Cortical computations and cortical depolarisations Per E. Roland Karolinska Institute, S-171 77 Stockholm, Sweden

The relation between cortical computations and measurements of the BOLD signal, cerebral metabolism and regional cerebral blood flow can be formulated in two ways. Knowing the cortical computations it might be possible to predict the changes in rCBF, BOLD and rCMRgl. This will take many modelling assumptions. Neuronal computing is defined as a depolarisation of the synaptic terminals in the dendrites subsequent depolarisation and the eventual transformations of these local dendritic depolarisations into action potentials. The computations are by nature non-linear and engage hundreds of thousands of neurons in the supragranular layers. This is apparent from real time measurements of the membrane potentials in the dendrites of layer 2/3 neurons in the visual cortex of the ferret. Here the distal dendrites of a vast population of neurons counting more than hundred thousand are more or less depolarised at least once following a natural stimulus. The resulting depolarisations have been termed dynamic depolarisation fields (DDF’s). The predictions following from the dynamic depolarisation field hypothesis are that all computations in the supragranular layers will give rise to such a DDF’s, and that different stimuli or tasks will give rise to different dynamics (Roland, 2002). This will be illustrated by measurements of the membrane potential changes to stationary and moving visual stimuli. The depolarisation fields last more than two hundred milliseconds even to very short stimuli. This long lasting depolarisation is presumably associated with large ionic transfers over the cell membrane and over different internals compartments, eventually leading to the release of pre capillary dilating signal molecules as well as the decrease in the ATP concentration. These events drive the oxiditative metabolism and the regional cerebral blood flow in a strongly non-linear fashion. The prediction from the animal experiments is that larger blobs of activity should be detected in measurements of the regional cerebral blood flow, oxygen metabolism or BOLD signal from similar stimulations in humans. The difficulties in making exact predictions of the depolarisation effects on rCBF and BOLD signal are further amplified by the long delays of the reaction of the pre-capillaries. Whereas it might theoretically be possible to model the events knowing the depolarisations and their dynamics, it is not possible from observing rCBF or BOLD signals or measurements of the regional cerebral glucose consumptions or similar in humans to infer the nature of the computations taking place.