Oscillatory Activity Reentrant Signaling among Simulated Neuronal Groups Leads to Coherency in their
O. Sporns,J. Gally,G. Reeke,G. Edelman
Abstract:Recent experiments have revealed tightly synchronized oscillatory discharges in local assemblies of cortical neurons as well as phase coherency of oscillations at distant cortical sites. These findings are consistent with the theory ofneuronal group selection, a population theory ofbrain function that is based on the properties of local groups of neurons. A set of computer simulations shows that cooperative interactions within and among neuronal groups can generate the observed phenomena. In the simulations, oscillations within neuronal groups are generated through local excitatory and inhibitory interactions. Different groups in general oscillate in an uncorrelated fashion. Coherency of the oscillatory activity of different neuronal groups depends crucially on reciprocal reentrant signaling and can reflect the spatial continuity of a stimulus. Separated or discontinuous features of a given stimulus can be transiently associated in a temporally coherent pattern through reentrant signaling between groups in networks responding to different aspects of that stimulus. A simulation of reentrant activity between arrays of neuronal groups selective for oriented lines and pattern motion displays cross-correlations between groups that are responsive to different parts of a stimulus contour if these parts move together. Such coherency among neuronal groups might be used in the discrimination of a stimulus from other stationary or differentially moving elements in a visual scene. The theory of neuronal group selection (1, 2) proposes that the basic functional unit of cortical processes is the neuronal group, a local collective of strongly interconnected neurons. The borders between groups as well as the receptive field properties of their constituent cells are a function of local neuroanatomy but can change dynamically under varying input conditions (3, 4). Groups compete with each other for the incorporation of additional cells, some only during a sensitive or critical period, others throughout the adult life of the organism. The cooperative effects resulting from local excitatory interactions between neurons in a group lead to a tendency for temporally correlated, synchronous firing. Indeed, groups are stabilized against competitive influences by such firing (4). Neurons belonging to different groups in general show asynchronous and uncorrelated firing. Signaling between neuronal groups occurs via excitatory connections that link cortical areas, usually in a reciprocal fashion (5, 6). According to the theory of neuronal group selection, selective dynamical links are formed between distant neuronal groups via reciprocal connections in a process called "reentry." Reentrant signaling establishes correlations between cortical maps, within or between different levels of the nervous system. It is more general than feedback in both connectivity and function in that it is not simply used for error correction or cybernetic control in the classical sense. Instead, reentrant interactions among multiple neuronal maps can give rise to new properties. Indeed, a recent computer model simulating reentry in visual cortex demonstrates that such interactions might serve as bases for complex visual illusions (7) achieving figural synthesis by recursion while competitively resolving conflicts between segregated cortical areas. This reentrant cortical integration model emphasized the reentrant circuits themselves and for economy represented groups as single units with fixed properties rather than as actual populations. It is the aim of the present study to investigate further the mechanism of reentrant cortical integration by making explicit use of the population properties of neuronal groups. At the same time, we wish to relate our analysis to results of recent experiments on visual cortex (8-10) indicative of group behavior. Gray and Singer (8, 9) have observed oscillatory neural activity within orientation columns in cat visual cortex. Appropriate stimulation produced rhythmic, temporally correlated discharges in areas 17 and 18 in adult cats and kittens, awake or under varying conditions of anesthesia. In many cases, it was found that multiunit activity as well as local field potentials oscillate at a frequency ofabout 40 Hz when a light bar of optimal orientation, velocity, and direction of movement is passed through the receptive field of one of the recorded neurons. Oscillatory activity decreased in the absence of a stimulus or during presentation of a light bar at other than the optimal orientation. Recordings from the lateral geniculate nucleus (LGN) showed no evidence of oscillations within the frequency range of20-70 Hz. Gray and Singer concluded that the observed cortical oscillations are generated by local excitation and recurrent inhibition within the cortex. As previously predicted (1), these results show "that groups of adjacent cortical neurons, when activated appropriately, engage in cooperative interactions" (ref. 8, p. 1702). Cross-correlation of multiunit recordings from spatially separate sites in area 17 revealed that oscillatory responses in orientation columns with similar specificity but with nonoverlapping receptive fields are synchronized if a single long light bar is moved across both receptive fields (9). This synchronization of oscillatory activity in remote columns may be mediated by intracortical, cross-columnar connections or by long projections from other cortical areas. An independent study (10) confirmed intracolumnar correlations and correlations between neighboring hypercolumns; in addition, synchronization of oscillatory activity was observed between different cortical areas. During the presentation of an appropriately oriented light bar, coherent and oscillatory local field potentials appeared in columns both in area 17 and area 18. These reports (9, 10) not only prompt the idea that the observed temporal correlations between spatially separate cortical sites in response to linked stimulus features may be of importance in the transient association of such features Abbreviations: LGN, lateral geniculate nucleus; OR, orientation network; MO, motion network; AAF, average activity function. 7265 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Proc. Natl. Acad. Sci. USA 86 (1989) but also provide evidence for the existence of neuronal groups and are suggestive of the role of reentry in correlating the behavior of different groups. To demonstrate the consistency of these observations with the theoretical concept of reentry, we present here a simplified computer model of groups in two interconnected visual areas. Models that generate oscillatory phenomena have previously been proposed (11, 12) but are not based on local populations of neurons and do not deal with their interactions and coherent behavior. The model described here lends support to the theory of neuronal group selection by generating phenomena similar to those observed (8-10) without the addition of special assumptions. DESCRIPTION OF THE MODEL The model is implemented using the Cortical Network Simulator program (for details see ref. 13), and run on an IBM 3090 at the Cornell National Supercomputer Facility. Fullscale simulations of 25,600 cells and over 3.5 million connections required up to 5 hr for 9,000 iterations (cycles). The model consists of two separate visual areas (Fig. 1): "OR" which contains 32 orientationand direction-selective neuronal groups consisting of 200 excitatory and 120 inhibitory