28 years ago, I discovered that any logical (physical in the TRIZ terminology) contradiction can be resolved by separating the contradictory requirements between somewhat opposite entities (dualities). Based on this discovery, I proposed several separation principles. Of them, only separation between parts and the whole was accepted by Altshuller and incorporated into TRIZ. The others were rejected because I could not provide examples from mechanical engineering (the only field of engineering that Altshuller understood) to illustrate them. TRIZ community never learned about them.
The paper "Inventiveness as belief revision ..." features some of the rejected separation principles. It is, for example, separation between space and time. Do not confuse it with separation in time or with separation in space. In the latter two, separation is performed between different points in time or between different locations in space. But the idea of the feeback circuit analysed in the paper, is specifically based on separation between space and time.
Another separation principle is separation between controller and controlled, which is demonstrated on the example of triode. Yet another one is separation between the total pressure and the partial pressure.
The latter principle is obviously specific to a particular area of technology. This indicates that the separation principles vary in the degree of their applicability. Some are applicable to any area of technology, whereas the others are area specific.
However, not all the above is the main point of the paper. The main point is that for the first time concepts of TRIZ were put into the framework of Artificial Intelligence (AI). So far, whatever was accumulated in AI during the decades of research was not used in TRIZ and vise versa. In this respect, the paper is a turning point.