Need help in understanding role of activation functions.

This question is being asked here after i had a discord with a professor of mine. As far as i understood it, activation functions are used to make neural networks non linear but he believes activates and de activates individual nodes of a NN.

I cant get my head around how it activates a neuron. Without using a threshold function. Taking the case of a neuron with a sigmoid activation the output will never be 0, though miniscule in case …

neuralnetworks role understanding