Binless estimation of mutual information in metric spaces

The quantification of the association between stimuli and the neural responses they evoke is one of the major concerns of sensory neuroscience. Currently, it is widely accepted that the temporal structure of a spike train is an essential aspect of the neuronal responses. For example, studies in the cat have shown that this temporal structure carries substantial information about the sensory environment (Middlebrooks et al. 1994; Furukawa et al. 2000). As the temporal structure of auditory stimuli is uniquely complex, the auditory system is an excellent model system to study the contribution of spike timing to the neural code (Nelken et al. 2005). In consequence, the development and validation of methods of measuring the amount of information carried by spike timing are highly relevant in the context of the auditory system.