In this project, we have demonstrated temporal signature extraction for eight different motor tasks. For this purpose, we has chosen a spatio-temporal architecture named as associated cluster-dependent sequence model, which divided the whole network into four different sub module in order to mimic biologically plausible somatosensory cortex (input), primary motor cortex (RNN layer), basal ganglia (GN neuronal gates), and thalamus (A action hub). To avoid fixed hardware connection and to introduce learning through training, we adopt Hebbian learning for RNN-G connectivity and delta learning for G-A connectivity. From our trained model with a particular motor sequence, we extract the temporal signature of the trained temporal networks. We have tested the robustness of the extracted temporal learning through temporal signature extraction, temporal shifting, temporal rescaling, temporal compositionality, and completely new pattern learning through multiplicative gain and weight learning mechanism as illustrated in the result analysis section. However, in future, our plan is to incorporate hebbian learning for both RNN-G connectivity and G-A connectivity and also find the relationship in motor actions versus learning sequence and training epoch.