Signal processing against the background noise in the brain

The brain is continuously active. It doesn’t matter whether we are awake or asleep, or whether we are thinking or relaxing, the nerve cells incessantly send signals. Scientists from the Bernstein Center for Computational Neuroscience (BCCN) at the University of Freiburg examined in a computer model how sensory stimuli or other information can be reliably processed and passed on in the face of such high background activity. Their work provides information on the specific form of neural information encoding which facilitates optimal information transfer.

When we see, hear or smell, the brain processes the received information step by step in progressively higher processing levels. Neurons at each level pass on signals to the next higher level in the form of electric impulses. The neural connections that form the basis of such a "feed forward system" have already been studied in many ways. Usually, however, it has not been considered that this “feed forward system” is embedded within the complex neural architecture of the brain, whose background activity provides feedback onto and, in turn, is influenced by this system. The question as to how signals can be passed on reliably in such a system has now been studied by Arvind Kumar, Stefan Rotter and Ad Aertsen from the BCCN Freiburg. Using an elaborate computer model, the scientists simulated the function of a network of 50,000 neurons as realistically as possible.

The transmission of activity in a neuronal network model can be compared to the propagation of individual waves in the sea. In this metaphor, background activity in the cortex would correspond to the two-dimensional surface of the sea, fluctuating irregularly and incessantly. The propagation of individual waves depends not only on their individual size, but also on the roughness of the sea. When the sea is rough, it is unlikely that the waves will get very far. By contrast, in a relatively calm sea, waves could travel long distances and interact with other travelling waves. Image: With kind permission of Prof. William M. Drennan, Rosenstiel School of Marine and Atmospheric Science, University of Miami, Florida, USA.

Each item of sensory information is translated in the brain into electric impulses in neurons. The principles determining the way in which information is encoded in this case are not yet known in detail and differ from case to case. A sensory stimulus for example can increase the impulse rate in certain neurons. The stronger the stimulus, the more impulses the neuron sends per unit time. This is referred to as a "rate code". However, a sensory stimulus can also result in several neurons sending out their signals simultaneously, such that synchronized "impulse packets" are transported through the Feed Forward System. Also the background activity of the brain is not always the same: depending on the mental state, cells in the brain may send their signals more or less regularly and/or synchronously. In their model the scientists now studied how these different forms of information transfer and background activity mutually influence each other.

As Kumar and his colleagues showed, not every form of information transfer is possible with every kind of background activity. A strongly synchronized neuronal background activity makes any targeted signal forwarding almost impossible. By contrast, an asynchronous background activity allows a reliable processing of sensory information and may even contribute constructively to a stable transmission of the signal. The researchers also showed that impulse packets of synchronized neuronal activity could be passed on far more reliably than increased impulse rates could be. Thus, their work has provided important information as to how sensory information must be encoded in order to be processed effectively in the brain.