Synthetic Psychology

Synthetic psychology is an approach to modeling or theorizing popularized by Braitenberg (1984). In synthetic psychology, models are not constructed after analyzing data. Instead, simple components or processes are combined to create a working system, the system is then embedded in an environment, and its behaviors are observed. That is, the model precedes data collection because the behaving model is the source of data. Synthetic psychology is a form of forward engineering.

Braitenberg has argued, in his law of downhill synthesis and uphill analysis, that synthetic models will be easier to create, simpler, and easier to understand than models created by the analytic approach. That is reverse engineering -- observing an intact system, and trying to infer its internal mechanisms -- will lead to more complex theories than necessary according to Braitenberg. The synthetic approach is clearly consistent with the work of behavior-based roboticists, but it is not restricted to such work. For instance, Dawson (2004) has argued that artificial neural networks are another interesting medium for conducting synthetic psychology.

References:

1. Braitenberg, V. (1984). Vehicles: Explorations In Synthetic Psychology. Cambridge, MA: MIT Press.
2. Dawson, M. R. W. (2004). Minds And Machines : Connectionism And Psychological Modeling. Malden, MA: Blackwell Pub.