Decision Chemistry Part II: The Emergence of Routines


Alessandro Lomi

Stefano Cacciaguerra


Last Updated September 2003





In the second part of the “Decision chemistry” model series, the objects that populate the stylized organizational world that we have designed are computationally more sophisticated than the objects in the first model. In particular, participants, problems, solutions and opportunities are now all endowed with communication capabilities and sensory mechanisms that allow more complex rules of search and motion, and hence induce more complex patterns of interaction. Objects communicate by leaving “information trails” that other objects can then sense and follow. In order to be followed, information trails must be (i) sufficiently strong, i.e., they must be above a certain sensitivity threshold, and (ii) visible, i.e., they must be within the angle of vision of the various objects.


Information trails are diffused in local neighbourhoods that are defined around individual objects. In the movie of the model’s behavior, the information trails can be seen as green a “halo” visible around each object. Information trails slowly disappear - or “evaporate” - if they are not followed. But if they are followed, they become stronger thus attracting more objects in the corresponding local neighbourhood. This (strictly local) mechanism of positive feed-back induces “routines” – relatively stable clusters of problems, solutions, opportunities, and participants. In the movie of the model’s behavior, routines can be seen as white areas of varying size.


As in part one – and in keeping with the original model of Cohen, March and Olsen - the objects that populate this more complex organizational world are: problems (yellow triangles), participants (blue squares) and solutions (red circles). The collision rules among the objects are those defined in the first model (Decision chemistry on a lattice part one). In particular, we recall that the collision between a “problem” and a “solution” generates an “opportunity.” Collisions between opportunities and participants may then give rise to different types of decisions depending on the relative energy level of individual objects. In the first version of the model the entities followed a random walk though the lattice.


The basic motivation behind this second and more advanced model is to show how aggregate entities called “routines” may emerge, grow, change and disappear as a consequence of simple rules of interaction among objects that are defined at lower levels. The main analytical insight offered by our modelling exercise is that conditions can be defined whereby garbage can-like decision processes evolve to ordered global states. Finally, from a theoretical point of view, the value of our model is to provide a structured context in which the implications of the mobility of agents for the structure of social systems can be discussed and explored. 


Visualizing the model


The following are links to an AVI movie of an illustrative run of the second model in the “Decision Chemistry” model series The movie is obtained by sampling one state change every second in order to save space. The underlying model produces a significantly less discrete behavior. The files large.avi (about 7.8MB) and small.avi (about 3.4MB) contain AVI movies of the model that can be viewed directly with any standard video player. The files (about 3.4MB) and (about 1.7 MB) contain zipped versions of the same movies.




The model described in “Decision Chemistry Part II: The Emergence of Routines” is programmed in Netlogo. The NetLogo User Manual reads: NetLogo is a programmable modeling environment for simulating natural and social phenomena. It is particularly well suited for modeling complex systems developing over time. Modelers can give instructions to hundreds or thousands of independent "agents" all operating in parallel. This makes it possible to explore the connection between the micro-level behavior of individuals and the macro-level patterns that emerge from the interaction of many individuals.” The NetLogo software, models and documentation are distributed free of charge for use by the public to explore and construct models. The most recent version of NetLogo can be downloaded from:





To refer to the present work in academic publications, please cite as:

Lomi, A., Cacciaguerra, S. 2003. Organizational decision chemistry on a lattice. Part II: The Emergence of Routines. Working Paper. Department of Computer Science. University of Bologna.




This work was supported in part by the Italian National Research Council (CNR) (grant number CNRC004195_001, and Grant Ex-60%, Scientific Committee number 13)