Neural information processing: Temporal features and spike train statistics

Sensory stimuli are received by sensory neurons and information about these stimuli is further transmitted throughout the brain as electrical signals. These electrical signals do not directly resemble the stimuli they represent, but instead are temporal sequences of discrete electrical impulses, known as action potentials (APs) or "spikes". This thesis aims to further the understanding of how the statistical properties of neural output is determined by the statistical properties of inputs; also it aims to understand what temporal features of the inputs are represented in AP output sequences, or "spike trains".;In the first part of this thesis, consisting of chapters 2 and 3, we study how input signals with long-range correlations impart slow correlations in output interspike intervals (ISIs). These correlations are important for information transmission at plastic synapses, which is dealt with in chapter 3. In the second part of this thesis we examine how the temporal structure of a specific class of relevant sensory stimuli affects the spike train patterning of the neurons they impinge on. Single cell recordings in weakly electric fish, presented in chapter 4 uncover a simple two-cell network responsible for transmitting a narrowband signal and high-order features (i.e. the time-varying contrast, or envelope) of the signal through parallel neural channels. In chapter 5 we examine, using experiments and theory, the single cell mechanism responsible for representing the envelope directly in the AP firing rate of neurons. In chapter 6 we extend these results to show that in parameter regimes where the firing rate cannot convey information about signal envelopes, networks of electrically coupled cells can convey this information through their precise relative spike times.