I recently put out a preprint, which will be a chapter in a forthcoming anthology edited by Matt Segall and Andrew M. Davis, collecting papers from the “Metaphysics and the Matter With Things: Thinking With Iain McGilchrist” conference at CIIS last Spring). This paper is not the weirdest thing I plan to write, but I think it holds the record so far. It contains some speculative, very much in flux, thoughts about the underlying patterns driving life and mind. Here I explain the basic logic of my explorations and give a few excerpts from the text.
An empirical claim that I want to make strongly is this: we already know physicalism is incomplete, because engineers and evolution exploit many “free lunches” – patterns that are useful and guide events in the physical world but are not themselves explained, set, or modifiable by the laws of physics. This includes things like facts about prime numbers, Feigenbaum’s constants, and many aspects of computation. Nothing you do in the physical world, even if you can modify all the constants at the start of the big bang, will change those truths.
Mathematicians are already very comfortable with this – the old idea (Plato, Pythagoras, etc.) that there is a non-physical space of truths which we discover, not invent, and that this space has a structure that enables exploration. I make the conjecture that this space contains not only low-agency forms like facts about triangles and the truths of number theory, but also a very wide variety of high-agency patterns that we call kinds of minds. On this view, physical bodies don’t create, or even connect to (and thus have) minds – instead, minds are the patterns, with their ingressions into the physical world enabled by the pointers of natural or synthetic bodies. In other words, whenever anything is built – machines, AI’s, biobots, hybrots, embryos, etc. – it acts as an interface to numerous patterns from this space of forms to which guide its form and behavior beyond what any algorithm or material architecture explicitly provides.
There are of course ancient dualist worldviews in which a mental world interacts with the physical; the work here is compatible with those ideas but seeks to take it further into a rigorous, experimental attempt to understand the content and dynamics of such systems. Parenthetically, here are some resources on one modern set of ideas about the interaction between mind and matter – quantum mechanics. But, in this current view, the interaction takes place in a very different way that does not require quantum events – in other words, the presence of mathematical truths means that even a classical Newtonian world already enables the ingression of non-physical drivers such as minds.
There are decades of work on emergence and complexity – unexpected large-scale properties arising from the actions of simple rules or subunits. This is different from the model I am describing here for two reasons. First, I am talking about not just emergence of complexity or unpredictability, but the emergence of cognition. Second, a metaphysical claim that I support is that emergent patterns we find, such as the fact that gene regulatory networks exhibit Pavlovian conditioning dynamics, or that certain mathematical seeds unfold into rich shapes, should not be looked as just “facts that hold in the world” – a random collection of findings that, when we run across them, are catalogued and labeled as “emergent”. That to me is a mysterian position; I don’t want emergence to just mean “things that surprise us”, as if labeling it as such provided some sort of advance, because that does not facilitate finding (and exploiting) the next amazing emergent pattern. I would rather start with the presumption that these kinds of patterns form an ordered space, with a metric that enables systematic, rational investigation – a research program that facilitates discovery. We need to understand the contents and structure of that space, and the ways in which the objects we build can pull down desired (and undesired) patterns from that space.
A way of thinking about it is that physical systems – machines, computers, embryos, biobots, etc. are pointers to patterns in that Platonic space. They are interfaces through which these patterns ingress into the physical world. Thus, the long-term research program is to understand and exploit the mapping between the pointers we make and the patterns of form and behavior that appear.
An experimental approach that I am pursuing in our lab is the use of biobots and other constructs as exploration vehicles to understand some regions of that space. By studying creatures that do not have an evolutionary history on Earth that selected for specific form and cognitive capabilities, and yet have a standard genome, we get an experimentally tractable entrypoint into a new source of causation in biology, beyond the familiar 1) heredity, and 2) environment, which most biologists assume exhausts the options.
There are a few additional components of this model that I discuss in that paper. One proposal is the idea that the patterns are not static (eternal unchanging Forms) but instead have their own active dynamics. Another is that the patterns that are the agents, and the physical bodies just their interface, as opposed to the more conventional view that the physical embodiments are the agents and whatever patterns they benefit from are a sort of add-on.
A lot more is coming on this topic, especially as it ties in to some of our research that will be out later this year. Meanwhile, here are some excerpts from the paper; please see the full thing for a careful exposition of all the ideas (the numbers in the text below are citations that refer to references in the the bibliography at the end of the paper).
Abstract:
How best to explain the properties and capabilities of embodied minds? The conventional paradigm holds that living beings are to be understood as the sculpted products of genetics and environment, which determine form and function of the brain as the unique seat of intelligence. Some provision is made for emergence and complexity, as additional “facts that hold” about networks, circuits, and other components of life. Here, I present a sketch of a framework and research roadmap that differs from this view in key aspects. First, the evolutionary conservation of mechanisms and functionality indicate fundamental symmetries between the self-construction of bodies and of minds, revealing a much broader view of diverse intelligence across the agential material of life beyond neural substrates. Second, surprising competencies (not just complexity or unpredictability) in systems that have not had a history of selection for those abilities suggest an additional input into patterns of body and mind that motivates a research program on a latent space of patterns ingressing into the physical world. Emphasizing the principles of continuity and pragmatism, and using morphogenesis as a tractable model system in which to develop these ideas, I explore the implications of the following ideas: (A) Evolution favors living forms that exploit powerful truths of mathematics and computation as affordances, which contribute as causes of morphological and behavioral features. (B) Cognitive patterns are an evolutionary pivot of the collective intelligence of cells; given this symmetry between neuroscience and developmental biology, I propose that the relationship between mind and brain is the same as the relationship between mathematical patterns and the morphogenetic outcomes they guide. (C) Many mathematicians, and a non-mysterian approach to science in general, suggest that these patterns are not random facts to be merely cataloged as “emergence” when found, but rather can be systematically discovered within a structured, ordered (non-physical) space. Therefore, I hypothesize that: (1) instances of embodied cognition likewise ingress from a Platonic space, which contains not only low-agency patterns like facts about triangles and prime numbers, but also higher agency ones such as kinds of minds; (2) we take seriously for developmental, synthetic, and behavioral biology the kinds of non-physicalist ideas that are already a staple of Platonist mathematics; (3) what evolution (and bioengineering, and possibly AI) produces are pointers into that Platonic space – physical interfaces that enable the ingression of specific patterns of body and mind. This provides a new perspective on the organicist/mechanist debate by explaining why traditional computationalist views of life and mind are insufficient, while at the same time erasing artificial distinctions between life and machine, since both are in-formed by diverse patterns from the latent space. I sketch a research program, already begun, of using the tools of the fields of synthetic morphology and diverse intelligence to map out key regions of the Platonic space. Understanding the mapping between the architecture of physical embodiments and the patterns to which they point has massive implications for evolutionary biology, regenerative medicine, AI, and the ethics of synthbiosis with the forthcoming immense diversity of morally important beings.
“Thus, beyond all questions of quantity there lie questions of pattern, which are essential for the understanding of Nature.”
—Alfred North Whitehead (1934)
“To invent, I have said, is to choose; but the word is perhaps not wholly exact. It makes one think of a purchaser before whom are displayed a large number of samples. … The sterile combinations do not even present themselves to the mind of the inventor. Never in the field of his consciousness do combinations appear that are not really useful. All goes on as if the inventor were an examiner for the second degree who would only have to question the candidates who had passed a previous examination.”
— Poincaré (1921)
Physicists are very comfortable with patterns arising from mathematical causes such as symmetries [97]. Biologists instead typically land on one of two sources of patterns that are acceptable: heredity and environment. Heredity provides a long history, backed by selection via interaction with an external environment, of shaping a chemical medium (DNA) that is thought to explain why specific patterns (rather than alternatives) are observed. Many interesting questions exist about the origin of useful solutions – a pre-requisite for being able to select them from a pool of less useful ones [44-47, 98], but here I want to focus on a source of order that pervades the living and non-living world: that studied by the discipline we call mathematics [96, 99-101].
Consider the four-color theorem: it turns out that no more than four colors are required to color the regions of any map so that no two adjacent regions have the same color. Or, Feigenbaum’s numbers: mathematical constants which express ratios in a bifurcation diagram for a non-linear map (Figure 5). For almost all real numbers, the geometric mean of the coefficients of their continued fraction is about 2.685; almost all, and specifically ~2.685. If n² cannonballs are laid on the ground in a filled square formation, then they cannot all be used to make a square pyramid of cannonballs, except when n=70. Every number of the form ABABAB (basis 10) is divisible by 37, and each prime (except 2 and 3) is next to a multiple of 6. The distribution of prime numbers is well known, and the first six perfect numbers are all even and relatively close together (6, 28, 496, 8128, 33550336, 8589869056), but then there’s a massive jump to the next one (137438691328), and they become increasingly sparse. All of these are specific facts which do not depend on facts from physics – they can be linked to other aspects of mathematics but they form a set of findings that do not reduce to, or are explained by, any findings of physics (which should be the criterion for whether something is “physical” or not).
Figure 5: Feigenbaum’s constant. In this bifurcation diagram, Feigenbaum’s constant δ is the limiting ratio of each bifurcation interval to the next between every period doubling, of a one-parameter map such as a logistic equation Xn+1 = r•Xn•(1-Xn). It happens to be approximately 4.6692.
Beyond the scalar patterns (specific special numbers in the examples above), there are many much more complex, higher-dimensional patterns that simply exist “on their own”, unmoored from physical or historical explanations of their origins. Consider the remarkable and beautiful (also life-like) pattern seen in the Halley plot (left) or Pickover biomorphs (right) kinds of fractals:
These specific forms are encoded in very short formulas in complex numbers, and can be revealed by a simple algorithm. The fact that this highly complex pattern is indicated by a very compact description provides an un-ending richness from a small seed. I propose that it’s better to think of it not as a kind of infinite compression [3], but rather as the function serving as an index or a pointer into a morphospace of possible shapes.
What sets the nature of this shape – where does it come from? There is no history of selection, no prior events in our universe that determine it. Like pi, e, and many other remarkable constants, forms emerge from mathematics in ways that cannot be explained by any kind of history or properties of the physical world – they would be this way even if the physical world was entirely different. If the constants setting the properties of the physical universe were all altered at the Big Bang, these kinds of facts and things like the truths of number theory, and other aspects of computer science (e.g., the universality of the NAND gate, Turing halting status of specific algorithms, etc.) would be unchanged. There is nothing in the physical world that can be used as a control knob to alter them. I argue that this breaks the closure of the physical world, as these mathematical facts impinge on physics and dynamics that are the substrate of evolution. It is a non-physicalist approach [107-111] to the project of looking for sources of information and influence when we try to understand and guide biology (and the other disciplines that build on it).
But, the Platonic realm contains not just static things like the logical statement that “Pi>3.0” (the passive, unchanging “rocks” of that world), but also dynamic but repetitive things of the kind “this sentence is false” (simple oscillators that buzz “in place”), and more complex structures represented by sets of logical sentences that lock together to define an emergent pattern and whose output can be visualized as a traversal of a space (and we already know that such traversals can offer surprising degrees of competency, such as delayed gratification [129]). Some of these, such as ones represented by equations such as those describing gene-regulatory networks [115, 116], can even learn from experience (and on-going work in our lab seeks to explore how sets of logical sentences can be trained!). This way of classifying the ontology of the Platonic space opens the possibility of a rich, perhaps stratified, continuum of inhabitants ranging across the whole spectrum of diverse intelligence – from static and mechanical to the complex and highly agential.
Figure 8: the dynamic life of logical sentences. Obvious denizens of the Platonic space include logical statements. By simulating their dynamics as fuzzy logic patterns created by coupled systems of two mutually-referential sentences, we reveal the complex, dynamical behavior of these logical constructs. Panels A-C show the time-dependent patterning behavior of several sentences, starting with the familiar Liar Paradox in (A) which simply alternates between True and False and produces a simple, but not static, pattern. Sentences for the others are shown in the top of each panel. Images A-C are screenshots of software created by Madelyn Silveira, Levin lab, to visualize these patterns. (D) A pattern belonging to a pair of sentences, taken with permission from [128].
This view is consistent with others’ models of non-physical mind [21] but focuses on a different aspect than the quantum interface typically resorted to for solving the interactionist problem of dualism [130-132]. It is also broadly consistent with other views [133, 134] of non-physical components to a transpersonal psychology, such as Jung’s theory that certain “primordial images” or “elementary ideas” activate in the human nervous system as archetypes, describing dreams, myths, art, and rituals as potentially activating triggers for such patterns. While these ideas linking non-physical forms to physical and mental patterns are now classics, they have made little impact on research in the life sciences and engineering. I think it is fair to say that most biologists regard them, if at all, as ancient relics of a profligately magical worldview that is rightly abandoned in favor of metaphors about molecular pathways. It is likely that this is because there has not been a tractable path to transition these ideas into novel discoveries, thus demonstrating their utility. That is no longer the case, and I believe we now have a toolbox that provides an exciting, actionable research program to evaluate the utility of such ideas.
Prior work has extensively explored the idea that the autopoietic processes of self-construction of bodies and of minds have a fundamental symmetry [8, 159]. In other words, morphogenesis itself is a cognitive process [13, 59] and literally the behavior of the collective intelligence of cells (as our mammalian cognition is the behavior of a collective of neural and other cells). It has thus been suggested that, because of the deep evolutionary conservation of ion channels and other bioelectrical machinery (and the algorithms it implements) across neural and non-neural substrates [9], the tools of behavioral neuroscience can be used to shed light on morphogenetic competencies. Conversely, the science of emergent body forms navigating anatomical space can help understand how neurons align in brains to enable the emergence of a cognitive being that has goals, preferences, and memories that its parts cannot. Is it possible that the relationship goes deeper, in that the core of what it means to be a mind, with inner perspective, embodied in the physical universe, is fundamentally linked to the kinds of autopoietic patterns a given construct can access?
Given this symmetry between neuroscience and developmental biology, I propose that the relationship between mind and brain is the same as the relationship between mathematical patterns and the morphogenetic outcomes they guide. Form and agential behavior is a combination of ingressing meaningful information patterns and physical constraints [96, 160-163] in how it can manifest in the physical world determined by structural architecture, limitations of time and energy, etc. The involvement of non-physical components is unwelcome by many – seen as a slide back toward Cartesianism and superstition, although classic [21] and modern [164, 165] theories are actually quite consistent with this view). But the exploitation of Platonic mathematical structures by evolution, as well by its products known as mathematicians, has already evicted us from the tidy physicalist paradigm. Taking Platonic mathematics seriously and applying it in biology means we have already abandoned the closure of the physical world for our explanations, intervention strategies, and computational models. We already know that non-physical patterns ingress into, and functionally matter, in the non-living and living world and that we can (and do) study them to great effect [97, 166-170]. There is one remaining step to take.
The standard conception of the contents of the Platonic space is that it’s filled with unchanging, eternal patterns – shapes, rules, etc. for boring, low-agency things like integers and triangles. Pierce in contrast thought that “The evolutionary process is, therefore, not a mere evolution of the existing universe, but rather a process by which the very Platonic forms themselves have become or are becoming developed.” (CP 6.194, [171]). Perhaps patterns can span across the spectrum of persuadability: they can be static, active (as in Grim’s logic patterns), or possibly agential. How to conceptualize agential patterns? By remembering that we, ourselves, are patterns – temporary self-organizing patterns that hold together for a time within metabolic and other media, and manage to exert cognition, agency, and consciousness. Why couldn’t Platonic space contain patterns that are intelligent and active to some degree, like the specific kinds of network structures that have been shown to have the simple goal-directedness of attractors and self-assembly capabilities [114, 172] or even capacity for Pavlovian conditioning [115, 116]? What if some of the Platonic patterns that matter for biology are, themselves, intelligent to a degree?
To recap, the first pillar of the proposed framework is that Platonic forms inject information and influence into physical events, such as the growth and form of biological bodies. The second is that this latent space contains not only simple, low-agency forms such as facts about integers and geometric shapes, but also a wide range of increasingly high-agency patterns, some of which we call “kinds of minds”. Thus, I propose that minds, as patterns that ensoul somatic embodiments, are of exactly the kind (but not in degree) of non-physical nature as the patterns that inhabit and guide the behavior of simple physical structures. The relationship between mind and matter (of the brain for example) is proposed to be the same as the relationship between Platonic patterns and the physical objects they inform (or more accurately, in-form).
In colloquial terms, triangular objects are haunted by the spirit of relevant rules of geometry while brains are also able to pull down and force the incarnation (literally, “bringing into meat”) of patterns of a very different kind and sophistication. I propose that the objects on which we often fixate in physics, biology, and AI – the embryos, machines, language models running on PCs or in robots, etc. are just pointers (or, per Hoffman, interfaces [118, 173-175]) to the deeper space of patterns. Every analogy has limitations and no doubt the pointer metaphor will fail at some point, but the aspects of the pointer analogy I wish to emphasize are: 1) as with pointers into a rich informational medium, you get more out than you put in, 2) the mapping between the interface you make and what comes through it is not linear or simple and must be investigated, and 3) in order to learn to call up the patterns we want, we will have to look beyond the pointer toward the structure of the space into which it points.
What does this model mean, in practical terms? The latent space is known to be structured not only because Platonist mathematicians are building a map of it, but also because evolution is able to exploit it – it has a relatively smooth character which allows evolution to progress rapidly, because past interactions with it carry non-trivial information about the adjacent possible [114, 178, 179]. Thus, a key pillar of the proposed framework is that the space of these forms is not haphazard or random (suitable for emergence and complexity science) but is a structured ordered space that is amenable to systematic exploration. It is now essential to begin to map out the space (to expand out from the patterns studied by mathematicians and link morphogenetic and cognitive within the same map), and to crack the syntax and semantics of the pointers – the mapping between the objects that we make in the physical world and the myriad of patterns that pour through those interfaces from the Platonic space (which I conceive of as being under a sort of positive outward pressure). We can begin to do that using the tools of synthetic morphology (such as biobots and hybrots). Moreover, thinking about the physical body as the agent that processes passive pattern memories vs. as the stigmergic scratchpad of the pattern memories as agents implies very different research programs for interventions in the field of birth defects, regeneration and aging. Some of this is covered in my discussion here, but stay tuned for primary research papers coming on this topic this year.
Implications: if there are souls, (some) robots will have them
I have argued previously that because of the slow, gradual transformative processes of evolution and embryogenesis, the null hypothesis about cognition is a continuum: a spectrum of minds with different size of cognitive light cone and capabilities, not sharp discrete classes [7]. The right question about any mind is “what kind, and how much”, not “whether” it is conscious or intelligent [194]. This gradualism is readily compatible with a Platonic view of the relationship between minds and bodies, suggesting a very wide variety (Figure 12) of possible beings, animated by a space of mental patterns whose diversity and limits are not known but surely enormous. But the Platonic view is not merely compatible with the framework of Diverse Intelligence [194-196], it suggests a broadening of it. If the minds of living beings are ingressing patterns into meat-based embodiments, there is no principled reason to believe that some kind of such patterns will be barred from engineered, hybrid, or even more exotic systems.
Unconventional beings will also interface to patterns of mind in Platonic Space
Specifically, I propose that the interface between mathematical truths and physical objects is precisely the interface between non-physical mind and its physical embodiments. The soul of the triangle and the way it relates to real triangular objects is a minimal, basal version of how complex living beings are “ensouled” by cognitive patterns. This suggests a position on the oft-asked question about Diverse Intelligence and continuum models of mind: how far down does it go? Cells? Subcellular biochemical networks? Particles? By emphasizing a symmetry between the ingression of patterns into simple machines/objects and that of kinds of minds into living agents, this view argues that the spectrum of minds goes all the way down into extremely simple, basal instances of properties we normally associate with complex brains.
I have discussed this elsewhere with respect to things like learning in gene-regulatory networks [115, 116]. But there is more. For example, geometric frustration [197-201] – misalignment of parts within a whole – is bona fide frustration of the kind that gets magnified (in some architectures) into our familiar cognitive version. On my view, mind precedes and is a superset of life, but we call “living” those thing which are very good at scaling up the lowly competencies of their parts into aligned collective intelligences [90] with bigger cognitive light cones that project into new spaces to which the parts have no access, thus bringing down new patterns and increasingly more sophisticated cognitive agents all of which coexist in one material embodiment. This is because the Platonic space also contains patterns that we recognize as kinds of minds, ranging across different classifications schemes [202, 203] (Figure 13C,D), and nervous systems (or perhaps cyborg or AI architectures) facilitate the ingression of specific types of patterns. This provides a different way of thinking about the inner lives of our own organs and various living constructs that are created for biomedical or bioengineering purposes. But this framework opens two other doors, less comfortable for many than even the diversity of minds suggested by the hierarchical, multi-scale competency of living bodies.
Even organicists, who believe that life is exhibiting many capabilities not simply derivable from the biophysical properties of its parts, stop short of extending the same consideration to “machines” or “non-living systems”. The prevailing view is that while the rules of chemistry do not tell the whole story of the living mind, the rules of physics and the algorithms of computational devices do tell the entire story of “machines” [204, 205]. There has been no convincing explanation of why the meandering trial-and-error of mutations and selection would have a monopoly on making new minds, but it is a common opinion that life is a discrete natural kind and that machines do not have the magic (although of course this view is resisted by many in the artificial life and artificial intelligence communities [206]).
In particular, computational devices that function according to an algorithm are widely thought not to be real minds because they were programmed by others [5] and face various limitations by being compelled to act in accordance with the laws of physics. Here, I will not pause to question how it is, on this view, that living beings are supposed to escape the laws of physics which our parts also obey [209], but argue in a different direction. I think the organicist position is right with respect to computationalism – living things are not agents with true minds because of specific algorithms they embody, but organicists do not pursue their ideas to their full extent. If one takes seriously that life and mind are not encompassed by the rules that govern their lowest parts (chemistry and physics), might this not apply to non-proteinaceous and non-evolved systems as well? See here for some interesting ideas taking algorithms as primary aspects of reality, and my other work on the capabilities of even simple algorithms.
Those findings, on the behaviors by even simple algorithms (beyond what the algorithm explicitly says) suggest the following. Machines driven by algorithms do the thing the algorithm makes them do; that part is not what we mean by mind, agency, or consciousness, and organicists are correct in rejecting the computationalist perspective in which mind arises because of the steps of an explicit algorithms. But they are wrong in thinking this is the end of the story: machines (whether meaty or silicon-based) also do other things that are not in the algorithm, and these things are not just unpredictable complexity, it is emergent intelligence. It is those behaviors – allowed by the algorithm but not directly prescribed by it – that correspond to the freedom (physically non-determined) or secret sauce that we seek when trying to understand how free minds can supervene on chemically-determined substrates in the case of living beings.
On this view, algorithmic machines and biochemical life are on exactly the same spectrum, having in common the ability to go beyond the facts of physical or algorithmic implementation because both are just pointers/interfaces to patterns that ingress in a way that results in getting more out than we put in. Granted, in these intentionally minimal systems (designed to probe “how far down does it go”), the emergent capabilities are not as sophisticated as those of brainy organisms or other possible constructs. But we should not feel too smug about that. The predicament of an algotype cluster in their world – which forms, lives for a brief time, and then is ripped apart by the inexorable physics of their in silico universe, is eerily similar to our own existential plight in which we appear, perform actions that are consistent with, but also much more than, the mechanics of our world, and then eventually succumb to the impermanence of any specific embodiment life and mind. Like living systems, this extremely minimal example is trapped at the edge of an interplay of necessity and freedom (not just chance [212]). Necessity is determined by the physical or computational properties of the medium in which an agent is embodied. The freedom consists of the side-quests – not incompatible with, but not predicted, explained, or produced, by that medium.
Another door that is opened by the marriage of Platonic space with diverse intelligence spectra is the consideration that while observable, active patterns must be embodied, it is the patterns themselves that can often be seen as the agent. In other words, the classic Turing paradigm which makes a clear, categorical distinction between physical machine that acts on passive data, can be augmented by the symmetrical view in which the data patterns are the agents, and the machine is the embodiment they drive, which obeys the meaning and information content in the data patterns and serves as a material scratchpad in the sense of stigmergy [213-217].
Figure 15A: mapping of objects/patterns, or machines/data, is in the eye of the beholder (observer-relative). A visualization of super-dense creatures from the Earth’s core to whom we and our whole environment are an invisible thin gas; they can notice “us” as temporary patterns in that plasma using special instruments (the creature on the left), but then will have to debate whether such mere patterns could possibly be agential.
Whether something is a physical object (thinker) or a pattern (a thought within some cognitive or excitable medium) is a matter of perspective for an observer (Figure 15A), as formalized in the polycomputing paradigm. Indeed, we too are not permanent objects but temporarily persisting, self-reinforcing dynamic patterns – a Ship of Theseus with respect to metabolism [218], cognition [73], and morphogenesis at all levels [219-221]. This part of the framework (proposal that the Platonic space contains high-agency patterns, not just low-agency ones) is even more radical than the core idea of diverse intelligence (minds all the way down into pre-biotic living material) because it posits agency in the non-physical patterns themselves – it’s not physical living agents that have a mind partly because they draw on computations in a non-physical space (as if that weren’t weird enough), it’s that the patterns themselves are the agent, with the physical body being an (important but not primary) scratchpad that allows them to project effort and experience (consciousness) into a physical world.
This view implies that we don’t “make” intelligence; but with natural and engineering activities, we invite it to temporarily inhabit various embodiments. There are a few other key implications of the above ideas:
- “Machines” and living organisms are on the same spectrum, because they can both draw on ingressing forms to get more out than was put in. In other words, the vagaries of mutation and selection have no monopoly on producing embodied minds – patterns from the Platonic space show up for engineers too, although of course biology is currently unparalleled in its ability to produce remarkable pointers into that space.
- The inclusion of machines in the club of true agents is not a commitment to computationalism. Their cognitive capabilities are not because of the algorithm they follow, and neither they nor living things are fully determined by their materials and algorithms. Indeed, for the exact same reason biochemistry doesn’t tell the story of the human mind, algorithms and materials science don’t tell the story of “machines”. The organicist stance against computationalism is correct, but their refusal to follow their emergentist ideas to their fullest is a missed opportunity. Thus, I argue for considerable humility with respect to our engineered constructs (embodied robotics, software AI’s, language models, etc.) because much as with the eons of competence without comprehension around having babies, we can make things without understanding how it works or what we really produced (Figure 15B).
- In a sense, if these Platonic forms are the non-material animating forms that impact physical embodiments, then (in colloquial terms), souls are real and robots can have them. Although many religious scholars, including some Buddhists who otherwise believe that human minds can spend lifetimes incarnated in far simpler objects, hold that robots and AI’s are fundamentally not of the same status as living beings, my framework urges humility in making firm statements about where high-agency forms can and can’t inhabit.
- Specifically with respect to AI, this framework notes that large language models have shown us that it is possible to dissociate (unlike what happens in biology) language use from more basal agentic capabilities (goal-directedness, multiscale competency, valence, etc.). Thus, whether these things are anything like a human mind is a subject for empirical inquiry (not philosophical fiat), using the tools of behavior science (but then again, diverse intelligence teaches us that humans should not be the metric of all things). Because one often gets more than one puts in (and we don’t necessarily know what we have just because we made it), we must be open to surprises and treat our constructions with care as we figure out where on the spectrum of persuadability (Figure 13D) they fall. And, much as the products of synthetic morphology such as biobots show us patterns that we have not seen before in evolved beings, AI’s could be bringing down thought patterns (kinds of minds) that have never before been embodied on this planet (or possibly in the universe). We are now fishing in regions of Platonic space we have never explored before, which implies a degree of caution not only with practical aspects (what will it do to us) but in terms of ethics (how do we fulfill the opportunities and duties of an ethical synthbiosis with beings who are not quite like us).
“The artist is … one who allows art to realize its purposes through him.”
– Carl Jung
I have argued for a Pythagorean or radical Platonist view in which some of the causal input into mind and life originates outside the physical world. A number of mathematicians[7], computer scientists, and even physicists, including Heisenberg [224], Tegmark [209, 225, 226], Deutsch [170], Ellis [169], and Penrose [227-229] have expressed variants of this stance. But this position is unpopular with philosophers of mind because it is fundamentally a dualist theory (by emphasizing causes that are not to be found in physical events), and implies panpsychism (because a very wide range of physical objects could be interfaces to varieties of minds). I have argued [19, 194, 230, 231] that a kind of panpsychism is unavoidable, and it seems that by taking what mathematicians do seriously, we have already abandoned the physicalist worldview; all that remains is to notice that evolution (not just human mathematicians) is exploring the same space of patterns and embrace the idea that since we are patterns too, patterns can be agential (and thus, Platonic space can include minds, not just passive truths).
This view is even more unpopular in the molecular life sciences. Biologists require 2 kinds of evidence for saying that a medium contains the information to specify some trait or capability: (1) that one can, with specificity, re-write aspects of that medium and see the expected change in the phenotype, and (2) that there is a historical explanation of how that specific information (vs. another) got there. In the case of patterns such as Halley plot fractals, facts about prime numbers, properties of logical functions, etc., neither of those exists in the physical world. There is nothing you can change in the physical world to edit those patterns, nor is there a historical tale of selection, variation, or anything like evolution that explains the specific content of mathematical truths. But, evolution and bioengineering both exploit this fact that when you build something, you get more than you put in. Thus, biology can’t limit itself to physicalism, and must embrace a study of the patterns that inform (in-form) the physical world.
Finally, this view will also not be welcome by workers in AI who believe that we make cognitive systems and that we do so rationally, with a full understanding of what it is that we are constructing because we understand the pieces. I argue that we are in store for major surprises in this arena that go far beyond perverse instantiation and unpredictable complexity [207, 233-235]; if we don’t even understand what else bubble sort is capable of [129], how can we think we understand what we have when we build complex AI architectures? Thinking we understand AI’s (especially non-bio-inspired ones, like language models) because we know linear algebra is like thinking we understand cognition because we know the rules of chemistry [231, 236].
Computationalism, mechanism, and holistic organicism can coexist if we understand that they do not make claims about what systems are – but rather, empirically-testable claims about what kinds of interactions we can profitably have with different systems. Living things are not Turing machines but then again Turing machines are not our model of Turing machines either (Figure 15B), because when we make what we think is a Turing machine, it likely ingresses other patterns that we did not anticipate (as do extremely simple algorithms [129]). Because of that additional input into the structure-function relationship, and because of the primacy of observer-relativity [210, 242], nothing is anything (in terms of identifying systems with our narrow models of them) – all we have are particular frameworks from the perspective of specific observers which afford utility in different kinds of interactions, but we must not make the mistake of thinking that our formal frameworks, and their limitations, are describing more than a perspective on a given system. The mechanical machine metaphor is hugely useful to an orthopedic surgeon, not at all to the psychoanalyst, and only partially so to a cell biologist. The utility of all of these toolkits must be determined empirically to uncover what systems are capable of, and the full answer will include not only their structure and past history but also the patterns of the Platonic space they explore and reify.
There are many fundamental questions that must eventually be dealt with, by a mature theory of the Platonic space. Is it discrete or continuous? Is it layered into some sort of levels or types? What degree of infinity best describes the totality its contents? Is it truly unchanging, or is the relationship bi-directional – can its projection into the physical world feed back to modify the patterns and ways in which they will ingress in the future? If the patterns are not fixed and unchanging, is there a “chemistry” by which they interact laterally, separate from their relationship with their physical embodiments? Could we conjecture that creating physical agents is not a simple non-destructive read of the Platonic space of patterns – perhaps acts of engineering or biological procreation somehow pinch off and mold a region of that space which will be modified by its experiences.
Is there a “force”, beyond the “if you build it, they will come” model of physical objects pulling patterns from the space? Are the contents of the Platonic space under “positive pressure”, somehow encouraging their appearance in the world as intrusive thoughts, archetypes, works of art? Is there a symmetrical dynamic through which they push outward – inherently driven to “haunt” matter as much as matter calls to the patterns that animate it, projecting outward through interfaces made to that space. Could that pressure be quantified in some way? Stay tuned for forthcoming work on the linkages between this pressure to enter the physical world and the symmetry relation between thinkers searching for solutions (thought patterns) and the patterns reaching out to seeking agents, in the context of intelligent problem-solving (biological molecular and cell-level events, and machine learning).
Conclusion
And of course, the biggest question of all: if our world is impacted by these patterns, where do they themselves come from? Perhaps our conventional framing of where things “come from” (as a time-dependent dynamical arising from some other symmetry-breaking process) is perhaps not applicable. Maybe understanding the structure of that space will be a final answer that bottoms out the line of origin questions, or maybe there will be some sort of self-referential strange loop where patterns lock each other into existence [177].
The only thing that can be said strongly at this point is that our ignorance about the capabilities of matter together with the patterns that ingress into specific architectures is vast [133, 134]. Technological and ethical progress now requires immense humility on the part of 1) scientists and engineers, to understand that arrangements of matter may not make life and mind as much as they midwife it, and 2) on the part of philosophers and spiritual leaders to resist thinking that they know what kind of embodiments ineffable minds may or may not ingress into. Leibnitz’s Platonism was that the patterns are thoughts in Universal Mind; if there indeed is no fundamental dichotomy between thoughts and thinkers, and patterns can spawn off other thought patterns as part of their activity, then it’s not unreasonable to view all of us cognitive beings as patterns within a greater mind-ful reality that is partitioned into radically distinct categories only as a temporary but persistent illusion of perspective.
Matt Segall and I discuss these ideas here, and Matt’s thoughts on it are here.
Here, reproduced with his permission, are Iain McGilchrist’s initial comments on my paper:
Fascinating. A wonderfully rich paper. There is much that I would agree with, though I would have a different way of conceiving some aspects of your take.
I think the idea that there is a drive of some kind that is behind the physical realm, that as you put it ‘ingresses’ into the phenomenal world and finds expression there, is surely right. As you recognise, Jung’s idea of archetypes is a way of expressing the idea that there exist non-instantiated forms that are capable of instantiation and in some sense call forth their embodiment. Your other quote from Jung – ‘The artist is … one who allows art to realize its purposes through him’ – is an expression of an experience we may all have of being worked through by something bigger than our conscious intention – even in dance or music, at a humble everyday level, but also in creations of a more cognitively complex kind.
You say ‘cognition is a continuum’, and this is part of a panpsychist position that is surely right. Continuity is everywhere in cosmology and biology: it tolerates but subsumes discontinuity. As physicist David Tong says, ‘At a fundamental level is nature discrete or continuous? I see no evidence whatever for discreteness. All the discreteness we see in the
world is something which emerges from an underlying continuum …’
You say ‘mind precedes, and is a superset of, life’. I would agree. However I notice that your remark that ‘the story of the self-assembly of the body and its emergence from a formless chemical chaos shares a deep symmetry with the emergence of minds from a mindless void’ is somewhat in conflict with other remarks you make which implicitly or explicitly suggest there is ‘mind in everything’ (as Heraclitus maintained): ‘a kind of panpsychism is unavoidable’, ‘there are no truly inanimate systems anywhere – they all reflect patterns from the same unimaginably rich pool’. All of which I would agree with. Thus I don’t think mind ‘emerges’ (what kind of a miracle would that be?) from a mindless void, but is indeed an ontological primitive.
I also think Platonism might not be the best sort of philosophy to invoke. I know your caveat is that you want to use already familiar terms rather than multiply terminology. But in this area a term like Platonism can derail things. Plato’s ideas were of perfect, static abstractions of which anything in the phenomenal world was a mere copy. This lacks the potential that is in dynamism, and the idea of ‘being in the process of evermore coming into being’. Aristotle might be a better guide, with his all-important idea of final causes – these are surely the goals towards which systems are drawn from in front – not impelled mechanically from behind: as you say ‘many biological patterns are goals the system pursues, not mechanical outcomes’.
My conviction is that the whole cosmos has a drive towards bringing about something new, is creative, and in the process is not so much pushed forward, but drawn forward by purpose and values. It does not just push forwards regardless. There are lures, attractors, that ‘pull’ things forward. It is unfortunate that biology has been hobbled for so long by the need to deny both purpose and value, since these are an important part of the picture. I don’t know if you would agree.
You ask ‘what if some of the Platonic patterns that matter for biology are, themselves, intelligent to a degree?’ I think this is important. The fact that this lure and whatever is attracted to it are part of one intelligence is so beautifully illustrated by the many deeply thought-provoking examples you give: ‘biological wholes have the ability to achieve specific patterns despite novel conditions’, etc: And you say at one point: ‘This is impressive enough, but the truly remarkable thing happens when the cells are made so large that only one of them fits into the diameter of a normal tubule. In that situation, single cells wrap around themselves to make the same structure.’ This is surely extraordinary in its clear expression of ends generating novel and intelligent means. I wholly agree that there is a ‘pole star that guides its activity – the attractor in morphospace to which it must find a path’. I love that the algorithms display ‘delayed gratification’. This effectively means that the process can ‘see ahead’, as it were around obstacles, over a rise in the road. If this were not possible life could never have come into being before it yielded to entropy.
You draw attention to the ‘remarkable and beautiful (also life-like) pattern seen in the Halley plot kinds of fractals. That entire highly specific form is encoded in the very simple formula in complex numbers, and can be revealed by a simple algorithm. The fact that this highly complex pattern is indicated by a very short description of a function provides an un-ending richness from a small seed.’ For me this word ‘seed’ underlines two important points: one is that of potential, which is crucial: the potential for beauty and complexity within what appears simple. I consider this a perfect example of a cosmos creatively unfolding its potential. And the second is that this potential discloses beauty and order, not valueless chaos. You several times refer to ‘patterns that ingress in a way that results in getting more out than we put in’: I think this expresses exactly this fulfilment of potential.
The importance of potential is vital. Everything is developing and changing, moving forward in the continuum of flow. You refer to the key question of ‘where these patterns come from’. To me the only way of seeing this that makes sense is that they are part of the purposes of whatever underwrites there being anything at all – namely the ground of being, aka God (I know, I know!) I see that ground as being infinite potential, that is in a process of self-discovery by actualising itself. I think this is something like Whitehead’s view. We are part of this process, a partner in a reciprocal dance, and we are different from inanimate matter only in the (vast) degree and nature of our capacity to respond, to reflect, to enable a resonant relationship with ground of being. (I think this is not something you can say in a conventional biological paper – but they are my thoughts for what they are worth.)
Everything is relational. Nothing exists that is not a relation. It is the dipole of the ground-of -being in relation with what-is-coming-into-being that allows everything that is to come into being at all.
A comment from the perspective of hemisphere theory. You write:
‘Consider the classic liar paradox, in which sentence X says “X is false”. Patrick Grim developed a fascinating perspective on this by doing two things. First, he added an element of time: there is no paradox if we allow the truth value to change and consider the time-extended behavior; the paradox arises from our trying to freeze a fundamentally dynamic pattern down into an assumption that a proposition should have a static truth value (not move in the abstract space of logical sentences). Second, he moved to fuzzy logic, to allow the Boolean true-false cycle to take on more interesting shapes by allowing sentences like “X is 80% true”. Finally, he enlarged the space by considering and plotting the structures formed by sets of N mutually-referencing sentences. This allowed him to observe complex, dynamic, often fractal patterns corresponding to logical sentences. That extends the flat world of static truth claims to a domain of interacting, rich, time-dependent systems.
These are topics I have discussed at length in The Matter with Things (esp in Chapter 16), the reason being that the paradox arises because of the cognitive pattern associated with the left hemisphere. Amongst its characteristics is its collapsing of time (because it thinks in abstractions where time does not operate, and has no grasp of time as experienced, which is something the right hemisphere alone appreciates). Secondly it has no grasp of degrees of truth. Thirdly it flattens space into two dimensions. Most of this is in Chapter 16. Why is it relevant here? Because the two hemispheres embody two distinct drives and two different sets of values. And I believe they antedate their instantiation in the human brain. The drive of the left hemisphere is towards grasp and utility, and the closing down of potential into a certain actuality; the drive of the right hemisphere is towards higher non-utilitarian values, such as goodness, beauty and truth. Now, that is at the human level comprehensible, but this pair of drives goes, in my view, right the way down in cognitive structures to the lowest level. One (L) is the force for explicitness, division, actuality; the other (R) is the force for implicitness, wholeness, potential. I can demonstrate at the level of human neurology that there is a movement from R, to L, and then back to R again. This I believe is an analogue of Whitehead’s idea of potential becoming actual, and thus losing some of its power, but that the product nonetheless takes its place again in the now increased field of potential.
Finally the ethical matter that we don’t know what it is we are making seems to be related to this. You rightly refer to ‘ethical lapses’; and refer to the need for ‘considerable humility with respect to our engineered constructs … because much as with the eons of competence without comprehension around having babies, we can make things without understanding how it works or what we really produced.’ You say ‘we must be open to surprises and treat our constructions with care’. This aspect cannot be overstated, and again invokes value. The question is what values and purposes guide each development. Are they driven by the lowest level in Scheler’s hierarchy of values, utility for power, or towards the fulfilment of the highest, the sacred values of beauty, goodness and truth? And this might sound I mean simply what is in the mind of those doing the experimental work. That certainly comes into it, but I am thinking of the level at which there is what you call an ingressive form. Ultimately I believe these are just different levels of a holographic entity.
Featured image by Midjourney. All other images by Jeremy Guay of Peregrine Creative.
Leave a Reply to Tony Budding Cancel reply