Neuronal coherence during selective attentional processing and sensory–motor integration

Abstract

Groups of neurons synchronize their activities during a variety of conditions, but whether this synchronization is functionally relevant has remained a matter of debate. Here, we survey recent findings showing that synchronization is dynamically modulated during cognitive processes. Based on this evidence, synchronization appears to reflect a general mechanism that renders interactions among selective subsets of neurons effective. We show that neuronal synchronization predicts which sensory input is processed and how efficient it is transmitted to postsynaptic target neurons during sensory–motor integration. Four lines of evidence are presented supporting the hypothesis that rhythmic neuronal synchronization, also called neuronal coherence, underlies effective and selective neuronal communication. (1) Findings from intracellular recordings strongly suggest that postsynaptic neurons are particularly sensitive to synaptic input that is synchronized in the gamma-frequency (30–90 Hz) range. (2) Neurophysiological studies in awake animals revealed enhanced rhythmic synchronization among neurons encoding task-relevant information. (3) The trial-by-trial variation in the precision of neuronal synchronization predicts part of the trial-by-trial variation in the speed of visuo-motor integration. (4) The planning and selection of specific movements can be predicted by the strength of coherent oscillations among local neuronal groups in frontal and parietal cortex.

Thus, neuronal coherence appears as a neuronal substrate of an effective neuronal communication structure that dynamically links neurons into functional groups processing task-relevant information and selecting appropriate actions during attention and effective sensory–motor integration.

Introduction

Oscillatory synchronization is a prevalent property encountered among groups of neurons throughout the mammalian brain (Buzsaki and Draguhn, 2004, Kruse and Julicher, 2005, Lachaux et al., 2003, Laurent, 2002, Steriade, 1999, Usrey, 2002, Usrey and Reid, 1999, Varela et al., 2001). Despite its widespread occurrence, only recent experimental evidence succeeded to show that synchronized oscillatory activity (or: coherence) among neuronal groups within and across cortical areas could be functionally relevant and support the dynamics of cognitive processes in a variety of tasks (Engel et al., 2001, Varela et al., 2001). The emerging view from these findings is that neuronal coherence subserves the selective and effective transmission of information among neuronal groups during the integration of sensory information to ultimately trigger adaptive motor performance (Fries, 2005, Laughlin and Sejnowski, 2003, Salinas and Sejnowski, 2001, Sejnowski and Paulsen, 2006). In this review, we will survey recent experimental results suggesting such a central role of neuronal coherence for effective neuronal communication.

Investigating the nature of effective neuronal communication is central to an understanding of the dynamics of cognitive processes. Neurons communicate via the transmission of action potentials along anatomical connections. During cognitive processing, only a selected subset of these connections become effective and convey task-relevant information. Moreover, varying task contexts require a flexible routing of information among varying groups of neurons in sensory and motor cortex. This is particularly evident during top-down attentional control: While the visual system is typically stimulated by a multitude of different stimuli, voluntary attention allows us to restrict processing resources to only one stimulus in the scene and to select the appropriate behavioral action. At the neuronal level, this involves the selective communication between those groups of sensory neurons processing the attended stimulus and ultimately those groups in motor cortex instructing the required behavioral response. At the same time, the communication between “unattended” sensory neurons and motor cortical neurons is prevented.

This example demonstrates the necessity of a dynamic mechanism that imposes an effective neuronal communication structure on top of the anatomical infrastructure as a function of the cognitive processing demands. In this review, we propose that coherent neuronal oscillations (i.e. phase synchronized oscillatory activity, from now on: “neuronal coherence”) between local groups of neurons is the basic ingredient to make neuronal communication effective and selective. In the following, we will first outline the hypothesis of “neuronal communication through neuronal coherence” (Fries, 2005) and survey its physiological evidence at a mechanistic level. We will then review studies demonstrating that neuronal coherence is modulated during cognitive processes in a variety of tasks. This evidence shows that within sensory cortex, selective attention enhances oscillatory activity among neurons processing attended sensory signals and reduces coherent activity among neurons processing distracting information. Additionally, for a fully attended sensory stimulus, the strength of neuronal coherence within a local group of visual cortical neurons predicts the processing speed, or efficiency, of sensory changes in that stimulus. Within motor-related parietal and frontal cortical areas, coherent neuronal coupling is especially enhanced during the planning of movements and can selectively predict which movement type and direction of movement will be selected.