At each step, the network combines the current input with the previous hidden state and emits a new state or output. Reusing the same transition across positions supports variable-length sequences but makes long-range learning and parallel training harder.