In general terms, our main interest is how neuronal circuitries of the brain support its cognitive capacities. Our goal is to provide rational, mechanistic explanations of cognitive functions at a descriptive level. In our view, the most promising area of cognitive faculties for scientific inquiry is memory, since it is a well-circumscribed term, can be studied in animals and substantial knowledge has accumulated on the molecular mechanisms of synaptic plasticity.

The theoretical framework can be summarized as follows. Bringing anatomically-distributed populations of neurons together in time is a major function provided by network oscillations.  Through their interconnectivity, GABAergic interneurons can maintain both localized and large-scale oscillations at various frequency ranges (theta, gamma, 200 Hz).  Networks of inhibitory interneurons within the forebrain impose coordinated oscillatory "contexts" for the "content" carried by networks of principal cells and provide the precise temporal structure necessary for ensembles of neurons to perform specific functions. The neocortico-hippocampal transfer of information and the modification process in neocortical circuitries by the hippocampal output take place in a temporally discontinuous manner. Acquisition of information may happen very fast during the activated state of the hippocampus associated with theta/gamma oscillations. Intrahippocampal consolidation and the hippocampal-neocortical transfer of the stored representations are protracted and may be carried by discrete quanta of cooperative 200 Hz bursts during slow wave sleep.  This general, two-stage framework of memory trace consolidation, is supported several experiments and computational models carried out by us and others. Most of our experimental investigations target one or several of the above issues.

Using large-scale recordings of single cells, we provided evidence that very slowly firing pyramidal neurons are engaged in transient, cooperative network oscillations (200 Hz) in association with hippocampal field sharp waves during sleep. In subsequent works, we found that 50,000-100,000 principal cells in the CA3-CA1-subiculum-entorhinal cortex axis are brought together with a high temporal precision during sharp wave burst events. In collaboration  we have shown that these fast oscillations are also present in the human hippocampal formation. Our intracellular experiments suggest that voltage-dependent oscillation of inhibitory interneurons is one mechanism responsible for the high precision timing of the action potentials of pyramidal cells. We hypothesized that long-term alterations in synaptic efficacy is the major physiological role of this organized network burst, since it 1) occurs within the appropriate anatomical pathway, so as to convey the outcome of hippocampal processing to neocortical networks, 2) occurs within a logical time domain (i. e., time constant of NMDA receptor), so as to be relevant to an “after the fact” consolidation process, and 3) may provide the prerequisite depolarizing force needed to produce synaptic modifications of neocortical networks.  These experiments are currently being expanded to record simultaneously from 64 to 128 sites with the second-generation of silicon probes in both hippocampus and neocortex. We are able to image extracellular current flow (spontaneous field potentials) at <1 msec time resolution in the depth of the brain in behaving rats. The field oscillations (population patterns) then are related to firing of simultaneously recorded individual neurons that carry behaviorally relevant information.

A general issue regarding the operations of cortical networks is the timing of action potentials of spatially distinct principal cells. Two competing mechanisms have been offered in the past: a) pyramidal cells are "coincidence detectors", therefore the occurrence and timing of their discharge depends on the temporal precision of presynaptic inputs in the msec range, and b) principal cells are "integrate-and-fire" elements and information is contained by the firing rate of neurons. We propose yet another explanation for the observed behavior of cortical neurons. In essence, we hypothesize that GABAergic interneuronal networks may superimpose a periodic fluctuation of the membrane potential close to, but below, threshold in principal cells. In such a framework, interneuronal networks provide precise timing of the action potentials of principal cells and the information is contained in the temporal sequence of their spike occurrences. The advancement of this hypothesis stems from two sources. In freely moving rats, we have observed zero-time synchrony among simultaneously recorded, spatially distant neurons and among field potentials recorded in far apart. The other supportive evidence derives from in vivo filled and completely reconstructed interneurons. These observations allow us to conclude that organization of GABAergic interneurons are fundamentally different from that of the principal cells. Interneurons in the forebrain are extensively connected, some of them are through gap junctions. This forebrain "interneuron supernetwork" functions as a "distributed clock" to time action potentials of principal neurons.

Another hypothesis that emerged from our experiments on interneurons is that subclasses of interneurons terminating on the dendrites of pyramidal cells may prevent the calcium influx into the dendrites without interfering with the sodium spike-generation mechanism. Given the pivotal role of calcium in synaptic plasticity, our observation suggests that the same neuronal circuitry may be used for information transmission without modification of its synapses or synapses may be altered in a use-dependent manner, and the mode of the circuit operation is controlled by the activity of these interneurons. Using simultaneous intradendritic and extracellular recordings from the same pyramidal cell in anesthetized rats, and simultaneous extracellular recordings from the soma and dendrites of the same cells in freely moving animals, we found that the active backpropagation of the sodium spike from soma to dendrites is under strong control of inhibitory interneurons. Voltage-dependent calcium influx fails to occur when the backpropagating sodium spike is attenuated, without any affect on the output function of the neuron. Calcium influx occurring during network bursts, as predicted by the general framework described above is currently under investigation using both large-scale recordings of unit activity and 2-photon laser scanning microscopy.

Research over the past decade revealed that neurons possess numerous active conductances. A contribution of our laboratory is an attempt to bridge the level between the enormous amount of knowledge accumulated from work in the brain slice preparation and the intact networks of behaving animals. We have demonstrated that some complex brain-behavior relationships whose mechanisms have been sought for at the network level can be explained by intrinsic, biophysical properties of single neurons.

These are good times for systems neuroscience since several critical levels of scientific enquiry (quantitative, single cell level anatomy, cellular biophysics, population physiology, genetics and computational neuroscience) have shown unprecedented progress and at the same time integration among the different levels has become a reality.