A single water molecule is not wet. Put enough of them together and wetness appears. A single neuron does not think.

That is the observation, and on its own it is completely uncontroversial. The disagreement is over what follows from it, and the whole subject turns on a distinction that the word itself hides.

Weak and strong, and why the difference is everything

Weak emergence is the claim that a system's behavior cannot in practice be predicted from a description of its parts, even though it is entirely fixed by them. Nothing new exists. The novelty is in our epistemic position — the computation is intractable, or the relevant pattern is only visible at a higher level of description.

Almost every honest use of the word means this. Traffic jams are weakly emergent. So are market prices, flocking, and the behavior of large codebases. Nobody claims a traffic jam exerts a force the cars do not.

Strong emergence is the claim that the whole has properties that are genuinely new — not fixed by the parts, and capable of making a difference to what the parts do. This is a substantial metaphysical thesis and it is far less often defended than the word's frequency suggests.

The two claims are stated in the same sentence. Consciousness emerges from the brain is compatible with both, and which one is meant usually determines whether the sentence is trivial or revolutionary.

The argument that makes strong emergence expensive

Suppose a high-level property does real causal work — the emergent mental state causes the arm to move.

The arm's movement also has a complete physical explanation in terms of muscle, nerve and neurotransmitter, each step of which is fixed by the step before it. So either the movement has two sufficient causes, which is overdetermination and looks like a coincidence repeated every time anyone moves, or the physical story is incomplete and there is a point at which physics is wrong about what happens next.

That is Jaegwon Kim's exclusion argument, and it is why strong emergence is not a moderate position. It is the denial of causal closure, and the cost of holding it is the same cost that dualism pays.

This is also why the wetness example is so misleading. Wetness does nothing the molecules are not already doing, so it is no model at all for the case people actually care about.

The case where the strong claim really is asserted

Consciousness is where this stops being an academic distinction.

Nobody is puzzled that a brain processes information; that is the kind of thing an organized system can obviously do. The puzzle is that there is something it is like to be the system — the qualia problem — and no description of neural organization seems to entail it, however complete.

Which leaves the choices bare. Either the felt quality is weakly emergent and a sufficiently complete physical story would explain it, and the appearance of an unbridgeable gap is a limitation in us. Or it is strongly emergent, genuinely new, and the physical account of the world is incomplete. Or it was never absent from the parts, which is panpsychism and is taken more seriously now than it was for most of the last century.

Notice that none of these is comfortable, which is a fair sign the question is real rather than verbal.

What to do with the word

Treat it as a flag rather than an explanation.

When something is called emergent, the useful question is: does this property do anything, or does it only describe? If it only describes, the claim is weak, probably true, and not very interesting — the higher-level vocabulary is a convenience and the argument is over.

If it is supposed to do something — to make a difference the parts would not otherwise have made — then a large bill has arrived, and it should be paid explicitly rather than smuggled in on a word that also covers traffic jams.