next up previous
Next: Dean Buonomano Up: No Title Previous: Naama BrennerRob de

Carlos Brody

National Autonomous University of Mexico

Cross-correlations do not imply synchrony

When the firing times of two neurons are synchronized, the cross-correlogram of their spike trains has a significant positive peak at zero. Such peaks at zero are thus usually taken as evidence of synchrony. However, strictly speaking, peaks at zero are merely evidence of lack of independence between the firings of the two neurons, and firing time synchronization is not the only way to depart from independence. I will describe two physiologically plausible ways to depart from independence, and the conditions under which the correlogram peaks they generate look misleadingly similar to spike synchronization peaks. I will go through some simulations constructed to illustrate the point, and I will also go in detail through at least one (possibly more) example from the experimental literature (``Feature-linked synchronization of thalamic relay cell firing induced by feedback from the visual cortex" by Sillito et al., Nature 369: 479-482 (1994), which I believe suffered very seriously from these confounds.

An important goal of my presentation will be to emphasize rule-of-thumb tell-tale signs of these confounds, to facilitate being alert to their possible presence.

Tony Zador
Sat Mar 27 10:58:21 PST 1999