Development Forum |
|
A bunch of questions about C2 dendrites | |
|
0renji
|
11/16/2018 | 2 |
So, I've been doing a hobby project in Unity to replicate a norn brain. I set up the lobes, SVRules and basic dendrites alright, but I'm scratching my head over dendrite connection/migration and instincts.
1. What's the deal with how the decision dendrites are connected? You can see that there's a wide swathe of concept neurons in the middle of the lobe which never connect to anything. Both standard and Canny norns look like this. Nova Subterra do look quite different, due to the larger lobe size.
Brain Images
At first I thought it was down to instinct processing making the brain structures similar, but a norn with no instincts and no decision dendrite migration looks the same.
2. In thís case, how does instinct processing even work? What if the norn's brain just doesn't have the right concept linkages at birth to result in the instinct's decision? Is that possible?
3. I figured the answer to 2 was dendrite migration, but I can't say I've actually seen a dendrite migrate, or seen conash/decash rise in a norn, while I've been monitoring. Given that the decision dendrites start at a low initial strength, shouldn't I be able to see SOME evidence of migration? It's unclear to me whether migration would be random or within the same fanout as the initial connections (in which case I'd never see a difference visually).
Cheers for any info you have, here's to everyone who's still developing for this game |
0renji
|
7/1/2020 | 2 |
So, 2 years later, I answered some of these questions for myself. Thought I'd better come back and update this thread in case anyone else has the same questions.
Here is an image I plotted myself of a C2 hatchery norn's concept lobe showing all connected dendrites at once.
So yes, in standard Norns, the entire middle of the concept lobe is unused. This means the base-game and Canny concept lobes are only ~40% the size they should be, and all decisions have to share the same concept cells rather than being spread across the concept lobe as presumably intended.
Dendrite migration only occurs within the initial 'spread', so dendrite migration cannot ever result in the usage of this extra space.
This is also the case for C1 Norns, except I think the situation is even worse, as the dendrites don't seem to wrap around the sides of the lobe as they do in C2. So the first decision cell gets a very small number of concept cells to work with. The add-on Norns (Sid and Nancy) do use the full size of their expanded concept lobes. |
bedalton
|
7/2/2020 | 1 |
I can't help, but I thought this post was really cool. I looked at the first post last year when I was trying to emulate Creatures in Unity. I have been trying trying to learn about dendrite migration myself, but haven't had much time to setup experiments to track things like this. So thank you for this post. It's great.
How is your brain emulation going two years later? |
|