope with their environmental challenges. This became known as transposons or “jumping genes.”
3. Many prokaryotes exchange DNA promiscuously with unrelated lineages, a phenomenon called horizontal gene transfer. This has the result that, for example, it is estimated that nearly 8% of the human genome is originally of viral origin.
4. Mutation rates are not intrinsic to DNA, but the result of cellular processes that reduce errors in replication by three orders of magnitude. As a result, how many errors are allowed to persist where is a function of cell-level, not gene-level processes. This also means that mutation rates are nonrandom with respect to environmental conditions and (therefore presumably) nonrandom with respect to function.
5. Prokaryotes have various mechanisms for writing and editing DNA, as well as reading it. This has been called natural genetic engineering. Crisper CAS-9 was developed from one of these mechanisms.
6. Inheritance of acquired characters, long assumed impossible, has proved to be quite common. Many examples have now been found in all sorts of lineages. This has led to the resurgence of interest in epigenetics as a source of therapeutic innovation. Neo-Lamarckism is obviously not a replacement for Darwinism but a supplement to it. Darwin himself presupposed in the Origin of Species, as did all biologists in his day, that “use and disuse” caused heritable changes. He even proposed a mechanism for this influence which he called “gemmules.” In the present day, extracellular vesicles have been discovered that function just like Darwin’s gemmules, carrying information from the soma to the germline.
7. Targeted experiments have shown that regeneration and recovery can happen from injuries that cannot conceivably resemble those in an organism’s past lineage. In two good examples, Michael Levin’s lab at Tufts has melted the heads of planaria with barium, an injury from which the planaria recover and develop a resistance to barium (a substance that they cannot have encountered in nature.) They have also surgically rearranged the faces of tadpoles and shown that the faces reorder themselves, even overcorrecting and rebounding as the eyes and mouths reach the correct configuration.
8. Many multicellular organisms, including humans, have been shown to be intimately dependent on their associated microbiota for cognitive, immune, and metabolic function. The microbiome is primarily acquired from the environment, not inherited, and forms a unique community which is largely stable against perturbations. It is hard to argue that specially-formed commensal behavior in individuals has been reliably scripted by natural selection.
9. Organisms have not only evolved through independent competition but also through symbiotic fusions, called symbiogenesis. This theory was rediscovered by Lynn Margulis, who applied it to explain the origin of mitochondria and chloroplasts, and now has been widely accepted. Margulis pointed to some lineages that she claimed demonstrated fusions of up to seven different simple life forms into one.
I want to reiterate that we are not claiming to falsify natural selection, but simply explore the possibility that there is an additional algorithm in all life that produces cellular learning. This learning-first alternative reinvents the teleology of life; it makes life about finding explanatory knowledge, with survival and reproduction as a side effect. Discovering that algorithm is our empirical research program.
If we think of intelligence as an algorithm that exists in all cells, it narrows the range of possibilities considerably. It has to be something that is widespread in the universe, perhaps present in nonlife as well but specially modified in life. It seems to me and to my colleagues Richard Watson and Steven Lehar that a form of harmonic resonance is the answer. All matter resonates, often forming stable patterns of oscillations we call atoms, molecules, or objects. Resonance is the only other physical process that we know of that makes order out of physical disorder, and it is formalizable. There appears to be a special form of resonance in life that not only finds new oscillations but retains a trace of the previous, older patterns. Imagine a trumpet that retains an oscillating influence from previous tunes. This is memory. An organism, in this concept, would be a memorizing resonator, a configuration of matter that can accept new forms of order it finds through its senses without entirely losing the traces of the previous forms of order it has found.
This conjecture deserves to be tested in a rigorous way. At the moment, only a small handful of thinkers are even following the problem here. As I’m sure you have noticed, the public is currently on fire with the idea that intelligence is a Bayesian inductivist phenomenon due to the commercial success of LLMs like ChatGPT. This is probably bound to frustrate you. I hope that by testing ideas like harmonic resonance as a cognitive process we can raise awareness that the problem of biological intelligence is not already solved.