Spiking activity propagation in neuronal networks: reconciling different perspectives on neural coding

• Published in: Nature reviews. Neuroscience Vol. 11; no. 9; pp. 615 - 627
• Main Authors: Kumar, Arvind, Rotter, Stefan, Aertsen, Ad
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 01.09.2010; Nature Publishing Group
• Subjects: 631/378/116/1925; 631/378/116/2395; 631/378/1697; Action Potentials; Animal Genetics and Genomics; Animals; Behavioral Sciences; Biological and medical sciences; Biological Techniques; Biomedical and Life Sciences; Biomedicine; Brain; Central nervous system; Electrophysiology; Fundamental and applied biological sciences. Psychology; General aspects. Models. Methods; Humans; Hypotheses; Information processing; Neural networks; Neural Networks (Computer); Neural Pathways; Neural Pathways - physiology; Neural transmission; Neurobiology; Neurons; Neurons - physiology; Neurosciences; Physiological aspects; Propagation; review-article; Vertebrates: nervous system and sense organs; Germany; Dichotomy; Propagation; Coding; Central nervous system; Attention; Strategy; Models; Review; Neural network; Feedforward; Encephalon
• ISSN: 1471-003X; 1471-0048; 1471-0048; 1469-3178
• DOI: 10.1038/nrn2886

Key Points One of the central problems in neuroscience is the characterization and understanding of the neural code. In 1968 Perkel and Bullock defined four key functions for a candidate neural code: stimulus representation, interpretation, transformation and transmission. Although the first three have been studied extensively, surprisingly, the fourth has been largely ignored in experiments. Yet, signal transmission is a vital functions for a neural code in ensuring communication among highly specialized brain regions. Feedforward networks with convergent or divergent connections between subsequent groups of neurons have been the model system of choice in the study of spiking-activity propagation. The simple feedforward topology captures key features of the modular architecture of the brain. Moreover, from a functional perspective, certain classes of recurrent networks can be treated as feedforward networks. Theoretical studies have identified two dominant modes for propagating spiking activity in feedforward networks: the aynchronous rate mode, in which the average spike count is propagated across the sub-networks; and the synchronous event mode, in which only synchronous volleys of spikes are propagated. Various properties of individual neurons and the structure of feedforward networks can amplify even weak correlations in spiking-activity propagation. Such amplification rapidly degenerates the fidelity of an asynchronous rate code. Thus, only feedforward networks with weak shared connectivity are suitable for propagating asynchronous firing rates. Large, shared connectivity favours the propagation of a synchrony code. Structural properties of feedforward networks, in particular connection probability and synaptic strengths, have a crucial role in determining whether asynchronous firing rates or synchronous spikes are propagated. Thus, appropriate architecture of the FFN may support stable propagation of asynchronous and synchronous neural codes simultaneously. Indirect experimental evidence suggests that neural networks in vivo may indeed induce synchrony in their propagating activity. However, a direct testing of theoretical predictions is currently lacking. Controlled stimulation of appropriately selected neural networks in vivo to generate activity patterns mimicking either asynchronous or synchronous input and monitoring of their temporal evolution downstream could provide an effective paradigm for testing these predicitions. Reliable propagation of spiking activity in the brain is vital for information processing. Kumar, Rotter and Aertsen now propose that under certain conditions asynchronous and synchronous propagation of spiking activity can co-exist in a modular neuronal network, and they suggest experimental strategies to test this hypothesis. The brain is a highly modular structure. To exploit modularity, it is necessary that spiking activity can propagate from one module to another while preserving the information it carries. Therefore, reliable propagation is one of the key properties of a candidate neural code. Surprisingly, the conditions under which spiking activity can be propagated have received comparatively little attention in the experimental literature. By contrast, several computational studies in the last decade have addressed this issue. Using feedforward networks (FFNs) as a generic network model, they have identified two dynamical activity modes that support the propagation of either asynchronous (rate code) or synchronous (temporal code) spiking. Here, we review the dichotomy of asynchronous and synchronous propagation in FFNs, propose their integration into a single extended conceptual framework and suggest experimental strategies to test our hypothesis.