Nerve cells pay attention to their neighbors
In an orchestra, the entries of the individual instruments must be very precisely coordinated. Also in the brain, the activity of billions of nerve cells (neurons) is “correlated”, as it is called in neuroscience. Only in this way, the brain can accomplish such amazing achievements as reading or listening to music. Despite the central importance of neural correlations, it has not been clarified yet how and under which conditions they occur. Scientists around Fred Wolf from the Bernstein Center for Computational Neuroscience in Göttingen and the Max Planck Institute for Dynamics and Self-Organization have developed a mathematical formula to precisely predict how and when neurons synchronize.
Each neuron in the cerebral cortex receives information from ca. 30 000 other cortical neurons and – in response to this – sends individual neural impulses. At least theoretically, a simple relation between input signals and neural response is conceivable: If, for example, two neurons share 1/10 of the input signals, 1/10 of their response signals would be the same as well. But neurons do not calculate that simply. Neurons receive a large number of electrical input signals which result in random membrane voltage fluctuations. When the membrane voltage reaches a threshold value the neuron sends out a signal. The Göttingen scientists investigated how the statistics of input and output signals are actually related, taking into account this mode of neuronal operation. They managed to summarize the neural conversion from input signals into output signals in a relatively simple mathematical formula. “The conversion of voltages into digital signals by means of microprocessors in computers works according to a similar principle,” says Tatjana Tchumatchenko, doctoral student at the Göttingen Graduate School for Neurosciences and Molecular Biosciences (GGNB), University of Göttingen, who carried out the mathematical analysis. “Also in computer technology, the correlation between digital signals plays a vital role. For example, if different parallel transistor elements receive similar input signals it is important to predict the stability of the output signals.”
As the researchers were able to show, the correlation of the response signals of two nerve cells depends not only on how similar the respective input signals are, but also on how active the cells are. If neurons send out many signals in rapid succession (their activity, the so-called firing rate, is high) also the response signals are more strongly correlated. However, this only applies if the neurons share just a fraction of their input signals. The rules change drastically if neurons are largely stimulated by common input signals and correspondingly produce similar response signals. In this case, the firing rate is no longer important. The scientists were able to directly experimentally confirm these theoretical predictions by stimulating cells with computer-simulated brain currents and measuring their response signals.
Moreover, this novel theoretical concept introduced by the Göttingen scientists offers a glimpse at how the cognitive information processing in the brain might work. For a long time, neuroscientists have been discussing the question of how the brain encodes information in the electrical activity of neural signals. In some cases, the firing rate seems to be crucial, in other cases the exact timing of a neural impulse in relation to other signals. In their work, the Göttingen scientists and their colleagues now demonstrated how closely these two concepts of neural coding are connected and which theoretical description can capture the sensory processing. For example, different neurons in the visual cortex specialize in certain aspects of image processing: they react to color, brightness, orientation or direction of motion. There is evidence that cells that encode the same object synchronize their signals, such that pieces of information that belong together are transferred together.
Tatjana Tchumatchenko, Aleksey Malyshev, Theo Geisel, Maxim Volgushev and Fred Wolf.
Correlations and Synchrony in Threshold Neuron Models.
Physical Review Letters, Vol.104, No.5 (5. Februar 2010),