Connected with this specific problem, there is also another variant that a parallel model to the main model of walking was created. This parallel model predicted that the stone will slip and built a saving model outside the line of the PSM. However the theory predicts only two lines: one of the PSM and one of the ZAM which controls the global activity. Possibly, the ZM could let the main ZAM act, and build in parallel other ZAMs models for new situations, which would be activated in special cases. This type of behaviour is not specifically excluded by the theory, but in reality it is not met sufficiently clearly, so that it can be sustained. Building a parallel model is an easy operation, but the question is, how does the main ZM know what other ZAM to activate, when the active ZAM does not correspond anymore. The implementation of this facility could be done if there were a 'pipeline' built by the main ZM, so that a specific order of activation of parallel ZAMs existed in special cases. But this would imply the existence of a new hardware. As I already said, the existence of this facility (pipeline of ZAMs) cannot be sustained yet, due to insufficient data, but could be a line of further hardware development of the brain. The issue of walking, jumping and running is inimaginably complicated and I do not believe that in predictible future, robots will come close to the performance of a chicken a few days old, running on a difficult terrain. Climbing trees is an even more complicated activity, than walking and jumping. The basic information is related to the lack of precise information about the resistance of the branches. The models are able to make an evaluation of the resistance of each branch, but the model will have enough simulations in which the branch will break. ZM will need to take this into account, based on various local models, in order to build a good strategy (the best ZAM reactualised very often). In this case, the stability in the tree will be given by the capacity of building alternative models, which could be activated, if a branch broke. The brain effort needed to ensure the stability of the person in a tree is huge. Not all brains have this capacity. Moreover the ZM should also build a 'saving' model, in which there should be at least three points of support at any moment, in the ideea that if at least two will behave as in the simulation, the system will have an acceptable level of stability. Walking on a difficult terrain, jumping and the stability in tree climbing are tests, which can show global performance of humans in the domain of image models. In animals these functions can be even more efficient. ETA 26: The brain evolves under our eyes. Generally all ETAs refer to the behaviour and evolution of the brain of a normal average human.