Personally, I've been studying ML for the past few months, especially neuroscience inspired architectures, and I stumbled across https://en.wikipedia.org/wiki/Modular_neural_network, which seem to be really useful for complex reasoning, though I can't find too much literature on the subject.
Do you think they could be useful?Some relevant reading https://vtechworks.lib.vt.edu/bitstream/handle/10919/27998/etd.pdf?sequence=1&isAllowed=y