The aim of this study was to investigate the dynamics of the horizontal functional connectivity within the visual cortex during spontaneous activity or during visual stimulation. Two arrays of 16 electrodes were inserted in the visual cortex of a rat. From these electrodes a depth profile was obtained of the local spiking activity. The cross-correlations between all electrodes were estimated. Three types of cross-correlation peaks were identified and classified as; 'thin peaks', 'fast waves' and 'slow waves'. Partialization was applied, a mathematical method to reduce the amount of common input in correlations, and its effect on the three types of correlation peaks was studied. Slow waves were found to be the most vulnerable to partialization and thin peaks the least. From these observations it was concluded that the three types of peaks represent synchronous neuronal assemblies of different magnitude; slow waves large, fast waves intermediate and thin peaks assemblies composed of small numbers of neurons. Large changes were observed in the types of cross-correlations and their spatial distribution within the set of interarray combinations of electrodes. These changes were spontaneous, and could not be related to the visual stimulation. Two states were identified; the 'thin-peak' state and the 'slow-wave' state. The 'thin-peak' state is interpreted as occurring at a light level of anaesthesia and is characterized by the presence of thin peaks in all combinations of electrodes. Thin peaks with the largest strength were found in the upper interarray electrode combinations. The 'slow-wave' state is interpreted as occurring at a deep level of anaesthesia and is characterized by the presence of exclusively slow waves, which were limited mostly to the middle and lower interarray combinations of electrodes. Activation of the cortex is thus associated with the appearance of synchrony between small groups of neurons (thin peaks) which, in contrast to the slow-wave state, include the upper layers of the cortex.