Information transmission efficiency in neuronal communication systems

The nature and efficiency of brain transmission processes, its high reliability and efficiency is one of the most elusive area of contemporary science [1]. We study information transmission efficiency by considering a neuronal communication as a Shannon-type channel. Thus, using high quality entropy estimators, we evaluate the mutual information between input and output signals. We assume model of neuron proposed by Levy and Baxter [2], which incorporates all essential qualitative mechanisms participating in neural transmission process. We analyze how the synaptic failure, activation threshold and characteristics of the input source affect the efficiency. Two types of network architectures are considered. We start by a single-layer feedforward network and next we study brain-like networks which contains components such as excitatory and inhibitory neurons or long-range connections. It turned out that, especially for lower activation thresholds, significant synaptic noise can lead even to twofold [Figure 1] increase of the transmission efficiency [3]. Moreover, the more amplifying the amplitude fluctuation is, the more positive is the role of synaptic noise [4]. Our research also shows that all brain-like network components, in broad range of conditions, significantly improve the information-energetic efficiency. It turned out that inhibitory neurons can improve the information-energetic transmission efficiency by 50 percent, while long-range connections can improve the efficiency even by 70 percent. The knowledge of the effects of the long-range connections could be particulary useful when we consider possible reconstruction or support of them applying biomaterials [5, 6]. We also showed that the most effective is the network with the smallest size: we found that two times increase of the size can cause even three times decrease of the information-energetic efficiency [7].

Figure 1

Mutual information dependency on synaptic success, s , in single-layer neural network. Maximal mutual information values (dotted line) and these achieved at s = 1 (solid). Size of a given dot is proportional to 1−s, indicating the bigger the dot, the corresponding mutual information value is achieved at lower s [3].