This study investigated the synaptic basis for various types of spiking response profiles, and how synaptic plasticity learning rules are important for shaping synaptic inputs for spike output and network performance. They leveraged cell-attached, extracellular, and whole-cell recordings from behaving animals alongside recurrent network modeling to explore the synaptic origins and functional contribution of heterogeneous response profiles. The investigators also combined first-order reduced and controlled error training with a dynamic network to create a novel recurrent neural network with spiking units and multiple spike-timing-dependent plasticity rules to solve a stimulus classification task similar to that of trained rats and mice. This dataset contains electrophysiology and behavioral data. In addition, the publication includes source data for underlying figures.