By default, it is assumed that each step reads the output of its predecessor. If necessary, a step's output data may be assigned to a symbol that is read by another step further down in the process flow.