A feedforward neural network processes data from input to output in a linear manner, without any loops or feedback connections. A recurrent neural network, on the other hand, has loops and feedback connections, which allow it to process sequential data and maintain information over time.