Like I've been saying all along, minds are built from the inside

Gary Lupyan and Andy Clark, Super-cooperators, Aeon. The lede: Clear and direct telepathic communication is unlikely to be developed. But brain-to-brain links still hold great promise

Definition: GOFT = Good Old Fashioned Telepathy.

The get off to a good start:

At the root of GOFT, however, is a problem. For it to work, our thoughts have to be aligned, to have a common format. Alice’s thoughts beamed into Bob’s brain need to be understandable to Bob. But would they be? To appreciate what real alignment actually entails, consider machine-to-machine communication that takes place when Bob sends an email to Alice. For this seemingly simple act to work, Bob and Alice’s computers have to encode letters in the same way (otherwise an ‘a’ typed by Bob would render as something different for Alice). The protocols used by Bob’s and Alice’s machines for transmitting the information (eg, SMTP, POP) also have to be matched. If that email has an attached photo, additional alignment must exist to ensure that the receiving machine can decode the image format (eg, JPG) used by the sender. It is these formats (known collectively as encodings and protocols) that allow machines to ‘understand’ one another. These formats are the products of deliberate engineering and they required universal buy-in. Just as postal systems around the world had to agree to honour each other’s stamps, companies and governments had to agree to use common encodings such as Unicode and protocols such as TCP/IP and SMTP.

But is there any reason to think that our thoughts are aligned in this way? At present, we have no reason to imagine that the neural activity constituting Bob’s thought – for example, I’m in the mood for some truffle risotto – would make any sense to anyone other than Bob (indeed, we are not even certain if Bob’s mental state could be interpreted by Bob himself in a year’s time). How then does Bob communicate his risotto desires to Alice? The obvious solution is to use a natural language like English. To be useful, these systems have to be learned. But, once learned, they allow us to use a common set of symbols (English words) to token particular thoughts in the minds of other English speakers.

It is tempting to assume that the reason why language works as well as it does is that our thoughts are already aligned and language is just a way of communicating them: our thoughts are ‘packaged’ into words and then ‘unpacked’ by a receiver. But this is an illusion. It is telling that even with natural language, conceptual alignment is hard work and drops off without actively using language.

Natural languages thus accomplish a version of what machine protocols and encodings do – they provide a common protocol that (to some extent) bridges the varied formats of our thoughts. Language on this view does not depend on prior conceptual alignment, it helps create it.

We can do this with language because we spend a great deal of time talking with one another and learning how to use language. We are continually negotiating the meanings of words. A bit later, they note:

Instead of viewing communication between people as a transfer of information, we can think about it as a series of actions we perform on one another (and often on ourselves) to bring about effects. The goal of language, thus understood, is not (or is not always) alignment of mental representations, but simply the informed coordination of action. On this picture, successful uses of language need not demand conceptual alignment. This view of language as a lever for coordination, a tool for practical action, can be found in research by Andy Clark (2006), Mark Dingmanse (2017), Christopher Gauker (2002) and Michael Reddy (1979).

That's what I've argued about music. I spelled this out in some detail in my paper, “Rhythm Changes” Notes on Some Genetic Elements in Musical Culture (2015).

They go on to speculate:

With this in mind, imagine now an alternative version of the sender-receiver setups used in Rao’s and Grau’s studies. Instead of instructing people to induce a particular mental state to communicate a predetermined meaning, there is simply a two-way brain-to-brain channel opened up between two or more individuals at a young age. The linked people then carry out various joint projects: they work on school assignments, move couches, fall in love. Might their brains learn to make use of the new channel to help them achieve their goals? This seems (to us, at least) to verge into more plausible territory. Something similar seems to occur when two people, or even a human and a pet, learn to pick up on body language as a clue to what the other person is thinking or intending to do. There, too, a different channel – in this case, vision – with a different target (small bodily motions) conveys an extra layer of useable information – and one not easily replicated by other means.

Setting aside the technical issues, could this work? Note that they stipulate that people be coupled together "at a young age" and that they "learn to make use of the new channel..." Learning is critical. Coordination would not be automatic through this new channel. It has to be learned, constructed.

My working paper, Direct Brain-to-Brain Thought Transfer A High Tech Fantasy that Won't Work (2020), deals with the same issues from a different perspective.

What’s it mean, minds are built from the inside?

I'm bumping this post from September 2014 to the top because it's my oldest post on this topic.

In my recent post arguing that “superintelligent” computers are somewhere between very unlikely to impossible, I asserted: “This hypothetical device has to acquire and construct its superknowledge ‘from the inside’ since no one is going to program it into superintelligence ...” Just what does that mean: from the inside?

The only case of an intelligent mind that we know of is the human mind, and the human mind is built from the “inside.” It isn’t programmed by external agents. To be sure, we sometime refer to people as being programmed to do this or that, and when we do so the implication is that the “programming” is somehow against the person’s best interests, that the behavior is in some way imposed on them.

And that, of course, is how computers are programmed. They are designed to be imposed upon by programmers. A programmer will survey the application domain, build a conceptual model of it, express that conceptual model in some design formalism, formulate computational processes in that formalism, and then produce code that implements those processes. To do this, of course, the programmer must also know something about how the computer works since it’s the computer’s operations that dictate the language in which the process design must be encoded.

To be a bit philosophical about this, the computer programmer has a “transcendental” relationship with the computer and the application domain. The programmer is outside and “above” both, surveying and commanding them from on high. All too frequently, this transcendence is flawed, the programmer’s knowledge of both domain and computer is faulty, and the resulting software is less than wonderful.

Things are a bit different with machine learning. Let us say that one uses a neural net to recognize speech sounds or recognize faces. The computer must be provided with a front end that transduces visual or sonic energy and presents the computer with some low-level representation of the sensory signal. The computer then undertakes a learning routine of some kind the result of which is a bunch of weightings on features in the net. Those weightings determine how the computer will classify inputs, whether mapping speech sounds to letters or faces to identifiers.

 

Now, it is possible to examine those feature weightings, but for the most part they will be opaque to human inspection. There won’t be any obvious relationship between those weightings and the inputs and outputs of the program. They aren’t meaningful to the “outside.” They make sense only from the “inside.” The programmer no longer has transcendental knowledge of the inner operations of the program that he or she built.

If we want a computer to hold vast intellectual resources at its command, it’s going to have to learn them, and learn them from the inside, just like we do. And we’re not going to know, in detail, how it does it, any more than we know, in detail, what goes on in one another’s minds.

How do we do it? It starts in utero. When neurons first differentiate they are, of course, living cells and further differentiation is determined in part by the neurons themselves. Each neuron “seeks” nutrients and generates outputs to that end. When we analyze neural activity we tend to treat it, and its activities, as components of a complicated circuit in service of the whole organism. But that’s not how neurons “see” the world. Each neuron is just trying to survive.

Think of ants in a colony or bees in a swarm. There may be some mysterious coherence to the whole, but that coherence is the result of each individual pursuing its own purposes, however limited those purposes may be. So it is with brains and neurons.

The nervous system develops in a highly constrained environment in utero, but it is still a living and active system. And the prenatal auditory system can hear and respond to sounds from the external world. When the infant is born its world changes dramatically. But the brain is sill learning and acting “from the inside.”

The structure of the brain is, of course, the result of millions of years of evolutionary history. The brain has been “designed” by evolution to operate in a certain world. It is not designed and built as a general purpose device, but yet becomes capable of many things, including designing and building general purpose computational devices.

But if we want those devices to be capable in an “intelligent” way we’re going to have to let them learn their way about in the world. We can design a machine to learn and provide it with an environment in which it can learn, an environment that most likely will entail interacting with us, but just what it will learn and how it will learn it, that’s taking place inside the machine outside of our purview. The details of that knowledge are not going to be transparent to external inspection.

We can imagine a machine that picks up a great deal of knowledge by reading books and articles. But that alone is not sufficient for deep knowledge of any domain. No human ever gained deep knowledge merely through reading. One must interact with the world through building things, talking with others, conducting experiments, and so forth. It may, in fact, have to be a highly capable robot, or at least have robotic appendages, so that it can move about in the world. I don’t see how our would-be intelligent computer can avoid doing this.

Just how much could a computer learn in this fashion? We don’t know. If, say, two different computers learned about more or less the same world in this fashion, would they be able to exchange knowledge simply by direct sharing of internal states? That’s a very interesting question, one for which I do not have an answer. I have some notes suggesting “why we'll never be able to build technology for Direct Brain-to-Brain Communication,” but that is a somewhat different situation since we didn’t design and construct our brains and they weren’t built for direct brain-to-brain communication. Perhaps things will go differently with computers.

By and large, we don’t know what future computing will bring. A computer with facilities roughly comparable to the computer in Star Trek’s Enterprise would be a marvelous thing to have. It wouldn’t be superintelligent, but its performance would, nonetheless, amaze us.

Consciousness, reorganization and polyviscosity, Part 4: Glia

Early on in my reading and studying neuroscience I read about the glia, brain cells between the neurons. Not much was known about them at the time and they seem not to have been much studied. With my recent interest in polyviscosity I decided to check up on the glia.

Things have changed. Quite a bit is now known about them. They seem to be crucial. One recent article:

Robertson JM. The Gliocentric Brain. Int J Mol Sci. 2018 Oct 5;19(10):3033. doi: 10.3390/ijms19103033. PMID: 30301132; PMCID: PMC6212929.

Abstract: The Neuron Doctrine, the cornerstone of research on normal and abnormal brain functions for over a century, has failed to discern the basis of complex cognitive functions. The location and mechanisms of memory storage and recall, consciousness, and learning, remain enigmatic. The purpose of this article is to critically review the Neuron Doctrine in light of empirical data over the past three decades. Similarly, the central role of the synapse and associated neural networks, as well as ancillary hypotheses, such as gamma synchrony and cortical minicolumns, are critically examined. It is concluded that each is fundamentally flawed and that, over the past three decades, the study of non-neuronal cells, particularly astrocytes, has shown that virtually all functions ascribed to neurons are largely the result of direct or indirect actions of glia continuously interacting with neurons and neural networks. Recognition of non-neural cells in higher brain functions is extremely important. The strict adherence of purely neurocentric ideas, deeply ingrained in the great majority of neuroscientists, remains a detriment to understanding normal and abnormal brain functions. By broadening brain information processing beyond neurons, progress in understanding higher level brain functions, as well as neurodegenerative and neurodevelopmental disorders, will progress beyond the impasse that has been evident for decades.

I take it, then, that the glia are central to consciousness, reorganization, and polyviscosity. And that’s only one article. 

It seems to me that one effect of the computational view of neural function has been implicitly to encourage treating networks of neurons as passive switching networks that just happen to be constituted of living cells. But the fact that cells are living has been treated as contingent and not essential to their switching functions. A great deal of neuroscience and cognitive reads like this and AI even more so. For that matter, that’s more or less how I thought about matters for years. That had begun to change by September of 2014 when I wrote a post, What’s it mean, minds are built from the inside? Here’s three paragraphs:

If we want a computer to hold vast intellectual resources at its command, it’s going to have to learn them, and learn them from the inside, just like we do. And we’re not going to know, in detail, how it does it, any more than we know, in detail, what goes on in one another’s minds.

How do we do it? It starts in utero. When neurons first differentiate they are, of course, living cells and further differentiation is determined in part by the neurons themselves. Each neuron “seeks” nutrients and generates outputs to that end. When we analyze neural activity we tend to treat it, and its activities, as components of a complicated circuit in service of the whole organism. But that’s not how neurons “see” the world. Each neuron is just trying to survive.

Think of ants in a colony or bees in a swarm. There may be some mysterious coherence to the whole, but that coherence is the result of each individual pursuing its own purposes, however limited those purposes may be. So it is with brains and neurons.

So, I’ve been moving away from the passive-switching-network view for a while.

But it took that 1988 paper by Fodor and Pylyshyn (pp. 34-45):

Classical theories are able to accommodate these sorts of considerations because they assume architectures in which there is a functional distinction between memory and program. In a system such as a Turing machine, where the length of the tape is not fixed in advance, changes in the amount of available memory can be affected without changing the computational structure of the machine; viz by making more tape available. By contrast, in a finite state automaton or a Connectionist machine, adding to the memory (e.g. by adding units to a network) alters the connectivity relations among nodes and thus does affect the machine’s computational structure. Connectionist cognitive architectures cannot, by their very nature, support an expandable memory, so they cannot support productive cognitive capacities. The long and short is that if productivity arguments are sound, then they show that the architecture of the mind can’t be Connectionist. Connectionists have, by and large, acknowledged this; so they are forced to reject productivity arguments.

Jerry A. Fodor; Zenon W. Pylyshyn (1988). Connectionism and cognitive architecture: A critical analysis. Cognition, 28(1-2), 0–71. doi:10.1016/0010-0277(88)90031-5.

THAT focused my attention on the problem of memory as a physical process. And that, in turn, led me back to Walter Freeman, which I discuss in this post, Physical constraints on computing, process and memory, Part 1 [LeCun]. And that in turn let me to my thoughts about polyviscosity – though I’d initially used the term “hyperviscosity,” but have abandoned it because it is already in use.

So, memory presents a physical problem (Fodor and Pylyshyn). That problem thus requires a physical solution: polyviscosity. And polyviscosity, it would seem, requires living tissue. Perhaps we’ll figure out how to implement it in inanimate materials, but at the moment living tissue is what we’ve got. And glial cells are central to the mechanisms of polyviscosity.

I’m tempted to say something like – and here I’m rambling again – that the glia implement consciousness. And the function of consciousness is to ‘operate’ the neuromolecular mechanisms of reorganization. And if that seems a bit circular, well, that’s the best I can do at the moment. The important point is that consciousness, reorganization, polyviscosity, and the glia and involved in the same phenomena.

More later.

* * * * *

Earlier posts in this series:

• Consciousness, reorganization and polyviscosity, Part 1: The link to Powers
• Consciousness, reorganization and polyviscosity, Part 2: ‘Fluidity’ & its requirements
• Consciousness, reorganization and polyviscosity, Part 3: Substrate independence, NOT

Another ramble: AI alchemy & the future, RationalityLand, life only, terminology, blues, NCIS...

I haven’t done one of this in a while. I do them when I’ve got a number of things jammed up in my mind and have trouble deciding what to do next. So I write a post in which I talk briefly about each of them, letting them rub up against one another.

Al alchemy and the future

This comes from a recent post, Superhuman AI, a 21st century Philosopher's Stone? The idea is that AGI and superintelligence play the same role in some people’s imagination that the Philosopher’s Stone played in an older imagination. I’m heading toward a 3QD post that might be titled, “Alan Turing and the Philosopher’s Stone.”

This involves the role of the imagination in thinking about the future: What of science fiction, scenario planning, and prediction? How do we direct our activities in exploring new (intellectual and imaginative) territory? Why did I turn to New York 2140 get a sense of our possible real future? And why do I think AGI is an imaginative fancy, like, well, the Philosopher’s Stone?

This leads to thoughts about...

RationalityLand and epistemic theater

By which I mean folks who tend to hang out a blogs such as LessWrong, Overcoming Bias, and Astral Codex Ten and are interested in things that include effective altruism, AGI, and AI alignment. These people see themselves as being deeply committed to rationality in all things and sometimes will refer to “the rationality community.”

I’m preparing posts in which I look at two posts from RationalityLand. One of them is from 2016 by Holden Karnofsky of Open Philanthropy: Some Background on Our Views Regarding Advanced Artificial Intelligence. He explains why he thinks AAI is likely within this century. Some of what he says strikes me as being what I’m calling epistemic theater, assertions couched in quasi-technical language that are not supportable by the underlying situation.

My other post is about a very long post by Scott Alexander at Astral Codex Ten: Biological Anchors: A Trick That Might Or Might Not Work. He’s taking a look at recent discussions about a long and complex research paper out of Open Philanthropy that attempts to predict the emergence of advanced AI. He’s very ambivalent about it. Does he fear that such research is, in effect, epistemic theatre?

Why fully human intelligence is impossible for a machine

At the end of his Chinese Room argument Searle suggests, but does not argue, that biology has something to do with it. Only living systems are capable of intentionality, which is required for meaning. How might one construct an argument on that point? I have an idea or two, but only that.

I’ve got part of such an argument in the idea that minds are built from the inside. I need to couple that with the observation that the distinction between hardware and software, which is central to digital computing, doesn’t hold for brains. I should probably toss in energy and thermodynamics as well. The human brain uses much less energy that digital computers, moreover, it is responsible to obtaining its own energy. If its operations require more energy than its actions allow it to obtain from the world, then it cannot survive. That brings up the issue of complexity that Hays and I explored some time ago.

Terminology

We need some terminology. What do we call what artificial minds do? Can we cover it with think, calculate, and compute, or do we need another term? For that matter, do we want to call them minds? Or do we just call them AIs? – “artificial intelligence” is too long to rattle off all the time.

I like the term “mentat” for AIs, but it’s already in use in Frank Herbert’s Dune universe. Still... If we adopt it, what is it that mentats do if not think?

Tell me about the blues

I’m planning a series of relatively short music post about the blues. Each post will comment on one, two, or maybe three, YouTube videos of blues performances. When I’ve written a half dozen or so posts I’ll collect them into a article for 3QD.

NCIS

Maybe I’ll do a post about how family is treated in NCIS. I’ve got notes on the topic. In what way is the NCIS team a family? How does that differ from real actual, you know, family?

Working papers

I’ve got a number of working papers to do. There’s another Seinfeld one in which I collect my most recent Seinfeld posts, the ones in which I analyze specific bits. I need to collect my Jaws posts, including the 3QD article, into a working paper. And I need to write a primer on attractor-nets.

What’s the opposite of substrate independence?

I suppose we could call it “substrate dependence” or “substrate linkage,” but this really isn’t about the term, it’s about the substance. As I recently noted, the idea of substrate independence is often invoked to indicate that it should be possible to construct a (proper) mind in silicon, though we’ve not yet figured out how to do it. I have my doubts, but who knows?

Substrate independence presupposes a fully explicit computational procedure, one whose structure is fully accessible to an external observer, one that can be constructed from the outside. The human mind, as I’ve argued, is constructed from the inside. I note as well that while digital computers recognize a distinction between addresses and content (data stored at an address), there’s no reason to think that such a distinction exists in natural nervous systems. Nor is there a distinction between memory units and processing units – something von Neumann recognized in Computers and the Brain. All neurons seem to be both processors and memory. And each neuron is a living agent.

What if all these things – built from the inside, no distinction between address and content, no distinction between memory and processing, living components – are necessary for the construction of a mind? Could they be realized in silicon, or some other inert substrate? That’s not at all obvious.

Minds are built from the inside [evolution, development]

Thinking about this post from last year in conjunction with my short note about Substrate Independence. What's the connection? Systems that are substrate independent are not built from the inside. Their structure is observable externally and are constructed externally. It's this externally observable structure that can be transferred from one substrate to another. 

More later.

* * * * *

What’s it mean, minds are built from the inside?

As far as I can tell, I’ve been arguing that minds and brains are built from the inside since (at least) January 2002, when I first imagined a thought experiment about coupling two brains together through a huge number of point-to-point connections. I put an argument (about inside) online in 2014 and elaborated on it in my current post at 3 Quarks Daily about the impossibility of direct point-to-point brain-to-brain communication. I’ve decided that the argument is important enough that I’ve excerpted it from the larger article and have placed it below.

The argument I make about brains is, I believe, true for all evolutionary and/or developmental systems, whether biological or cultural. They start at some point in time, maintain a boundary between themselves and their surrounding, and develop from the inside. They are self-organizing.

Note that I first developed the diagrams for a presentation I made before the Linguistics Association of Canada and the United States and were intended to convey an idea I learned from David Hays, who, in turn, learned it from Sydney Lamb: that the meaning of an idea in a network is a function of its location in the entire network. As such, it is difficult to observe such meanings from outside the system.

Minds are built from the inside

Let us start with this simple diagram:

It makes a very simple point: that the central nervous system (CNS), with the brain as the largest component, functions in two worlds. There is the external world: the physical world, the world of plants, animals, and other people. And there is the internal milieu: the body’s interior (in which the brain itself is situated). The brain senses the external world through the vision, hearing, smell, taste, touch, and a other senses; it directs action in the world by control of the skeletal muscles. Similarly, it senses the internal milieu through the bloodstream and acts on it through the endocrine system – there is more to it than that, but we don’t need it all; it is the principle that I’m interested in.

This is basically the same diagram, but with just a bit more detail in the central box, the CNS.

Now we have distinct regions for receiving input from the external world (A), from the internal milieu (B), for acting on the internal milieu (C) and for acting in the external world (D). Finally, there is a central area containing neurons connected to neurons in the other areas. While no real nervous system is that simple, they are all elaborations of that basic organization.

Closed organization

And that organization is CLOSED. What do I mean by that?

The brain isn’t something that is assembled from a bunch of parts scattered throughout a bunch of bins from which they are fetched by some Transcendental Maker. The process is quite different from what I did years ago when I assembled my stereo amplifier from a kit. When I did that I laid all the part out and assembled the basic sub-circuits. I then connected those together on the chassis and, when it was all connected, plugged it in and turned in on, a magic moment when all the dead elements suddenly came to life.

No, the brain develops through an organic process that starts with a single fertilized egg which then begins dividing and differentiating. At some point about three weeks into the process specifically neural cells appear in the embryo and then differentiate into the brain and the peripheral nervous system.

The brain is far from fully developed at birth, but its operating environment changes drastically at that time. Until that point its external world had been the womb. After birth it is exposed to the larger external world. In humans the brain continues developing into the early twenties, at which time the sutures in the skull finally close completely.

At every point in the process the cells are living cells. The neurons are receiving inputs and generating inputs. They are, in effect, becoming used to one another, “learning” one another’s ways. They are mutually adjusted.

THAT’s what I mean when I say that the system is closed. There is no external meddling going on. How could there be?

External meddling

External meddling, that is what happens when two brains are hooked together by a high tech coupling linking tens or even hundreds of millions of neurons across two brains. Both brains are now subject to considerable external meddling. Each brain is receiving inputs from a source it has no experience with, and generating outputs to that source as well. As I have indicated before, it has no way of even recognizing the presence of all this foreign input as foreign. It is just there. It is noise, electrochemical energy with no traceable linkage to the organism itself.

I suppose one could imagine that in time, weeks, months, years, the two brains would somehow sort things out. Maybe. But that’s very different from the instantaneous perception and recognition that Koch is talking about. But maybe not. Maybe the initial shock of all that noise is too much to overcome. We simply don’t know.

No two brains are alike

Bu..bu..but aren’t the two brains alike?

Only approximately, only approximately.

It is easy to identify gross body parts for one individual with the same parts for another. Here’s my left leg, there’s your left leg, here’s my left thumb, there’s your left thumb. But you can’t do that with hairs on the head. For one thing, people don’t even have the exact same number of hairs on their heads.

It is the same with neurons. It is not as though we could link neuron #7,983,004,512 from one brain to neuron #7,983,004,512 in the other brain, and so on for tens or even hundreds of millions of neurons. We have no way of making such identifications between neurons in different brains. Brains are sufficiently different from one another that it is difficult to identify corresponding areas with a high level of precision. Gross correspondence at the scale of centimeters or millimeters is all we can do. That’s not very high precision.

What we have at best, then, is some miraculous technology linking millions or even billions of neurons in one brain with of millions or billions of neurons in another brain in some quasi-ordered pattern based on approximate brain geometry. This technology allows the two brains to send signals to one another, signals which neither brain is prepared to deal with and which therefore interrupt the normal processes of both brains. I do not see how anything resembling coherent thought can emerge from the resulting electrochemical chaos. Even the best of magicians is incapable of pulling a rabbit out of a poorly constructed hat.

A quick note on the ‘neural code’ [AI meets neuroscience]

In The End of Science John Horgan has suggested that something called the ‘neural code’ as the Holy Grail of neuroscience, indeed, perhaps of all of science (from the preface to the 2015 edition): “The neural code is arguably the most important problem in science—and the hardest.” But it is so far proving elusive. Here are some passages from an article he posted in 2016:

Koch doubts, however, that the neural code “will be anything as simple and as universal as the genetic code.” Neural codes seem to vary in different species, he notes, and even in different sensory modes within the same species. “The code for hearing is not the same as that for smelling,” he explains, ”in part because the phonemes that make up words change within a tiny fraction of a second, while smells wax and wane much more slowly.”

“There may be no universal principle” governing neural-information processing, Koch says, “above and beyond the insight that brains are amazingly adaptive and can extract every bit of information possible, inventing new codes as necessary.” So little is known about how the brain processes information that “it’s difficult to rule out any coding scheme at this time.”

A bit later:

British neurobiologist Steven Rose suspects that the brain processes information at scales both above and below the level of individual neurons and synapses, via genetic, hormonal, and other processes. He therefore challenges a key assumption of Singularitarians, that spikes represent the sum total of the brain's computational output. The brain’s information-processing power may be many orders of magnitude greater than action potentials alone suggest.

Moreover, decoding neural signals from individual brains will always be extraordinarily difficult, Rose argues, because each individual’s brain is unique and ever-changing. To dramatize this point, Rose poses a thought experiment involving a “cerebroscope,” which can record everything that happens in a brain, at micro and macro levels, in real time.

Let's say the cerebroscope records all of Rose's neural activity as he watches a red bus coming down a street. Could the cerebroscope reconstruct what Rose is feeling? No, because his neural response to even that simple stimulus grows out of his brain's entire previous history, including a childhood incident when a bus almost ran him over.

To interpret the neural activity corresponding to any moment, Rose elaborates, scientists would need “access to my entire neural and hormonal life history” as well as to all his corresponding experiences.

That resonates with remarks by the late Walter Freeman that Horgan mentioned in an earlier piece on the neural code:

Then there is the chaotic code championed by Walter J. Freeman of the University of California at Berkeley. For decades, he has contended that far too much emphasis has been placed on individual neurons and action potentials, for reasons that are less empirical than pedagogical. The action potential “organizes data, it is easy to teach, and the data are so compelling in terms of the immediacy of spikes on a screen.” But spikes are ultimately just “errand boys,” Freeman says; they serve to convey raw sensory information into the brain, but then much more subtle, larger-scale processes immediately take over.

The most vital components of cognition, Freeman believes, are the electrical and magnetic fields, generated by synaptic currents, that constantly ripple through the brain. [...]

The uniqueness of each individual represents a fundamental barrier to science’s attempts to understand and control the mind. Although all humans share a “universal mode of operation,” says Freeman, even identical twins have divergent life histories and hence unique memories, perceptions, predilections. The patterns of neural activity underpinning our selves keep changing throughout our lives as we learn to play checkers, read Thus Spoke Zarathustra, fall in love, lose a job, win the lottery, get divorced, take Prozac.

AI researcher Yann LeCun has made some remarks that seem relevant to me. This is from a podcast quoted by Kenneth Church and Mark Liberman in a recent article, The Future of Computational Linguistics: On Beyond Alchemy:

All of AI relies on representations. The question is where do those representations come from? So, uh, the classical way to build a pattern recognition system was . . . to build what’s called a feature extractor . . . a whole lot of papers on what features you should extract if you want to recognize, uh, written digits and other features you should extract if you want to recognize like a chair from the table or something or detect...

If you can train the entire thing end to end—that means the system learns its own features. You don’t have to engineer the features anymore, you know, they just emerge from the learning process. So that, that, that’s what was really appealing to me.

That second paragraph is the important one. It means, to use a phrase I’ve come to favor, that the mind is built from the inside.

And that, it seems to me, is what makes artificial neural nets (ANNs) so interesting and powerful. Individual ‘neurons’ in these nets resemble real neurons about as much as the smiley face emoticon resembles the Mona Lisa. ANNs depend on backpropagation, which doesn’t seem to exist in real brains (see, for example, Grace Lindsey, Models of the Mind, p. 82). But ANNs learn ‘from the inside’ because, like real nervous systems, they are fully structured ‘end-to-end’ (receiving external inputs and generating external outputs).

Why do we need a genotype-phenotype distinction for cultural evolution? Because minds are built from the inside.

When I first published about the process of cultural evolution a quarter of a century ago, in Culture as an Evolutionary Arena [1], I insisted on distinguishing between a cultural analog to the gene and genotype and a cultural analog to the phenotypic trait and phenotype, but didn’t have a well-thought out way of characterizing either. I made the distinction because, well, it was there in biology so it ought to be there in culture, no? And it made a kind of sense, having one context in which selection is made and a different one in which variation is made.

Since then I’ve devoted a fair amount of effort to clarifying the distinction [2], though it’s not clear to me that things have settled down. I like the notion of coordinator for the genetic element, but I’m still not sure about the phenotypic element, but it seems that cultural being is the term I’ve provisionally settled on. No matter, not do I intend to settle that now.

Rather, I remain interested in the issue of why make the distinction at all. Sure, there’s the parallel with biology, but as culture is otherwise so very different, is that a sufficient reason? Is the logical structure of the mechanism all that important? Well, if it is causal, then yes it is.

The issue is one of empirical method. We have many studies of cultural evolution in which things are counted and measured. As far as I can tell, these things are never characterized as either genetic elements of phenotypic elements. They’re simply the stuff of culture. If the distinction is without empirical consequences, then what’s the point?

Good question. And I’m not sure of the empirical consequences. But I do think there is a reason to maintain the distinction.

It is because minds are built from the inside [3]. From that it follows that cultural beings must be built from the inside as well. Cultural beings are constructed over coordinators, which are physical phenomena in the external world. Those phenomena trigger cascades of neural activity which then coalesce into cultural beings.

The Kuhnian notion of the paradigm is relevant here, where paradigms are collections of coordinators and cultural beings. Paradigms are more or less impervious to one another because of, on account of, those coordinators. What functions as a coordinator in one paradigm doesn’t necessarily function as a coordinator in another.

But what are the observable consequences? What forces us to treat a countable ‘thing’ as a coordinator rather than as a phenotypic element, or even as a Dawkinsian meme? 

Note: I seem to have made this argument back in 2014 as well.*

References

[1] William Benzon, Culture as an Evolutionary Arena, Journal of Social and Evolutionary Systems 19(4), 1996, 321-362, https://www.academia.edu/235113/Culture_as_an_Evolutionary_Arena.

[2] Here, in reverse chronological order, are some steps in that process.

William Benzon, Cultural Evolution: Literary History, Popular Music, Cultural Beings, Temporality, and the Mesh, Working Paper, 2015, https://www.academia.edu/10263479/William Benzon, Cultural_Evolution_Literary_History_Popular_Music_Cultural_Beings_Temporality_and_the_Mesh.

See section 7, The Construction of Cultural Beings, pp. 46-54.

William Benzon, Cultural Beings Evolving in the Mesh, blog post, New Savanna, Jan 5, 2015 https://new-savanna.blogspot.com/2015/01/cultural-beings-evolving-in-mesh.html.

William Benzon, Cultural Beings & Intertextuality: Information, blog post, New Savanna, Jan 28, 2015 https://new-savanna.blogspot.com/2015/01/cultural-beings-intertextuality.html.

William Benzon, Cultural Beings, the Ontology of Culture, and a Return to Books and Blues, blog post, New Savanna, December 15, 2014, https://new-savanna.blogspot.com/2014/12/cultural-beings-ontology-of-culture-and.html.

*William Benzon, Why Cultural Evolution Needs a Distinction Between “Genes” and “Phenotypes”, blog post, New Savanna, November 21, 2014 https://new-savanna.blogspot.com/2014/11/why-cultural-evolution-needs.html.

William Benzon, Terminology for Cultural Evolution: Coordinators and Phantasms, blog post, New Savanna, November 8, 2014 https://new-savanna.blogspot.com/2014/11/terminology-for-cultural-evolution.html.

William Benzon, Cultural Evolution, Memes, and the Trouble with Dan Dennett, Working Paper, August 2013, 67 pp., https://www.academia.edu/4204175/Cultural_Evolution_Memes_and_the_Trouble_with_Dan_Dennett.

In particular, see the section, Memes as Data: Targets, Couplets, and Designators, pp. 44-46.

[3] As I’ve argued most recently in this post, Minds are built from the inside [evolution, development], blog post, New Savanna, Sept. 15, 2020, https://new-savanna.blogspot.com/2020/09/minds-are-built-from-inside-evolution.html.

Reading the Human Swarm 9: What about the brain?

There is at least one place where Moffett mentions the brain, in the beginning of Chapter 12, “Sensing Others”, where he mentions the work of Uri Hasson, a neuroscientist who investigates neural activity of people intereacting with one another through conversation; in this case, they were watching a film and chatting together. When that happens their brain activity becomes synchronized.

I’d like to introduce some comments on the brain on a different theme: The brain is built from the inside? By thinking that through we can appreciate the (conjectural) neural underpinnings of some findings Moffett has introduced: societies are more likely to form through fission rather than merger; the development of factions; and the distinction between group and society.

Brains are built from the inside

First, what does that mean: the brain is built from the “inside”? Automobiles, for example, aren’t built from the inside. They are designed by engineers and then constructed, to design, by teams of workers, from the “outside”. The same is true of computers. We design them and then build them. Software is like that as well. In these cases, and many more, what is being designed and constructed is one thing, over there, if you will, while the team doing the designing and constructing is another thing, over here.

Brains aren’t like that. Brains and nervous systems are embedded in animals and the process through which an animal comes into being is a biological process of development and growth that starts with a single fertilized cell. There is no separation between a design-build team and the thing being constructed. They are one and the same. It is in that sense that I am asserting the brains are built from the inside. There is no external agent doing the constructing.

While it is common to use computing as a source of metaphors for mind and brain (brain as wetware, mind as software), much of that usage doesn’t hold beyond casual conversation. It is easy to give a computer new capacities by loading a new software package, but nothing like that is possible for humans. We can learn new skills, but that takes time, hours, months, years, depending on the skill. In contrast, a software upload takes whatever time is required for the installation, seconds, minutes, maybe hours; but once the software is installed, the new capacity if fully there and ready to use. Conversely, software can be quickly uninstalled, or simply erased, without damage to the computer itself; but no such thing is possible for knowledge and skills a person has learned. Once the knowledge or skill has been acquired it is more or less permanent, though it may degrade over time if not used.

Implications

Let’s begin with the observation that new societies rarely arise from the merger of existing ones (pp. 283-285). Rather, they arise from the fissioning of existing societies. Thus the members of the societies that result after the split already know one another. They’d grown up together and lived with one another for years. When any one of them was maturing, and thus their brain was developing, the other members of the society where automatically incorporated into their neural representation of the world.

Similarly, as societies become larger, the scope of individual contacts relative to membership in the society as a whole becomes smaller, making internal cohesion more difficult. We can conjecture that this is a function of the capacity of single brains to maintain highly detailed individualized (“face-to-face”) representations of others. Factions develop, Moffett argues (243 ff.). At first they’re amicable, in time, though, that frays. The society must then split in order to maintain peace, peace in two new societies.

Then we have the distinction Moffett makes between a group and a society. Members of groups recognize one another as individuals and interact with one another in individualized ways. Societies do not. Societies divide between the world into US (members of a society) and THEM (non-members). Ants do not recognized one another as individuals, but they do recognize other members of their society (coloney); they do so by scent. Humans typically live in societies consisting of multiple (face-to-face) grounds and use markers of various kinds – hair styles, body markings, clothing, language – to recognize and interact with society members who are not in their local group. They interact with non-group society members through stereotyped social roles. The neural requirements (again, my conjecture) of interaction governed by markers and roles are lower than those governed by detailed individual knowledge. The existence of markers and roles thus allows people to form societies that are larger than face-to-face groups without making burdensome neural requirements.

A concluding note on identity is a different sense

Identity is an important theme in the book where it is about an individual’s link to their group and society. There’s another aspect of identity which Moffett doesn’t deal with and which I’d like to link to the brain, but that would require more of an argument than I can undertake here.

The argument that needs to be made is that our nervous system affords us open-ended awareness of the world. I suspect that’s a joint product of the active nature of the nervous system and the emergence of language. On that active nature, the nervous system doesn’t passively take the world in, but rather actively probes the world through continuously projecting expectations – think, for example, of the model William Powers developed almost a half century ago in Behavior: The Control of Perception (1973). Thus perception is a process of verifying those projections (or, to use a more current language, updating Baysian priors).

The emergence of language leads to an endless curiosity about everything: What’s that? How does it work? Where’d it come from? Living becomes thus becomes a dialog with the world. And the question, Where did WE come from? will arise in that process. The answer initially takes the form of myth, of stories about origins. And those stories, in effect, establish the link between a society and world. That too is a matter of identity.

Why Cultural Evolution Needs a Distinction Between “Genes” and “Phenotypes”

I’m thinking I’m about to burn out on cultural evolution, so this will be relatively short and informal.

* * * * *

Ever since I began thinking about a Darwinian process for cultural evolution back in the mid-1990s I’ve insisted on making a distinction between phenotypic entities (which I’m now calling “phantasms”) and genotypic entities (which I’m now calling “coordinators”). Why? My basic reason was to preserve the analogy between the cultural evolutionary process and biological.

That’s understandable, and it was a reasonable thing to do – back then. But there’s been a great deal of discussion about whether or not such a distinction needs to be made for cultural evolution, and if so: how do we make it? Some thinkers, like Dennett and Blackmore don’t make such a distinction at all, being content to theorize about memes, which are thus more like viruses than genes. To be sure, they’ve not gotten very far, nor for that matter has anyone else. But still, the issue must be faced, for there needs to be a better reason for such a distinction than the mere logic of analogy.

After all, what if the underlying logic of cultural evolution is different from that of biological evolution? What if there is no distinction comparable to the genotype-phenotype distinction?

My contention is that there is such a distinction and I’m now prepared to offer a reason for it:

Culture resides in people’s minds and the mind is in the head. We cannot read one another’s minds.

The environment to which cultural entities must adapt is the collective human mind. That’s been clear to me for a long time. And, of course, various conceptions of collective minds have been around for a long time as well. The problem is to formulate a conception in contemporary terms, terms which admit of no mystification.

I did that in the second and third chapters of Beethoven’s Anvil (2001) where I argued that when people make music together, and dance as well, their actions and perceptions are so closely coupled that we can think of a collective mind existing for the duration of that coupling. There are no mystical emanations engulfing the group. It’s all done through physical signals, electro-chemical signals inside brains, visual and auditory signals between individuals.

In this model the genetic elements of culture are the physical coordinators that support this interpersonal coupling. These coordinators are the properties of physical things – streams of sound, visual configurations, whatever – and as such are in the public sphere where everyone has access to them. Correspondingly, the phenotypic elements are the mental phantasms that arise within individual brains during the coupling. These phantasms are necessarily private though, in the case of music making, each person’s phantasm is coordinated with those of others.

If those phantasms are pleasurable – I defined pleasure in terms of neural flow in chapter four of Beethoven’s Anvil – then people will be motivated to repeat the activity and those phantasms will thus be repeated. One of the factors that lead to pleasure is precisely the capacity to share the experience with others. The function of coordinators is to support the sharing of activities and experiences. Just as genes survive only if the phenotypes carrying them are able to reproduce, so coordinators survive only if they give rise to sharable phantasms.

Culture is sharable. That’s the point. If it weren’t sharable it couldn’t be able to function as a storehouse of knowledge and values.

* * * * *

That, briefly and informally, is it. Obviously more needs to be done, a lot more. I can do some of it, though not now. But much of the heavy lifting is going to have to be done by people with technical skills that I don’t have.