natselrox wrote:May I suggest something as basic as Sean B.Carroll's 'Endless Forms Most Beautiful', twistor? It's very hard to try and frame an answer to your questions.

Already sitting in my cart at Amazon!

Mr.Samsa wrote:Yeah I understand that complexity of behavior doesn't necessarily equate to more genetic material, but I thought Twistor was asking about more (and distinctly different) behaviors. So rather than it being a case of a simple code that is like a "keeping adding one more section" rule that gives you the scales on a snake and their source of locomotion, it's more like having a code for a heart, a liver, lungs, etc. So I would have thought that the more biological components you need to create, then the more material you need (that is, when it can't be built using some simple rule; for example, a heart can't be created by adding a rule to a liver that says "create a second liver but make it pump blood").

Is that wrong?

It's invalid if not entirely wrong. Consider stem cells. They are each identical, but depending on the environment they find themselves in will differentiate into heart cells, brain cells, liver cells, etc. Each cell of your body contain your whole genetic code, and had a common initial beginning. At least early on in development. So your heart, liver, etc., is the same code finding a different environment to respond to. Now once differentiated each system of your body works together. So making a single change in how one region differentiates changes how it responds to interacting systems in the body, thus producing many functional changes well beyond that one feature of that one system the code regulated.

Consider megacolon in rats, where their bowel builds up in the body resulting in death. It results genetically from a failure of cells to migrate from the spinal region to their place in the body. Thus one of the effects is white blazes or spots in their coloration in abnormal places. Though this is fatal, here you have a genetic problem with the colon where risk factors can be seen in the color of their fur. In canines, the genes that code for flight distance also code for maturity. Hence dogs maintain puppy like behavior in adulthood in comparison to their wolf ancestors. Russian experiments with domesticating foxes have lead to the same result. Including floppy ears and changes in fur coloration. So simply breeding them to be less fearful of people prevents them from fully maturing in many ways.

The biological complexity verses DNA code size is a lot like the "n verses np problem". It really has no answer, and the fact that biological systems are massively parallel chemical systems means that very simple code sets/sizes can lead to arbitrarily complex systems. We would have to know every possible combination allowed in nature to answer to these questions numerically.

Good points, I see what you mean now.

Mr.Samsa wrote:
Anyway, how do these complicated behaviors like (possibly) nest building and kelp gull pecking get passed on? As I said earlier, even though our brains are modified across our lifetime through learning, we are also born with a set of neural networks already in place.....
Both processes rely on selectionist principles, the former through the culling and strengthening of neural connections, and the latter by the culling and selection of individuals that have more adaptive neural networks in place.

[my emphasis]

I know this isn't your area, but can you or someone in that area expand on the concept of "neural networks/connections" and how they determine behaviours, and also how they are "culled"? I imagine it is something like sight, which follows a distinct pathway from eye to the visual cortex. Is that right? If so, what is the mechanism that makes the flow of information/chemicals take that path? Do individual neurons only have two connections - in and out?

To understand neural networks you need to understand Hebbian learning, also anti-Hebbian learning. A highly simplified description of the Hebbian rule is "cells that fire together, wire together". So when your a baby a slaps the toy airplane hanging over the crib, the neurons associated with arm muscle control and the visual effects of setting the airplane in motion wire together, because these sets of neurons fire together due to being tied to sensory data.

Now when you are born there is no sensory experience in order for learning to be a significant factor. However, there is a general rough mapping that wires together different brain regions, corresponding to different sensory functions, in a general pattern. This happens even without learning, even though much of this early wiring can be halted by a lack of sensory data. The first rough draft of this wiring can wire emotional states to hunger, pain, crying, etc. It also defines certain reflexes like the diving reflex, suckling, etc. In humans this prewiring seems, with a few exceptions, to mostly associate emotional states to sensory data, rather than specific actions. This wiring is then refined through learning. The things you learn long term will effect the physical development and wiring of the brain. It's simply more advantageous to learn about the world you live in than for evolution to try to predict and hardwire what you need to know to survive, because the environment is not that predictable. Learning also reduces the number of details, required for instincts, that has to be encoded in the initial conditions defined by DNA.

------------------------
I like using self syncing metronomes as a toy model to help intuitively understand Hebbian learning.
[youtube]http://www.youtube.com/watch?v=W1TMZASCR-I[/youtube]
Given the solid base, these metronomes are all essentially wired together from the start. But imagine that the wooden base had a variable elasticity, and that each metronome increased frequency due to sensory input. Like a metronome tied to each pixel of a camera. When sensory data excited some subset of the metronomes the rigidity of the base between those metronomes, while the rigidity of the base between metronomes that don't match decreases. Basically this would store a single experience. So increasing the frequency of any one metronome associated with that experience will cause all the metronomes associated with that experience to sync up, but not the ones not associated with that experience because the placidity is too weak. So triggering any part of the memory trace will activate the entire memory trace.

Now when you overlay a new experience, the parts of the the new experience that match the old experience will share the same metronomes, because they are tied to the same sensory connections. Meanwhile the propensity for all metronomes to self sync will, over time, reduce the complexity of the connections sets to a minimum. Which is why memories tend to fade if not reinforced with new experiences, or false memories can be reconstructed from associative memories.

I can take this qualitative analogy much much farther, but that's the basics in the simplest form.