On the Physiological Basis of the 15−30 Hz Motor-Cortex Rhythm Jensen, O. 1,2 , Pohja, M. 1 , Goel, P. 3 , Ermentrout, B. 4 , Kopell, N. 5 , and Hari, R. 1 1 Brain Research Unit, Low Temp. Lab., Helsinki Univ. of Technology, Finland, 2 F.C. Donders Centre for Cogn. Neuroimaging, The Netherlands, 3 Dept. of Physics, Univ. of Pittsburgh, USA, 4 Dept. of Mathematics, Univ. of Pittsburgh, USA, 5 Dept. of Mathematics, Boston Univ., USA Abstract In vitro work on the rat brain and theoretical investigations have established that networks of GABAergic cou- pled interneurons play a fundamental role in creating the rhythmicity and synchrony necessary for neuronal os- cillations in the gamma (30−80 Hz) and beta (15−30 Hz) bands. To test if related mechanisms are involved in producing the human ~20 Hz cortical rhythms, we recorded the magnetoencephalogram (MEG) from 8 subjects before and after orally administered benzodiazepine (0.08 mg/kg) which is known to upregulate GABAergic transmission. The power of the 15−25 Hz rhythm increased dramatically with benzodiazepine, whereas the mean frequency decreased slightly (~1 Hz). The most prominent sources of the benzodiazepine-induced rhythm were identified in the primary sensorimotor cortex bilaterally, suggesting that the sensorimotor cortex is the primary effector site of benzodiazepines. Numerical simulations of a physiologically plausible network model, in which excitatory neurons fire at 20 Hz at every second cycle of a 40 Hz rhythm produced by inhibitory interneurons, could account for these experimental findings. Thus the human sensorimotor-cortex beta rhythm and the rat hippocampal beta rhythm could be modelled according to similar biophysical mechanisms. 1 Introduction The physiological mechanisms responsible for gener- ating spontaneous neuronal oscillatory activity have been studied extensively both in hippocampal slice preparations and theoretically. The GABAergic in- terneurons seem to be crucially involved in synchro- nizing the neuronal population and determining the network frequency of spontaneous 3080 Hz (gamma) oscillations [1,2,3]. Several studies suggest that related mechanisms are responsible for the gen- eration of spontaneous 1530 Hz (beta) oscillations [4,5,6]. How do these results relate to beta oscillations produced in the human brain? It is well known from clinical EEG recordings that benzodiazepines, which are GABA agonists, increase the beta band power [7]. Thus the GABAergic interneurons could play an im- portant role in the production of spontaneous beta os- cillations also in humans. The aim of this study is first to characterize how beta power and frequency are modulated by ben- zodiazepines and to identify the brain regions respon- sible for producing the beta oscillations sensitive to benzodiazepine. These observations will be used to constrain physiologically realistic computational models, which aim at explaining the experimental findings. 2 Methods Magnetoencephalographic (MEG) signals were re- corded from 8 healthy subjects using a 306-channel Vectorview neuromagnetometer (Neuromag Ltd.; 204 planar gradiometers and 102 magnetometers). The subjects, seated in a relaxed position under the MEG helmet, were instructed to keep their eyes closed while ongoing MEG signals were recorded for 3 min. Benzodiazepine (Diapam) was then administrated orally to the subjects (~80 µg/kg, i.e. 4−7.5 mg). Fol- lowing a 1-hour break, the MEG measurement was repeated. Power spectra estimates of the neuromag- netic signals were calculated using Welch method. To identify the sources of the beta activity, the minimum current estimates were calculated in the frequency domain [8]. This allowed us to locate the sources of the 1530 Hz oscillations and to superimpose them onto the subjects own MRIs. The network model was constructed of 20 exciatory and 8 inhibitory Hodgkin-Huxley type model neurons with all-to-all connection (see Fig.3A) The synaptic connections were modelled using realis- tic kinetics for AMPA and GABA receptors. During the numerical integration, noise was added to the ex- citatory neurons. The total field produced by the exci- tatory neurons was calculated as the sum of the exci- tatory currents lowpass filtered at 20 Hz.