It has been argued that synchronous oscillations might not be important for neural coding.47 One problem with encoding information in synchronous oscillations is that it can be difficult to distinguish true synchrony from chance correlations. A high level of stimulus-induced activity could create false synchrony simply due to the fact that the high number of spikes produces more coincident action potentials. Neurons would have to distinguish background synchrony from true synchrony for synchronous oscillations to encode useful information.
A property of neurons that can help to distinguish background from true synchrony is adaptation in the spiking threshold.48,49 As the level of activity rises, the mean level of depolarization increases. The increase in threshold for spiking emphasizes rapid depolarization and increases the sensitivity to coincident inputs. This adaptive threshold mechanism narrows the window for coincidence and filters out temporally unpatterned input. This would decrease background synchrony and emphasize synchronous oscillations. Moreover, neurons can be sensitive to periodic input due to intrinsic or synaptic resonances at the appropriate frequency; in this case, aperiodic chance synchrony might be more easily distinguished from meaningful synchrony.
For synchronous oscillations to play a role in information processing, neurons must have mechanisms for discriminating between synchronous and asynchronous inputs. There have been several experiments that indicate that synchronously arriving postsynaptic potentials (PSPs) summate more effectively. Kenyon cells (KC) in the locust mushroom body have an intrinsic voltage-dependent mechanism for detecting coincident excitatory postsynaptic potentials (EPSPs).35 When two different inputs to a cell arrived less than 12 ms from each other, a spike was evoked, but when the interval between inputs was greater than 12 ms or when the active conductances other than Na+ and K+ were blocked, no spike was produced.35 In the hippocampal-slice preparation, correlated pre- and postsynaptic spikes that produced either long-term potentiation (LTP) or long-term depression (LTD) also changed how the postsynaptic cell integrated its input. Prepost pairings that produced LTP enhanced summation, while those that caused LTD decreased summation.50
Another issue is the theoretical problem with using synchrony to represent information across a group of neurons51-53 as a population code. Averaging over a group of neurons can overcome some of the variability in individual neuronal firing. However, correlations between neurons could limit the advantage gained by averaging.51-53 An illustrative analogy is a group of people independently estimating the weight of something. While the estimates may vary over a wide range, their mean will get closer to the true value as the number of individual estimates increases. However, when the estimates are not independent, for example in the stock market, where there are long-range correlations as everyone takes into account what everyone else thinks, there is a limit to the value of increasing the number of estimates. Synchrony in the responses would therefore limit the accuracy of the information encoded by a population of neurons in their average activity and would be a problem that the brain would need to overcome. However, information is not necessarily lost when the correlations themselves encode information.54 Indeed, information lost due to correlations can be regained when the synchronous oscillations are stimulus dependent.55
Was this article helpful?