next up previous
Next:

Tsodyks: Duality Up: No Title Previous: SmirnakisWarland, Berry,

Quantifying information

Steve Strong

NEC

We show how to quantify the information a given neuron's spike train provides about an experimenter controlled stimulus. Our method is free from any assumptions about which features of the spike train or input waveform are most important and is based on systematic correction for finite data effects and correlation effects as well as on the use of bounds that are relatively insensitive to sampling problems. We apply this approach to the analysis of experiments on a motion-sensitive neuron in the fly visual system.



Tony Zador
Tue Oct 22 16:34:57 PDT 1996