Item

Abstract

Given the connectivity of its inputs, the neocortex is the largest part of the brain which is mainly self referencing. This bears several consequences for the organization of neuronal activity. One relatively simple consequence is that ongoing activity is best predicted by correlating it to other cortical activity. In contrast, stimulus-related activity which is mediated by subcortical inputs may only contribute a small fraction to the total activity measured. Such a system design can be considered an advantage, because high degrees of freedom for the dynamical evolvement of brain activity require relative independence from external signals. As the output of cortical neurons is highly sensitive to the timing of the many inputs, relative timing of neuronal activity e.g. determined by the phase relation of oscillatory processes may play an important role for integrating and propagating activity through the cortex. By analyzing field potentials recorded simultaneously in multiple cortical areas during a visuomotor task that requires multiple subsequent steps of sensorimotor integration, we found that oscillations occur in many of the recorded sites and persist almost invariably throughout the task, often beginning before the monkey actually started the actual visuomotor transformation. In contrast, signal correlation across sites, as revealed by time resolved coherence or cross-correlation, increased only during particular epochs during which precise coordination between visual cue and motor output is needed. In contrast, during much less transient behavior like short term memory of static objects, in which it is assumed that highly persistent signals like sustained neuronal firing and/or oscillations carry information across the memory delay, it is not straight forward to find signals which really bridge the gap and therefore provide a potential mechanism for memory maintenance, even in prefrontal cortex. We therefore devised a bootstrapping method with which we analyzed differential effects of behavioral performance or the memorized stimuli on the occurrence of oscillations. With these methods it is more feasible to track continuing oscillatory signals, even if their frequency changes over time. We found that in the high gamma-frequency (60-95 Hz) band there is more or less continuously activity which bridges the memory delay and from which several times during the delay oscillatory processes emerge which rapidly reduced their frequency. In our study, performance and stimulus related oscillatory processes reduced their frequency into the beta-frequency (15-30 Hz) range at the time when the monkey had to compare test stimuli and the content of short-term memory in order to generate a behavioral response. There is growing evidence that cortical oscillations occur in rather continuous form and adapt more their frequency or coupling to the dynamics of task-related processes. Future work will have to show how the dynamical changes of oscillations can modulate the spiking output of neuronal populations which determines signal transmission and ultimately behavior.