bf404

That's true - and we see that sort of "seeking" behaviour in single celled organisms as well (e.g. https://www.mit.edu/~kardar/teaching/projects/chemotaxis(And...) - but this can be explained by a pretty simple feedback loop, and so the the difference in complexity between that and, say, the ability to keep track of different individuals in a social group and reacti differently, is orders of magnitude.

That was perhaps misleading phrasing. Animals have both true multicellularity (shared with plants, fungi, and some other algae), and much higher behavioural complexity. Multicellularity is almost certainly necessary for behavioural complexity, but it obviously isn't sufficient, because as cool as plants and fungi are, you don't see any of them doing anything that one can imagine leading to intelligence in any form that we'd recognise.

Now, broadly, I think (in my capacity as an armchair biologist) what we care about here is an nervous system, or something like it that is capable of fast communication and complex feedback loops that can be built into complex "computation". Interestingly, nervous systems aren't present in sponges, which are most likely the basal animal group (and which are immobile), but are present in all other animals. I'm therefore speculating that "able to develop a nervous system" could be the filter we care about for behavioural complexity.

Being non-photosynthetic (so more advantage to being able to move around and find food), not having cell walls (so potentially more able to be motile, and form flexible structures), as well as just whether there are existing cellular pathways that control ion gradients in a way that can be co-opted for communication (which are present in sponges) all seem like plausible things that could predispose a multicellular organism to develop nerves, but if it is a filter, it could well also be that it's really hard to develop something as complicated as nerves, and an early animal got very lucky.

I think it's important, in this case, to distinguish between simple multicellularity where identical cells form clumps (known to have evolved at least 25 times in eukaryotes, as well as in prokaryotes according to https://en.wikipedia.org/wiki/Multicellular_organism - and which has been observed evolving in the lab before your linked paper), multicellularity that has cell differentiation (evolved in 6 groups), and multicellularity that has enough behavioural complexity that we might reasonably expect the emergence of intelligence (only in animals).

I don't think simple multicellularity has ever been something that seemed like a likely candidate as a particularly strong filter (although you could argue that it's a weaker filter combined with others)