Hi Roye,
From R2020a it is recommended to not use "rlRepresentation". I understand that you want to create a representation for a critic that takes both observation and action as input, and the agent you are using is DDPG. For MATLAB R2021a, based on your requirement, it is recommended that you use the "rlQValueRepresentation" function for your inputs as follows:
actor = rlQValueRepresentation(actorNetwork,env.getObservationInfo,env.getActionInfo,'Observation',{'observation'},'Action',{'ActorTanh1'},actorOptions);
You may refer to the following documentation for more information: https://www.mathworks.com/help/reinforcement-learning/ref/rlrepresentation.html#mw_a6277225-fecf-4d97-9549-1fc4799bf5b6.
