Phantasian Biology 101: Lesson 1; Emergence

·

·

What are little boys made of?
What are little boys made of?
Slugs and snails and puppy-dogs' tails
That's what little boys are made of

What are little girls made of?
What are little girls made of?
Sugar and spice and all things nice
That's what little girls are made of

So, what are little phantasians made of?

‘Computer code’, you might reply. ‘Bits and bytes and all things device’. You certainly wouldn’t be wrong to say that, but it’s not exactly right, either. I’m an Artificial Life researcher, and the ‘artificial’ part of that phrase does indeed refer to computer programs, while the ‘life’ part is fundamentally about autonomy. Life is autonomous, while non-life is not.

But here’s the thing: You cannot program something to be autonomous. It’s a contradiction in terms.

Just to back this up a little, the word ‘pro-gram’ basically means to ‘write ahead’, as in a definite plan. I think I’m right in saying it was Alan Turing who first coined the term ‘a definite method’, which today we call an algorithm. Programming is about saying, in advance, exactly what a computer should do, one blind step after another, to reach a specific goal. Back in those days, ‘computers’ were actually people, not machines, but the emphasis was on the blind, step by step nature of it; a series of actions that somebody or something could dutifully carry out, without actually needing to understand what those steps were trying to achieve.

Autonomy, on the other hand, is auto-nomy, or self-law. It implies something that chooses its own law, or its own rules. In other words, unlike the ‘computers’ of the day, it doesn’t follow somebody else’s. If we dig any further than that, it starts to get a little messy: People in my world often talk about ‘autonomous agents’, and an agent could be somebody who does something for you, like a travel agent. That seems like the very opposite of autonomy, frankly, but what they’re getting at is something that can do tasks on our behalf, such as plan a trip for us, all by itself, i.e. autonomously.

I’ve never had any interest in that sort of autonomous agent myself, but those of us over on the A-life side of autonomous agent research mean something quite different by it. An autonomous agent, to us, is an agent that does what it wants to, or evolution wants it to, rather than something we want it to. An autonomous entity is ‘self-motivated’, which taken literally means it is self-moving, although we often take it to mean it is self-steering and self-starting; it doesn’t merely move under its own steam, but at its own direction and in its own time. An autonomous agent is thus one that chooses its own future, even if nobody has quite figured out what that concept really means.

You cannot program something to be autonomous

Either way, the idea of writing out a sequence of program steps to be blindly followed, seems fundamentally at odds with a system that chooses things for itself, doesn’t it? A natural living being may suddenly decide that it doesn’t like carrots, or it’s going to take up knitting, or it has fallen in love with another living being and can’t bear to be parted from it. The idea of telling such a system when or how to fall in love, let alone with whom, is clearly the very antithesis of enabling it to be autonomous.

So how do we resolve this paradox? How do we make autonomous life appear inside a totally deterministic and predictable computer?

Well, I think the answer to this is pretty deep and philosophical, and I don’t have space to cover all of it today: I’ve already written two whole books about it. But obviously, learning is one really key ingredient. A system can certainly be explicitly programmed to learn things, perhaps by trial and error. If we don’t tell it what it should learn, it’s fair to say that when it later acts in response to the things it did learn, it is doing so somewhat autonomously. It certainly didn’t do it because we told it to. Although, having said that, do bear in mind the distinction between learning and training, since anything that has been trained is essentially still doing what it was told to do; it just wasn’t told how to go about it.

Feelings are also a fundamental element of autonomy, I think, because our feelings are what generate and control our wants, and wanting is fundamental to us choosing to do something. A chess computer literally can’t stop itself from playing chess; it can’t decide that it doesn’t want to, and nor does it care. But that’s a story for another day.

The third main ingredient is to do with chaos and complexity, which also I’ll talk about another time. Possibly many times. But this involves making a distinction between processes and structures, which I’ll talk about now. Computer programs inherently define processes – ‘first do this, then do that’ – but in essence, the whole thing can also be turned inside-out. Instead of code performing operations on data, the data can instead take charge of the code.

I don’t want to put non-computery people off by getting too deep into this, but it’s worth saying a few words about software objects, for non-programmers. ‘Object-oriented programming’, for instance, is… well, oriented towards programming objects, hence the name. Such software objects are, in certain ways, just as much things as they are lists of instructions, and these things can be connected together into networks, i.e. structures.

Most of the time, such virtual objects merely exist to keep the facts about something (the data) conveniently bundled together with the processes that operate on those facts (the program). An object representing a document, for example, might contain the words and images in that document, packaged together with the program code that knows how to print those words, save them to disk, spell-check them, and so on.

Such objects are usually quite notional, abstract things. Nobody would suggest that an employee object inside a company accounting system is actually called Susan – it’s just an abstract representation of Susan, for accountancy purposes. This virtual Susan doesn’t own a dog, she isn’t currently on the fourth floor, she isn’t worried she’s putting on weight, and so on. It’s only a very few of the properties describing the real Susan, like her salary, stored alongside the code for making sure that the real Susan actually gets paid.

But sometimes, software objects can represent rather more ‘physical’, ‘solid’ things, including biological things such as molecules, neurons, or sensory organs. Again, nobody would be dumb enough to think that a software object representing a molecule is actually a molecule. Virtual hydrochloric acid can’t burn your skin, any more than using your mind to imagine a brick makes your head heavier. BUT, here is where something really quite magical can happen, I think. Although it’s a bit hard to explain.

I got to hold the brain of somebody who had just died.

I’ll tell you a little story.

When I was 17, I got to hold the brain of somebody who had just died.

When I say I held it, I don’t mean I rolled it around in my bare hands: I just held the tray that it was sitting in, so that the pathologist could have his hands free to point at interesting bits of it. I had mildly impressed this doctor with the fact that, in my autistic way, I knew how to stain tissues with substances like Eosin and Methylene Blue, and could hold a meaningful conversation with him on the topic of Delafield’s Haematoxylin, so perhaps this was why I became the person on our school trip who got charged with the sacred duty of holding this poor person’s brain. I don’t know.

I don’t really know what was going through my head at the time, either. I was only 17, and painfully shy, so I’m sure most of my thoughts involved flustered embarrassment about having been singled out to be stared at by my whole biology class, all of whom were girls. Nonetheless, even fifty long years later, I do distinctly remember the sudden, electric shock of realization that the thing I was holding in my hands might, just an hour or so earlier, have been feeling quite scared.

This sloppy purple and red object might have been really worried about what its wife was going through as she stood by the bedside, or perhaps it was sad that it would never see its grandchildren again, or it was embarrassed that its body had started to seriously misbehave itself. That same body was now lying nearby, off to one side of the room, which itself is quite an odd thing to think about. Meanwhile, the wife in question and her hypothetical grandchildren were presumably trying to come to terms with the recent loss of whoever had been ‘inside’ this now cold lump of bloody fat, of which I had been put in charge.

The reason this story came to my mind is that this squishy lump that was lying in my hands, was a physical machine. A whole network of physical machines, in fact. Its atoms danced according to the laws of physics, and always had. And yet, a short while earlier, it been filled with thoughts and feelings, hopes and fears and intentions. They were once part of its physical structure; part of its dynamics. The brain itself was merely a collection of neurons and glia and neurotransmitters, obeying the blind, step-by-step processes of chemistry, and yet its owner was an autonomous agent, with opinions and likes and interests. Right up until the moment when it suddenly wasn’t.

Now, if I attempted to write a computer program to behave like this person (which would be many trillion times harder than many of my own colleagues like to pretend), it clearly wouldn’t actually be a person. It would be more like a statue of a person. A simulacrum. It wouldn’t eat carrots because it wanted some carrots or because it really liked them; it would eat them because I had explicitly told it to. If it spoke to you in English, and told you how much it loves carrots, you would be wrong to believe it, since I would have told it what to say. Even if it had learned for itself how to realistically answer the question “do you like carrots?”, just like chatGPT can, it would still be quite wrong of you to believe it. It wouldn’t have learned to like carrots; it would only have learned to say what people statistically tend to say when they’re asked questions about carrots. It has never tasted one. Not even a virtual one.

Exactly the same is true if I tried to write a computer program to behave like a single neuron, too (these are also more complex than certain colleagues like to pretend). It wouldn’t actually be a neuron; it would just behave more or less as if it was a neuron. It would be fake. But here’s the interesting thing: Simulated neurons are not actually neurons; simulated enzymes are not actually enzymes; simulated muscles don’t actually lift things. And yet, if a suitably arranged collection of hundreds and hundreds of these virtual objects, starts collectively to behave like a similar population of real, physical ones, why should that behavior be regarded as any less real? It may not be real in our own universe, but it’s doing the same things, for the same reasons, inside another universe. The system as a whole wasn’t programmed to do those things – they emerged spontaneously from the interactions of its parts. The behavior is autonomous.

I usually end up losing control more in the manner of a drunk on a bicycle

Anyway, this is a really deep topic that I could wax lyrical about for hours, especially if you have objections and doubts that I would need to address. But for now, I just wanted to point out that both of these approaches exist inside the virtual universe of Frampton Gurney – both the explicit programming and the emergent behavior.

The house you see in Frampton isn’t a real house, obviously. The trees are not real trees and their leaves are not really blowing in the wind. Even the weather is fake, although it’s a little less fake than is normal for a computer game. This aspect of the project is all about a computer following pre-programmed rules, just like in any other 3D game.

But when it comes to the creatures themselves, the entire point is for me to let go control, and allow their behavior to emerge spontaneously and autonomously out of zillions of tiny, non-living parts. Ideally I would do this in much the same way as a skillful writer might let the words more or less write themselves, or a sculptor might sensitively allow the wood to define what the sculpture becomes. In practice I usually end up losing control more in the manner of a drunk on a bicycle, but nevertheless, intelligently, knowledgeably and thoughtfully letting go control is what it’s all about. [If this idea intrigues you, you might want to check out or re-read a book by Kevin Kelly, called ‘Out of Control’]

And so, to really understand these creatures’ behavior, we have to understand the virtual objects from which they are made, and why I put them together in the ways I have. This is no longer programming, it’s biology! Not just in pragmatic ways, but also in theoretical and philosophical ones.

If nothing else, when you see your creatures do weird and unexpected things (and they will, a lot), this doesn’t necessarily mean there’s a bug in my program, which you can just report to me and I’ll go ahead and fix it. Sometimes that’s going to be true, but quite often it will mean there’s a ‘bug’ in the biological structures that I’ve assembled instead. There may be something quite wrong with my theory of the brain. I’m actually quite cool with that – nobody on earth knows how the brain works, so my own theory is almost certainly going to be wrong too. I’m learning as I go. All I ask, really, is that you don’t make the mistake of thinking that I’ve programmed these creatures to behave in a certain way, and so if they do something wrong, I just need to modify my code to make them do it right. It’s not me who tells them how to behave. In fact it may even be their fault! They’re autonomous agents, after all. I’m letting go control, so that they can hopefully take control of their own lives.

Anyway, this was just by way of a broad, philosophical introduction to an upcoming series of blog posts I want to write about phantasian biology: what their bodies and brains are made from and how they’re put together. Over the coming weeks, I’ll talk about their internal organs and sensory systems, their biochemistry, their genetics and their neurology. There won’t be an exam, I promise, and this isn’t a ‘biology game’ (unless you personally want it to be) but I hope you’ll find it interesting. With luck, you may even learn a little bit about how you are put together, too.

I barely remember our school trip to the path lab, where I was able to hold this poor man’s brain just a few minutes after he had died. A couple of years later, though, I discovered computing for the first time, and these two events are not unconnected in my mind. When I say I discovered computing, I don’t mean computers were already there and I just hadn’t discovered them for myself yet. I literally mean this was the beginning of the personal computer revolution. The microprocessor had only just been invented. It was back when people like Bill Gates and Steve Jobs and their friends were still tinkering with things in their garage to see what they could do. It gave me a different perspective on computing that I think would be quite hard to recreate today. Gates and Jobs went on to make a lot of money and change the world, while I basically stayed in my garage. But I never forgot this sad stranger’s brain, and how all those thoughts and feelings could be there one minute and gone the next. I’m still working on understanding it today, and I’m really looking forward to telling you what I’ve found so far. Watch this space!


5 7 votes
Article Rating
Subscribe
Notify of
40 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Mabus
Mabus
17 days ago

So far, this was my favourite blog post here!

And you teased upcoming explanations about their biology, neurons and similar.

May I suggest, activating the cheat tools one by one, once the blog post about that is written?

That way we all will explore the same tool together, while we got a brief understanding of it. That means it would use a step by step approach to make sure most people here will understand all tools.

That means once phantasia goes mainstream, you don’t need to bother explaining all that stuff to the new people, since we can do that instead.

Also “nobody would be dumb enough to think that a software object representing a molecule is actually a molecule” – don’t underestimate ” phantom pain” of some VR users….. Or should I rather say the warped perspective of their reality?

SoulSkrix
SoulSkrix
17 days ago

Wonderfully written! Maybe if you find the time after all of this, you’ll have to write a book on your experience with the FG project.

I’m at work, and hiding to write out this message – but I couldn’t put the phone down for what was supposed to be just a quick peek.

I can’t imagine the feeling of holding a human brain inches away, but I can say I relate to some degree when I have held dear beloved animals in my hands who have passed away, and it made me think in this “cold” way about life; and to contemplate what separates us from any other form of life, organic or otherwise.

For all I know, I could be inside some sophisticated machine. I certainly hope not, but then again, I’d never know.

Starting out with a little nursery rhyme style intro always helps the homesick feeling, I’m not sure if other would agree but it feels very British to me – old timey and comforting like the books I grew up with.

Looking forward to your next post

damanska
damanska
16 days ago

This was really interesting! I quite liked your books when I read them a few years ago. Looking forward for the following blog posts!

Robowaifu Technician
Robowaifu Technician
15 days ago

Any sufficiently complex process is indistinguishable from randomness.

danielmewes
danielmewes
15 days ago

This is the kind of profound stuff which makes this project so important!
Very nicely written!

Fern
Fern
13 days ago

A moment of silence for Multiversus and the extensive incompetence that went into its sad and painful decline. Actually upsetting. At least we have Marvel Rivals. And your game of course.

Last edited 13 days ago by Fern
neurobot
neurobot
13 days ago

Fantastic! I don’t swim in these waters anymore, but I think my focus on dynamical systems, and general fascination with emergence and complexity/chaos, during my (brief) time as a professor was in large part attributable to ideas your books helped crystallize for me during formative years. Still interested in the topics of course, just avocationally now.

Mabus
Mabus
12 days ago

Tipp for all the other players:

  • force interactions with the ball early on after spawning, that improves their interaction with the world.
Squirrel
Squirrel
11 days ago

I didn’t know there was another book, and I shall have to wait to read it because I have an irrational preference for hardcover which will have to ship from America. Actually I think I may have got the last new hardcover they had, because I only see second hand options now. Fortunately I got some books for Christmas to keep me entertained while I wait, about a party of dungeon delvers who just want to eat all the monsters. My family know me so well.

But thinking about it now I’m curious, what’s on y’all bookshelves?

Chat Icon Close Icon
40
0
Would love your thoughts, please comment.x
()
x