We propose a new account of concepts as specifically structured and handled mental files. We argue that concepts consist of two components, (a) an associative network of integrated information used for property based categorization and recognition; and (b) a handling system that organizes and sorts through this associative network. A certain type of concept is determined by the package of associated information integrated in a mental file and the specific structure of this information including the specific way this information is handled. With this framework, we can account for the large variety of concepts including everyday concepts of individual objects and properties, scientific concepts, natural kind concepts and phenomenal concepts.