The Difference Between Creatures and Phantasia
This document will be relying heavily on my own personal understanding of how these A-Life simulations work. Everything is from a layman’s perspective, I barely made it to being a college drop-out. I am humbled to have been asked to share what knowledge I have gleaned and I will do my best to break things down in the simplest terms I can.
Everything that I say can be assumed to be 40-80 percent accurate. I have no formal training pertaining to this and began dissecting Steve’s life’s work for my own edification and nothing more. I have been snooping through the code and source files for an hour or two a day going on about a week now whilst Steve has been putting in 14 hour days for more than a decade, so please excuse any inaccuracies.
Creatures:
Norns and Grendels were, and frankly, still are an incredibly impressive attempt to distill the essence of life into a knowable quantifiable system. Given the computational constraints of the day (I ran it for the first time on 33 Mhz 486 pc with a massive 400MB hard drive) things in Albia were still very complex. For me personally, the most accessible part of Creatures was the chemistry system. Simple reactions within the creatures would give rise to different behaviors by affecting the strength of their individual drives.
For example, starch converts into glucose which then is converted to energy which fuels the body. Each and every chemical has a half life which determines the rate of decay or how long it tends to hang around in the body, this combined with the reaction rates between individual chemicals leads to a dynamic system which was carefully crafted to parallel actual biological systems. There’s a ton of really thorough information on the creatures wiki.
https://creatures.wiki/Biochemistry for anyone interested in the details.
Along with the chemicals which might be familiar to us, there were also discreet chemicals which would indicate each drive or “desire”. This includes such things as loneliness, hunger, sleepiness and sex drive. I think of them as crude neurotransmitters, these chemicals would come together to determine what the particular creature was “thinking”. Norn brains were essentially evaluating the various drives and outputting simple scripts/actions like “when hungry, move towards food” but they are still incredibly interesting in their own right. I remember reading about someone (probably Steve) explaining that the Norns fundamentally couldn’t think, they could react in seemingly intelligent ways but they don’t have an inner monologue or cartesian theater(if you haven’t read it yet, and even if you have, this https://phantasia.life/phantasian-biology-101-lesson-4-brain blog post is very enlightening.) Again, plenty of people smarter than me have discussed this at length already as it pertains to C1, C2, and C3. https://creatures.wiki/Brain
Another point that bears mentioning in terms of understanding the differences between the two systems (Creatures and Phantasia) is that in the 2d world, nothing had any substance. Norns were only able to interact in very limited ways and nothing had physical properties, Norns could never find an object that was too heavy to lift for example and multiple objects could occupy the same space.
Phantasia:
The chemical system in Phantasia is pretty similar to that of Creatures, though it is still a work in progress. For example, there’s really no need to start introducing toxins and medicines until the creatures themselves are finished. One noteworthy exception is that drive chemicals are a bit more streamlined now. Instead of separate channels for “crowded” and “loneliness” they are now combined into a single chemical which the brain interprets based on it’s concentration meaning that when the level in a bolly’s brain is low, they will be more likely to seek out company and when it’s high, they may prefer to take a walk by themselves to try and return the drive to a balanced state. All of the drives operate like this, with a “baseline” which the brain will always try and bring them back to. I like to think about it in the sense that the further from baseline each drive is, the less comfortable the creature is so it should try to respond accordingly.
In contrast to the Norns, Phantasians live in a fully simulated physical world. The leap in complexity from a 2d system to 3d one is actually a requirement for the new generation of brains, they wouldn’t work as they are if they were transplanted into a 2d system.
The Phantasian’s brains are fascinating. These creatures unlike Norns have actual physical bodies with around fourty to fifty individual components from ears and eyes to bones and skin, each of which have their own region of the brain or “lobe” assigned to them. Sensorimotor organs associated with each body region feed data into their respective lobes given various environmental stimulus.
This is where it gets a bit murky. Every lobe has multiple layers which in Steve’s terminology deal with “yin”(incoming, sensory inputs or present state) and “yang” (outgoing, what’s comfortable to us, the preferred state, or “baseline” I referred to my in drives example).
Essentially some witchcraft / black magic happens in the intermediary layers and the results become “yang” which is passed as “yin” to the next lobe in sequence. https://phantasia.life/docs/ep/biology/neuroscience/brain-maps/
There are approximately 100 of these lobes, constantly communicating with each other and this leads to all kinds of emergent behavior. Phantasians have short term and long term memory, they have empathy towards their parents and offspring and are able to actually think about the world they live in in meaningful ways.
Like I said at the beginning, I’ve only been immersed for about a week, What I’m interested in within the simulation changes from day to day because the more I learn, the more I realize I didn’t understand. Currently I’m thinking a lot about the visual classification system of objects and I intend to write up something on that in the next few days.
I’ve learned a lot that I haven’t included here because I don’t want to be inaccurate and really don’t know where to start, but I’ll do my best to answer any questions and research the answers I don’t have.
Wow! You learned all that in a week?!?!?! I’m going to start sending you my bugs…
I already had a good grounding in creatures, but i’m not quite ready for bug hunting yet, I’ve only just gotten through my introduction to c#, next I’m going to do any intro to game design with unity.
Reminds me of your blog about coding by learning one new instrument at a time.
I take it that you approve of my assumptions?
@foggygoofball May i suggest a compromise? a seperate mutant village where steve stores all the messed up brains. For all neuroscientists.
That way the mutants won’t mess up the population of us normal players (that the day comes when i label myself as normal player…) – and you brain scientists can have your playground?
I think could be a good idea, the normal flock and then the island of Dr Moreau.
@foggygoofball Your assumptions are mostly right. You figured out the bipolar nature of (most) drive chemicals. The chemoreceptors that detect these (in the affect layer of maps) turn the concentration into a degree of discomfort, so full_hungry, for instance, will produce -1 when very full (concentration is 0), 0 when neither full nor hungry (concentration is 0.5), and +1 when very hungry (concentration is 1). -1 and +1 essentially mean the same level of discomfort, which the map wants to reduce, but I keep it signed, because otherwise a change in hunger from 0.4 to 0.6 would seem like nothing has changed, because they’re the same distance from neutral.
The affect layer of maps learns the average drive changes that occur when the map is in a given state (if that state has no consequence then they’ll tend to average out to zero). It records this by migrating dendrites around in 2D space. The affect surface of that map shows how much the current drive levels are predicted to become less or more uncomfortable, for every state the map has ever experienced. The map wants to get into the state that it thinks will produce the most comfort or least discomfort.
It’s quite like Creatures, but often bipolar, as you say, and it tells each map which state it would like to be in, rather than the whole brain at once. Sometimes a map can simply command the state it wants to be in and it will happen. So, if looking far to one side is uncomfortable then the map directly controlling the eye muscles can just choose to look straight ahead instead. But deeper into the brain this becomes more of a problem for the map. It might need to send yang commands to several child maps, to tell them which state it wants them to be in. They in turn will decide whether they want to respond to this top-down command or whether they have plans of their own.
So, if a very high-level map thinks “I’m hungry. I know that eating some food will make me less hungry”, then it has to learn how to tell its child maps things like “I want you to go to where you last saw food”, and “I want you to eat the food”. The first of these will then have to tell other maps things like “plan a route from where we are to where we last saw food, avoiding those places where the scary monsters live or the terrain is very rough.” And the map that knows about those things needs to tell another map, “We’re in the right region now, so plan a route towards where we remember there being some food, avoiding any obstacles you’re aware of”. And that map will need to tell its own children things like, “walk in this direction”, which will involve other maps demanding that their children put the body in a sequence of poses, and adjust the turn rate.
So a high-level desire turns into a hierarchy of commands for lower level maps. Meanwhile, down at the bottom, yin inputs are busy saying things like “our body is almost in this pose”, or “we’re turning left at this rate”. And these get turned into more abstract facts as they move up the tree, such as “we’re currently over here”, or “our eyes are pointing to the left but our head is pointing to the right, so we’re actually looking straight ahead”. Right at the top of the tree, we will hopefully end up with a very abstract yin input that says “we’ve successfully found lunch and we’re eating it”. The level of hunger will then drop, the top-level map will be pleased that it achieved its goals, and it will be even more convinced that doing lunch is a good idea when you’re hungry. Although the lower maps might have their own opinions and send up signals saying “yeah, but that was a long walk”, and “I didn’t like going near the monsters”.
All put together, the top-down yang signals turn abstract ideas into particular intentions, and the bottom-up yin signals turn particular facts into abstract facts. And the affect layers learn the consequences. Meanwhile, everything is negotiating with everything else until the brain as a whole decides what it’s going to do.
How the maps know which pattern of commands to send down to their child maps in the first place is a whole other kind of learning, that I won’t blow your mind with right now! 🙂
@steve I really appreciate your explanation of the maps, It’s all so much to visualise, so i’m building my own mental map from all of the concepts first rather than approach it as a whole. I’m coming to realize what a task it will be to make a physical model of these brains, but I still think it will help me understand better.
Like how I read about 3d printers for over a decade, but never really understood it until I put one together myself. There’s a certain benefit to learning literally hands on, at least for me, I can’t stand learning things from a YouTube video. Just hand me the textbook please! 😉
@foggygoofball Believe me, it took YEARS for me to visualize it myself! But of course I was doing it the other way round.
I totally agree with you about the value of hands-on experience!!! I’ve spent my life trying to build brains, and it’s very notable how almost nobody ever actually tries – they just wave their hands in the air and write fancy books about their theories, complete with the infamous “then a miracle happens” box hidden away deep in their flowchart… But if we want to understand something, we have to build one and see if it actually works. And then if it doesn’t, we know that our theory was wrong!
Ooh, it would be interesting to 3D print, I think! It’s just stacks of plates connected by wires, after all. What printer do you have? Back when I was well-connected to the scientific community (in England, before I left for the US and abandoned everything, which is too painful a story to tell), I used to know Adrian Bowyer, who was just starting out trying to make RepRap into a thing. Do either of those names mean anything to you? We still laugh at each other’s jokes on social media, and I built a RepRap of my own, but I never had the time to keep up to date with the firmware and stuff, because of this killer of a project. I was going to use it to make some creatures to give to people, but the best laid plans of mice…
I have a Prusa mk3 which is currently in need of a little TLC and I recently received an elegoo saturn UV printer as a gift, but it’s a whole different beast and I’ve yet to fully tame it.
I’d dreamed of building a reprap for years but it wasn’t until i convinced my wife on the practicality and the money it could save us around the house that I finally got one (technically not a reprap but it’s still open source). That also led into learning to use CAD software in 3d because I wasn’t allowed to buy a printer until I could design replacement parts. All in all, a worthwhile endeavor if you can find the time, though the dream of the reprap is slowly being eroded by commercial interests.
