Correction: In the article, “A Model of High-Frequency Ripples in the Hippocampus Based on Synaptic Coupling Plus Axon-Axon Gap Junctions between Pyramidal Neurons,” by R. D. Traub and A. Bibbig, which appeared on pages 2086–2093 of the March 15, 2000 issue, two numbers were reported incorrectly. At the end of the first paragraph of Results and in the legend of Figure 2C, the mean pyramidal cell signal leads the mean interneuronal signal by 6.0 msec, not the 1.3 msec stated. In the legend of Figure 3, the local mean interneuron potential lags this axonal signal by 5.6 msec, not the 1.5 msec stated. The authors regret these errors.
In interpreting these numbers, we note the following. Inin vivo experimental studies (e.g., Csicsvári et al., 1999) examining the temporal relations of pyramidal cells and interneurons during ripples, cell pairs are recorded that are usually within 300 μm of each other, a distance for which axonal conduction delays are small; in addition, to obtain sufficient unit firings to collect data for statistical analysis, large numbers (>100) of separate ripple events were recorded. In the model, data are pooled from the activity of all of the neurons in our distributed array, during the course of a single ripple. This introduces axon conduction delays, which average ∼2 msec but can last up to 3.8 msec. Thus, the average axonal signal in Figure 3 leads the AMPA signal by 3 msec, of which 2 msec is average conduction time, and 1 msec is the time to peak of a unitary EPSC [note that this 1 msec may also be on the high side (Geiger et al., 1997)]. The AMPA signal leads the voltage of the particular interneuron by 2.1 msec. Without the conduction delays, the time from the axonal signal to firing of this interneuron would hence be 3.1 msec, less if a faster EPSC time course on interneurons were used.