Life quality and life span are seriously compromised by numerous brain diseases. Currently, rehabilitation is based largely on behavioral manipulations directed at activation of brain 'self-repair' processes. Future advances are expected to include biological manipulations such as genetic manipulation and stem cell-based therapy that promote neuronal recovery. Another feasible strategy is replacement of defined neuronal microcircuits by synthetic analogs. Decades of iterations between scientific inquiry, technological development and escalating clinical demands have advanced the techniques of monitoring and stimulation of localized brain sites. To date, deep brain stimulation successfully ameliorates a range of Parkinson and OCD symptoms and deep brain recording is used to detect the source of epileptic fits. The hope is that these two techniques can be interfaced by a real-time processor and used as a closed loop system with the brain. Our objective was to test the feasibility of the closed-loop hybrid methodology for rehabilitation of brain functions by replacement of a damaged brain microcircuit. Specifically, we reasoned that a function lost after localized brain damage can be rehabilitated by a biomimetic device that analyzes on-line the inputs to the damaged microcircuit and sends the outcome of this analysis to the output of the damaged microcircuit. The cerebellum is a good choice for testing the feasibility of the closed-loop rehabilitation methodology. In particular, the cerebellum is well conserved across the mammalian kingdom and is comprised of recurrent microcircuits sharing similar anatomical and physiological architecture. Cortical Purkinje and deep nuclei neurons are sites of convergence of sensory signals originating at two brainstem nuclei; the pontine nucleus (PN) conveys telesensory signals and the inferior olive (IO) conveys somatosensory signals. Associative exposure to the above signals initiates cerebellar learning, expressed as long term depression (LTD) of the synapse conveying the telesensory signals to the Purkinje and as emergence of a neuronal model of the learned response at the cerebellar deep nuclei. In the present study we replaced in a rat the cerebellar microcircuit essential for acquisition of eyeblink response by a biomimetic cerebellar model that received its sensory inputs from the PN and IO precerebellar nuclei and which send its output to the brainstem motor nucleus. Results demonstrated synaptic plasticity in the model and behavior which was compatible to the dynamics of acquisition and habituation in a normal rat. We now look to advance the technologies necessary to support years long hybrids, to advance the understanding of the sensory coding that is required for the detection of the sensory events from electrophysiological records. Finally, we will test whether animals will adopt the hybrid technology or regress to alternative behavioral strategies in ecological conditions. *This study was supported by the European Community’s 7th Framework Program (216809), the Converging Technologies – ISF Grant (1709/07), and by The Center for Complexity Science (GR2004-065). PIs were del Giudice P., Guger C., Marcus M., Mintz M., Shacham Y., Silmon A., Vershure P., Yaron H.M.