Open Access

The physiology of our most complex organ, the brain, is still not comprehensively understood. The brain basically serves the processing, storing and binding of external and internal information, and thereby generates amazing phenomena like the understanding of oneself as an individual entitiy. How exactly information is encoded and represented, how individual neurons or networks of neurons actually interact, is a gigantic puzzle, whose pieces were collected since many decades. Subject of scientific discussions are the basic spatiotemporal structures of neuronal representations. Suggestions and observations reach hereby from simple rate coding of individual neurons to synchronous activity of larger ensembles. To approach answers to these questions, our working group has used a combination of different recording techniques that allowed for the comparison of neuronal interactions on different spatial scales. We focused on prefrontal neuronal interactions during visual short-term memory. Herefore two rhesus monkeys had been trained to perform a visual short-term memory task. We measured and recorded their neuronal activity by means of a microelectrode matrix that could be inserted into the cortex via a closable chamber, which had been previously implanted above prefrontal cortex. The acquired signal was separated into two components: a high-frequency component, that represents the spiking output activity of few neurons in the vicinity of each electrode tip (multi-unit activity), and a low-frequency component, that results from dendritic input activity of larger neuronal assemblies (local field potential). From one of the experimental animals we also recorded mass signals of even larger neuronal populations by means of small silverball electrodes, that had been implated into the skull above prefrontal cortex (skull EEG) in the context of a pilot project. In the first subproject, we analyzed the selectivity of output signals with respect to the memorized stimulus and task performance. We compared selectivities of local recording sites (multi-unit activity) with the selectivities of patterns created by the combined activity of all recording sites, thus representing the activity of large and distributed ensembles. Local neuronal activity correlated with the course of the visual short-term memory task, but was not highly discriminative with respect to different visual stimuli. We could show that the population activity was significantly more specific. Concerning task performance, we obtained the same result, albeit less pronounced. Further analyses revealed that the patterns of distributed ensemble activity were only partly based on realtime coordination of neuronal activity, and in addition, did not remain stable across the time course of the short-term memory task. In the second subproject, we focused on the oscillatory behavior of the local field potential. After a time-frequency analysis, we studied different frequency bands concerning stimulus selectivity and task performance of the monkey. We hereby found significant modulations of oscillations in the beta- and gamma-frequency range, that correlated with different periods of the task. Especially for oscillations in beta- and low-gamma-range, we observed phase-locking of oscillations between different recording sites, which could play an important role as internal clock to coordinate spatially separate activity. Local high-gamma oscillations themselves seemed to be important for the maintenance of information. These results could be partly confirmed by mass signals of EEG. In sum, our results support the hypothesis that information is represented in the brain by means of concerted activity of spatially distributed neuronal ensembles. This activity again appears to be coordinated by oscillatory activity in beta- and low-gamma-frequency ranges. A deeper understanding of central nervous information processing could contribute to better treatment of diseases like Parkinson’s, Alzheimer’s as well as epilepsy, and neuropsychiatric disorders like schizophrenia.