This paper argues that another microeconomics, methodologically different from neoclassical reductionism, is rapidly developing: systemic and evolutionary microeconomics. The former is based on clear-sighted agents that have access to information but do not really learn, and always proceed with substantive rationality. The latter is based on partially blind agents who learn and act rationally in an adaptive way. The study demonstrates that the crux of the difference between these two alternatives is the way they connect the parts among themselves and the parts to the whole. In the former, the agents are independent of each other and global properties are obtained by aggregation. In the latter, the agents are organized by social structures and they form compositions that have emergent properties. Therefore, the parts and the whole belong to each other; that is, they are inseparable. In the former, the agents are defined only by their intrinsic properties, but in the latter, they are also defined by their relations.
evolutionary microeconomics; systemic microeconomics; adaptive rationality; procedural rationality; composition problem