The invention provides a two-step approach to providing a BCI system. In a first step the invention provides a low-power implantable platform for amplifying and filtering the extracellular recording, performing analogue to digital conversion (ADC) and detecting action potentials in real-time, which is connected to a remote device capable of performing the processor-intensive tasks of feature extraction and spike classification, thus generating a plurality of predetermined templates for each neuron to be used in a second processing step. In the second step the low-power implantable platform amplifies and filters the extracellular recording, performs ADC and detects action potentials, which can be matched on-chip to the predetermined templates generated by the external receiver in the first step. This two-step approach exploits the advantages of both offline and online processing, providing an effective and safe method for performing multiple recordings of single-neuron activity, for research or monitoring applications or for control of a remote device.
