Nobody really LIKES doing experiments. Even in vitro work is slow, tedious, and expensive. Anything involving living organisms is orders of magnitude more aggrevating.

If we could simulate things instead, we absolutely would: cheaper, faster, less suffering, etc. But we can't.

The canonical example is a nematode called C. elegans. It's small: 959 cells, 302 of which are neurons. It has been studied thoroughly. We have a complete "wiring diagram" of its entire nervous system, we know the origin and fate of every single cell, it has been sequenced and probed and imaged with every technology under the sun. If there is an organism we should be able to simulate, C. elegans is it. But there aren't any good models of it, in the sense that you start a simulation and it behaves like a "real" worm. There are a lot of factors that we know we don't understand/can't model well: the passive diffusion of neuromodulators is probably important but is a nightmare to simulate. There are undoubtedly other factors we don't even know matter yet too.

The worm is, of course, really simple. A human brain has about 300M times more neurons (and combinatorially-many more connections between them).