Such models are not changed by direct interaction with the external reality. If a ZAM fails to do something, the upper ZAM or the local-ZM will build and activate another ZAM. During their activity, they have full access to any resource of the brain (internal or external, by a ZM model). AZM: these short-range models are direct-action models. They are connected directly to different organs which can act on the external reality (hands, legs..) They keep precise information about such organs. When a ZAM model has to make a simulation of a specific action, the AZM models have to give precise information about every feature of the associated organs. Referring to ZAM and AZM models, any action on the external reality is based on a simulation. Without a successful simulation, the brain is not able to do any activity. Once a simulation is a success, the main ZAM will activate the action (in connection with local-ZM). The action will be the same as the simulated action. This procedure is followed by any ZAM in any activity. There are some exceptions. When the PSM takes control, some reflex-models are activated without initialization and without simulation. The reason is the highest speed of action, even if the action is not the right one. ACTIVE-TYPE MODELS (normal ZM-models) ZM models are made mainly by direct interaction with the external reality. They generate the knowledge and consciousness. They are self-activated in any situation when they are able to predict in a good way the possible evolution of a given external reality. A ZM, which interacts directly with the external reality, is also a model, which controls the whole body. They build, activate and deactivate any ZAM, based on a set of goals. A long-range ZM model also controls a local-ZM. This is able to modify a local-ZM model, based on long-range predictions. The whole activity is supervised by PSM. Usually the PSM is activated only when all the available ZMs fail to control the interaction with the external reality. A ZM model has full access to all the resources of the brain. It can take any information from any part of the brain and can make any model based on any available information. By analyzing the normal activity of the brain, one understands that some activities use only a limited number of models. The access to some models is easier than to others. We can find very fast some information and find more slowly other. This suggests that some models, which are usually employed together, are grouped. We define a quasi-structure of models called MZM. A MZM is a group of models, which are used often together. They can be associated with some specific complex activities (job, hobby, family life, car driving and so on). STORY-TYPE MODELS These are transition-ZMs. When there are some information from the external reality, the normal activity of the brain is to build a normal ZM-model or to find the best available ZM to integrate that information.