Faced with a new external reality, the brain will activate the model which makes the best predictions associated with that external reality. If a model makes wrong predictions, we have to change the model with another one or to modify the model. Example: Newton's symbolic model of Mechanics makes wrong predictions associated with the objects moving at a speed comparable to the speed of light, but its predictions are good (correct) at lower speed. In any situation, the terms "correct" and "wrong" must be associated with a model or with a structure of models. We already described the first basic hardware facility associated with the brain (human or animal). It generates truth, reality, knowledge and consciousness. Now we will describe the second basic hardware facility of the brain. This is the action on the external reality. We already saw that faced with a section of the external reality, the brain makes at least one ZM model. A ZM model works in association with any available (or several) M-model and with any other ZMs of that brain. The main ZM is able to predict in a correct way the evolution of a section of the external reality. Such a ZM is able to make a new class of long-range models called ZAMs. ZAMs are artificial and invariant. An artificial model is made without any direct interaction with the external reality. An invariant model is a model, which cannot be changed by direct interaction with the external reality. A ZM model will make a ZAM model in order to modify the external reality. Once a ZAM is made, it becomes a reference model in changing the external reality. To do this, the ZAM-model works in connection with a number of AZM models. An AZM is a model which is already connected to the execution organs of a being (for human beings these are legs, hands and so on). Once a ZAM is activated, it will simulate the requested action using any information from all models of the brain. Based on simulations, ZAM will determine if it is able or not to meet the goal. If the simulation shows that the action is possible, then the ZAM will activate AZM models for action on the external reality. The ZAM will control the AZMs to act on the external reality exactly as in the successful simulation, with good chances of success. If by any simulation the objective is impossible to reach, the brain will be blocked to do that activity. Example: if a person has to jump over an obstacle, that person will know very fast if the jump is possible or not. The person knows this, because a ZM makes a ZAM-model, which is associated to the external reality (the person itself, the supporting surface and the obstacle, as main elements). The ZAM then simulates the jump on the model. If the simulated jump fails, the brain is blocked to do the action. If the jump is done with success in the simulation, the ZAM will control the body during the jump exactly as it was in the simulation, with good chance of success.