Precise timing is critical for motor control such as singing and speech.Chain networks,in which groups of neurons form feed-forward networks that support sequential activations of successive neurons,are shown to operate in the pre-motor nucleus HVC of songbirds during singing.They are also prime candidates for sequential firings observed in mammalian prefrontal cortex and basal ganglia.How such networks are formed is not well understood.In this talk,I will present a self-organization mechanism via computational modeling.Through synaptic plasticity and network activity,the chain network gradually emerges when a small group of “training neurons” are intermittently activated.The process requires spike-timing dependent plasticity(STDP).However,STDP must be supplemented by other synaptic plasticity rules,including synaptic decay and strength-dependent axon remodeling.Furthermore,it helps that the chain network forms during the development,in which the GABA inputs to the excitatory neurons change from excitatory to inhibitory as the excitatory neurons mature.The chain network formation is then guided by the structure of the inhibitory-excitatory neuron connections.With these synaptic plasticity rules and neuron properties,the chain network grows through successive attachments of neuronal groups.I will discuss the possibility of observing such a process in experiments.