ARTICLE Communicated by Misha Tsodyks Short-Term Synaptic Plasticity Can Enhance Weak Signal Detectability in Nonrenewal Spike Trains

We study the encoding of weak signals in spike trains with interspike interval (ISI) correlations and the signals ’ subsequent detection in sen-sory neurons. Motivated by the observation of negative ISI correlations in auditory and electrosensory afferents, we assess the theoretical per-formance limits of an individual detector neuron receiving a weak sig-nal distributed across multiple afferent inputs. We assess the functional role of ISI correlations in the detection process using statistical detec-tion theory and derive two sequential likelihood ratio detector models: one for afferents with renewal statistics; the other for afferents with nega-tively correlated ISIs. We suggest a mechanism that might enable sensory neurons to implicitly compute conditional probabilities of presynaptic spikes by means of short-term synaptic plasticity. We demonstrate how this mechanism can enhance a postsynaptic neuron’s sensitivity to weak signals by exploiting the correlation structure of the input spike trains. Our model not only captures fundamental aspects of early electrosensory signal processing in weakly electric fish, but may also bear relevance to the mammalian auditory system and other sensory modalities.