Maui said:
Self-organization. SO is not an inherent property of matter(well at least as far as we know), though it's reasonable to suspect that it might be. At the moment it isn't possible to predict, based solely on the laws of physics and chemistry, that life would form out of the interactions of atoms or molecules. Though it happens somehow.
I think it helps to see the issue in terms of three levels of mechanism.
1) atomism: you just have local components and local action.
2) self-organisation: you now recognise that the collective action of local components can develop some self-organised pattern. This is simple self-organisation as we see in chaos modelling and self-organised criticality for instance.
3) intelligent and adapative self-organisation: now something extra is added. By virtue of some kind of systems memory or symbol processing (semiosis), we have rate independent information in control of rate dependent processes (see Howard Pattee).
So life and mind (bios, complex adaptive systems, etc) are distinctive in that they can take SO processes (like a metabolic reaction) and stand back and harness them. The way that DNA can remember when to throw enzymes into a mixture to change its rate. The metabolic process itself is just level 2 SO. But it is controlled by level 3 semiotic organisation.
If you are talking emergence, you will thus have a simple kind of collective property emergence at level 2, but a quite different symbolic, or semiotic, emergence at level 3. And what goes on at level 3 seems now very different because, indeed, it demands a strong disconnection between the rate independent information and the rate dependent process.
So DNA is a very protected and stable molecule that exists "independently", while cellular metabolic processes just run to equilibrium as fast as they can.
Level 3 is quite computational (and not very dynamic). It is just the same as the way the silicon circuits of a computer are engineered to as isolated from real world dynamism as possible. This allows the symbolic code to run as if it had nothing to do with real world rate dependent processes, like battery life, cosmic rays, and other things that would interfere with the pure symbol processing.
I would say that levels 1 and 2 can really be collapsed into the one thing - look closely and all dynamical processes are SO within boundary constraints. QM is an example of that as Pythagorean says.
And level 3 is where the real novelty comes in that "physics" really cannot see at all. But on the other hand, we know all about computers and languages.
Where science is at - when it comes to complex systems, such as organisms with "consciousness" in its many grades and guises - is in trying to put together a combined story on 1/2 and 3.
You can't directly explain 3 with just SO principles (and thus simple emergence). But you still want to be able to build some composite model, some biological ToE, that puts both these aspects of nature, the dynamic and the computational, the rate dependent and the rate independent, into the same story.
Dissipative structure theory is one approach that does seem to bridge the divide. Especially with newer ideas like infodynamics - where the relation between structure and process can be describe more mathematically by combining information theory with hierarchy theory.
Theoretical biology has been talking about this kind of stuff for 30 years now. But reductionism still has such a grip on the popular imagination that even reaching level 2 is a step too far for most.