Low dimensional attractors in discharges of sensory neurons of the rat spinal dorsal horn are maintained by supraspinal descending systems.

Background activity was recorded from sensory neurons in laminae I-V of the lumbar spinal dorsal horn of the rat prior to and during cold block spinalization at the cervical cord. To graphically and quantitatively describe the complexity of the discharge patterns, phase space portraits were plotted and the correlation dimension D2 was calculated by using the Grassberger-Procaccia algorithm adopted for point processes, i.e. for series of interspike intervals. The algorithm is validated both for the Baker transformation, which is a simple point process, and for the Lorenz model, whereby a transformation from continuous to point process variables is achieved. A method of surrogate data is provided in order to show the difference between original neuronal patterns and their surrogate stochastic data. Therefore, this method shows that neuronal discharge patterns cannot be fully described in terms of interspike interval histograms. However, in the intact cord most (73%) of the neurons displayed background activity with low (0.28-4.3) D2 values. During spinalization, D2 values significantly increased in 68% of the neurons showing previously low D2 values, irrespective of classification and laminar location of neurons, thus proving that tonic descending systems may maintain a high order in the discharge of sensory dorsal horn neurons.