If Intelligence could be Artificially Made...

On expanding scientific explanation to include Platonic Form as a cause

Biological computer fusion by ChatGPT

If any innovations are to occur on the artificial intelligence / transhumanism front, they will be inspired by biologists who do not make the mistake of assuming that the mind and body work just like machines.

I teach a course called, “We Are Not Machines,” which, like my essays, explores all the ways in which artificial intelligence is unlike biological intelligence. My main point is that biological cells can use interpretation in their responses to signals—they have agency—while computer network nodes are passively changed by the information that flows through them.

Computer nodes have to be exposed to a lot of data (many examples of cats, or introductory SAT paragraphs, or responses to greetings) to be changed, or “weighted” accordingly. Neural networks are good at consensus, not creativity. Biological cells can adapt to new signals almost instantly, which opens the door for creativity. What cells are able to do on a primitive level is magnified by the brain and further expanded by the use of language.

All that said, however, I often note how living creatures can behave like machines; once they acquire habits, they might as well just be automatons. Mostly, Large Language Models (LLMs) mimic human behavior that is in that automaton mode.

Reductionists dominated biology in the 20th century. They assumed that creatures are really like machines and so they weren’t interested in biological creativity, assuming it didn’t exist. With the rise of emergence theory and complex systems science, that began to change. More and more biologists (Denis Noble is a great example) have produced very good work showing how organisms aren’t like machines and aren’t merely programmed by their genes.

As we learn more about biological creativity, could we invent computers that can also be creative?

In theory, processes that contribute to creativity and real intelligence might be induced artificially, given world enough and time. Maybe we will develop different kinds of computing systems that work more like living systems do. Instead of algorithms, computers could use chemical reactions. Instead of switches and wires, computer nodes could have membranes and receptors which could selectively emit ions that produce electro-chemical fields. Instead of being made of discrete circuits, computer networks could be suspended in a gel, through which the electro-chemical waves might propagate and the intersections of those waves might be harnessed as further signals, indicating the overall state of the network. Instead of trial and error training being used to strengthen the connections for the artificial cell nodes, there could be internal feedback that would allow novel responses to signals to be immediately self-sustaining. Instead of a network using “hallucinations” or random mutations to experiment with, a network could be semiosically constrained to only try out that which is likely to work. Instead of the structure of neural network connections itself acting as memory, computers could have an unalterable code sequestered somehow in the artificial cell, which its artificial organelles could access on demand and use as a tool for generating raw materials for further growth and repair.

If researchers were to succeed at creating such an artificial cell—all the parts of which would have to be self-replicating and self-assembling—then you might end up with an artificial life form that is capable of basic creative activities, like pursuing new goals (not just finding different means to the same end) and even behaving irrationally, such as slime mold can do.

But if you made such an artificial creature, you would lose all of the advantages that machines posses in terms of precision and speed. Biological creativity and machine precision are opposite kinds of information processing. It’s best, I think, to only loosely couple man and machine. Computers are wonderful tools that we need to be able to turn off and walk away from when we want to. Then we will have the best of what machines and life offer.

Throughout the rest of this essay, I will explore how understanding biology better could lead to breakthroughs in more efficient smart technologies.

Platonic In-Formation

Above I mention how electrochemical fields form in living tissue and how intersections of fields might be harnessed as further signals. This is a very important tool for creativity: the process creates reliable “ready-mades” that organisms can use in novel ways. New structures do not have to be built bit by bit through selection processes. Butterfly wing “eyespots” are formed by the intersection of electrochemical wave fronts. Also the placement of eyes during embryonic development is mapped out in this way. Genes don’t code for spatial or temporal patterns. Instead, the mathematical laws of spontaneous pattern formation provide the constraints that shape biological development and temporal rhythms. This could be considered a Platonic idea. It seems that Form is a causal force that guides natural processes.

Before Darwin and Mendel, such theories of biological form were common. They’ve been all but forgotten or dismissed in favor of the selectionist and connectionist theories of evolution and intelligence, which can be modeled by machines. (I say evolution by natural selection can be modeled, but not very well, in my opinion. It’s a bit ludicrous to assign a value to a trait in terms of its reproductive fitness when that trait is only one among countless traits, all interacting with dynamic environments: in other words, the confounding factors are infinite.) There are some structures in nature that exist and persist, not because they have been selected as reproductively fit. They may come to find a use in producing more offspring at some point. But that’s not why they are so prevalent.

Biology for the last century or so has been focused on genes and the molecular world—mostly because this focus has been fruitful. We have learned much about how biological signal pathways work and we have some understanding about why they don’t when a person becomes ill. But now we’re missing the forest for focus on the trees.

The field of medicine would benefit from zooming out from the molecular world to the wider world of bioelectric fields. Biologist Dr Michael Levin at the Alan Discovery Center at Tufts University is holding a symposium this fall, The Platonic Space, with researchers from many different fields arguing, in one way or another, that scientific explanation needs to be expanded to include Form as a cause.

Mostly these days, scientists are aware of the existence of forms that emerge in dynamical systems, but Levin says that to call these forms “emergent” might be misleading. The mathematical laws that define such forms are atemporal; they’re Platonic. The contention is that “mind-like” law, in addition to genes and environment, is a cause of biological form and behavior.

Some of Levin’s most promising work involves manipulating bioelectricity (instead of trying to rewrite genes) to stimulate regeneration. The hope is that it will be possible to get missing fingers or amputated limbs to grow back. If this is the future of medicine, it looks bright to me.

I discovered Levin because I do research in the evolutionary mechanisms of butterfly mimicry, and I came to understand how patterns in biology—body plans, skeletal structures, sleep-wake cycles, butterfly wing patterns—are determined, not by the genes per se: genes just make the ingredients. Spatial forms and regular behaviors are in part structured by mathematical laws underlying the processes that create patterns. This fact lends a direction to evolution that is not due to selection for reproductive fitness.

In short, form comes before function. Darwinists have is backwards.

Here is a short video of my work in this field.

About two years ago, Michael Levin and I teamed up to write science fiction together exploring questions about biological life/mind and artificial life/mind. Our first story, “Animism,”—about life in the future when all our appliances, gadgets, and tools talk to us—was just published in the fall issue of Interconnections: journal of posthumanism.

Although I didn’t have this in mind when I named my course, mentioned above, Levin argues that “Living things are not machines (also, they totally are).” I very much appreciate why a two-handed argument is necessary. On the one hand, life forms are unlike machines; on the other hand life forms can sometimes be manipulated almost as if they were machines. Living beings and machines have physical processes that can, in theory, be observed, and imitated. However, the complexity of biology is more than daunting, and even machines are so complicated now, with various levels of neural nets interacting with each other, that they might be on the verge of being considered “complex.” (Complex in science means capable of emergent behavior that the designers didn’t plan for.)

You might say that Levin’s glass is half full on these issues and mine is half empty, but we agree about the amount of fluid in the glass. He writes,

It’s very clear that our formal models, like Turing machine paradigms, are not sufficient to capture what is special about life.

But also, it turns out that there are no machines that aren’t, to some degree, also doing more than our simplistic stories of algorithms and materials lead us to expect.

Elsewhere, Levin also observes that both life and computing machines may be “in-formed by diverse patterns from the latent space” of Platonic forms.

Life already takes full advantage of the Forms. If machines were designed such that they did too, that could be a game changer. Such machines, I bet, would not need nuclear power plants or rivers of water.

The Future of Medicine

In the same essay mentioned above, Levin calls attention to the way our bodies can seem like machines when transhuman-like alterations are made to them, when for example, a broken bone is patched up with metal plates and screws. But then, he notes, only the body does know how to heal itself, close the incision.

Unfortunately, as some patients have discovered, the body can mount an immune response to even supposedly “inert” metals such as titanium, leading to fatigue, brain fog, headaches, and depression among other problems. If there are two kinds of metal in the body, a galvanic reaction can cause a host of other weird problems, including paralysis.

Medical interventions too often do not take into consideration the complexity of the body. There are, after all, sensitive ion-gated channels in each cell that modulate cell state, and if we mess with those, it could disrupt biological processes in significant ways. Fortunately for beings with bones that need mending or spines that need straightening, medical technology can adapt to new findings. Different implant materials can be substituted for metal, such as customizable 3-D printed ceramics or Ossio fiber that the body processes and replaces with actual bone.

But I do not see how brain implant technology is going to overcome this type of problem. It’s unavoidable that implants will disrupt bioelectric fields. There aren’t any new materials or alternative processes, so far as I know, that would make that technology safe.

Engineers need to learn more about biology if they want to try to imitate it or intervene in its processes. Machine metaphors have for too long imposed themselves on biology. It was the mechanists and the selectionists who banished Platonic Forms from biology. Mechanism and selectionism are only partial explanations.

For over a hundred years, we’ve been obsessed with molecules. We’ve also been obsessed with the idea that competition leads to innovation. Not everything that is good in nature exists because it beat out something else. I hope the 21st century is one of fields and Forms.