Darwinbots Forum
General => Biology => Topic started by: jknilinux on November 28, 2008, 09:55:53 PM
-
I was just wondering- why are all DB species we've ever evolved r-selected, as in being similar to insects?
Why do we never evolve an elephant-like species that is few in number, lives a very long time, has few offspring, and cares for them?
See this: r/K selection theory (http://en.wikipedia.org/wiki/K-selected)
Or, has anyone evolved a K-selected species? If so, I'd love to see it.
If not, maybe this points to a problem in DB?
Maybe we should make it a bot challenge, replacing the conditional-evolution challenge? That challenge is really old (years?), there's a winner, so it's over, right?
-
There's nothing to teach the offspring, that's the issue. Most bots are hardwired with any behavior or constants, so they don't need their parents to teach them.
-
Hmm. That's definitely a possibility. Maybe there should be a way to change certain bits of code from the offspring automatically, like setting all offspring's ANN weights to 0 at birth. That way, they'll need to learn from their parents.
-
Or, has anyone evolved a K-selected species? If so, I'd love to see it.
Been there, done that. I increased zerobot ''efficiency'' from costx .5 to costx 1.5 in .7 M cycles. How? r/k selection: with dynamic cost toe focus is more on r selection, so don't use it. I started wit a strain of my zerobot @0.5 Population remained stationary @ 1400 bots. I increased costx a little. Population would fall a little, and species diversity some more: K selection (btw: I need a graph of species diversity/population).
-
So, you mean your zerobots would, on average, live a very long time, have lots of nrg, few young, and actually cared for them? I'd love to see a sim!
-
So, you mean your zerobots would, on average, live a very long time, have lots of nrg, few young, and actually cared for them? I'd love to see a sim!
I think you are overstating the implication of the term k selection. None of that is a prerequisite.
K-selection
In stable or predictable environments K-selection predominates, as the ability to compete successfully for limited resources is crucial, and populations of K-selected organisms typically are very constant and close to the maximum that the environment can bear.
is definitely applicable
Traits that are thought to be characteristic of K-selection include:
this does not mean that the following are essential prerequisites for k selection
-
Oh- OK. Well, in that case, I meant "why do we never evolve a bot that has the characteristics we commonly see in K-selected species?"
-
Oh- OK. Well, in that case, I meant "why do we never evolve a bot that has the characteristics we commonly see in K-selected species?"
- Would we recognise them if we saw them?
- is the environment complex enough to warrant teaching and learning?
- is growing big and old a viable let alone favourable strategy withing our stable environment?
The rules define the outcome. I think the IB league discussion is a good example of the issues faced. here we have intelligent beings trying to design a system where designed, complex behaviour has to outcompete ''simple'' r type strategies. Given the discussion this is no simple task.
Here we are discussing not the next step but a mile further down the road: having this behaviour emerge spontaneous.
-
Well, once we make the IBBL, we should be able to run an evosim under those conditions and make complex behavior emerge spontaneously, right?
Anyway,
- For one thing, they'd be physically huge, so we should have noticed them by now. In fact, maybe we have- big berthas sort of are a K-selected species, because they have a small population and are very well suited to their environment, so they become huge and kill everything else...
- Maybe not teaching and learning, since there isn't a single bot in existence that learns during it's lifetime, but at least the basic attributes of a K-selected species.
- It should be a preferable strategy in such an environment. R-selection dominates mainly in new or unstable environments.