Repeating spatiotemporal spike patterns reflect functional network states in the visual cortex
Kai Gansel & Wolf Singer
Max-Planck-Institute for Brain Research, Frankfurt am Main

Do the neuromodulatory setting and the level of excitation inside a cortical network generate distinctive states of functionality, and can these states be characterized by the emerging spatiotemporal patterns of multineuronal suprathreshold activity?

To approach these issues, we analyzed single unit spiking activity from acute slices of rat visual cortex under two pharmacological conditions, namely without and with activation of cholinergic receptors using Carbachol. Recordings were performed with a 32-channel extracellular electrode array ("Utah probe"), yielding on average about 90 isolated single units in parallel. A second electrode matrix with flat electrodes was used to apply short current pulses to transiently raise the population's spike rate. Spatiotemporal firing sequences were detected using sliding windows with lengths of 5 to 50 ms and checked for significance by common Monte Carlo methods. In this way, we were able to collect thousands of statistically significant spatiotemporal spike patterns from ten slices, repeating with different timing precisions of up to 0.5 ms. The question now was, if the mere identities of the patterns - taken as indicators of the functional state of the network - as well as their qualities like complexity, spatial dimensions and variance change systematically depending on the experimental condition.

It turned out that both pattern complexities (in terms of participating units) and the average spatial distance between pattern-forming peers did not change significantly following pharmacological treatment or electrical stimulation of the neuronal tissue. The repeats of individual patterns were also constant across pharmacological conditions but increased two-fold during electrical stimulation, indicating a decreased variance among patterns. However, most importantly, about 77% of all patterns selectively occurred during a certain experimental condition (chance level: 51 24 SD), showing an average specificity of about 90% (chance level: 78 ± 4.4 SD). About 23% of all patterns also selectively recurred and hence may be viewed as the "fingerprints" of that particular condition. Likewise, the functional connectivity maps, as derived from the patterns, display a clear dependence on the experimental condition, although unsystematic changes are present as well.

As a bottom line, changing the neuronal dynamics in a cortical network by cholinergic modulation or electrically evoked activity produces systematic changes in the network's functional connectivity, leading to the emergence of selectively recurring spatiotemporal spike patterns. Concomitant unsystematic changes, provoking non-selective or non-recurring spike sequences, may well be interpreted as the result of ongoing plasticity.