
A wide spectrum of noise effects was observed in NL coincidence detector neurons and neocortical neurons. The left panels are f–I relationships at increasing noise levels (crosses, circles, diamonds, squares, and triangles). The right panels are contour plots of the firing rate versus the mean current (horizontal axis) and noise amplitude (vertical axis). Hot colors indicate the high firing rate. Numbers are the firing rate (hertz). A, An NL neuron absolutely requires stimulus fluctuation to fire, and its gain increases with noise amplitude. B, A pyramidal neuron (layer 2/3) with a large sAHP fires continuously during DC stimuli, but its gain is increased by noise. C, A pyramidal neuron with a small sAHP (layer 5) shows little gain change by noise. D, An FS interneuron exhibits gain reduction by noise. E, An ideal differentiator (with rectified output) shows an increase in firing rate with noise amplitude. Similar behavior occurs primarily at high mean current in gain-increasing neurons and at low mean current in gain-reducing cells. F, An ideal temporal integrator fires at an average rate that is unaffected by input noise. The pyramidal neuron with a small sAHP is the cell closest to this ideal.




