It can decide. It cannot learn.

The neuron adds its inputs and checks the threshold. Give it the same inputs twice and it gives the same answer twice.

It does not know whether the answer helped. It cannot notice a mistake. It cannot change its own settings.

So this is a decision-making machine, but not yet a learning machine.

Rosenblatt's question was simple: what could change after a wrong answer?