I was speaking against reductionism but while we are there, as your examples alude to, counting neurons to determine sentience is like counting GHz to determine computational power.
It's really not. Frequency is nearly meaningless as you can have as many computational elements as you like operating at a given frequency meaning it tells you nearly nothing. It's like counting transistors which actually DOES work even if imperfectly.
But neurons are like compitational elements. There are different sorts with different power. In fact some human neurons have features that no other neurons have.
