As often happens for the rotational world, the Newton's law can be restated in terms of angles, moment of inertia and torques replacing displacements, masses and forces (this is one source among millions on the web).
While for the translational case, I have no problem to figure out which are the forces due to the Newton's third principle, with a rotating system composed by many bodies, I have some problem.
To overcome my problem, I try to figure out some "mental experiments" which I think may help me to understand the topic. Here is one of these experiments.
Consider a simple DC motor able to generate a torque. We can assume that this motor is a dimensionless object with mass $M_1$. We fix this motor on a horizontal surface (moment of inertia: $I_2$, mass: $M_2$), which can freely rotate around a point.
We put a bar (moment of inertia: $I$, mass: $M_3$) on the top of the motor, like a gramophone.
The scheme is represented in the following figure:
The bar is fixed, so we cannot move it (high forces may destroy it). The only way to move the bar is to turn on the motor.
If we switch on the motor, then it will create a constant torque $\tau$.
The system I'm proposing is composed by $3$ bodies: $2$ rigid bodies (the surface, which is not necessarily fixed on the ground, and the bar) and $1$ dimensionless (the motor).
The question is:
Which are the torques on the bar, on the motor and on the rotating surface due to the Newton's third principle?
We can also consider the floor as a forth body. In that case, which is the torque on the floor?
The intuition suggests me that when I turn on the motor, then both bar and rotating surface will start to rotate.