Biophysical basis for Differentiate-and-Burst neurons
Brief bursts of high-frequency action potentials are commonly observed in neuronal recordings and are thought to represent a special neural code. Here I discuss detection of specific temporal features of stimuli by spike bursts. Using a computational model we show that as a consequence of the biophysical mechanisms that lead to burst generation, bursts act as a differentiator and signal the slope of the input current rather than its amplitude. Furthermore, the number of spikes per burst signals the magnitude of the slope in a graded manner, establishing a burst duration code. Because the firing rate during bursts is so high, variations in spike number can efficiently transmit information in brief periods. I illustrate this general result by the example of upstroke detection by bursting cells in the electric fish. Thus some neurons may differentiate and burst and thereby provide a useful computation quite distinct from neurons that integrate and fire.