Composing music from neuronal activity: The spikiss project

Abstract

We describe here an attempt to compose music, constrained by recordings of brain activity where excitatory and inhibitory neurons were discriminated and used to trigger simple tones or more complex sounds. We used experimental recordings of brain activity under the form of "spikes", recorded with microelectrodes in human subjects. The recordings come from different sources, which have been all published in the neuroscience literature. We emphasize here the natural rhythmical activity of neurons, in particular, that of inhibitory neurons. Inhibitory neurons are thus naturally suited for driving bass sounds and rhythmic sections. The sparser activity of excitatory neurons is exploited here to reveal melodic capabilities, which are sometimes exacerbated by subjective choice of scales and tones. We explain step by step how this can be done and provide examples of musical sequences and tracks composed from neuronal activity during different brain states, such as wakefulness, deep sleep, or paradoxical sleep (dreaming). We suggest to extend this approach to more global signals, such as the electroencephalogram or neuroimaging signals.