Problem solving model for analyzing strategies of plastic regulators

We present an application of an analytical model of the constraints on problem solving in the professional domain of plastic injection press regulators who must assess the causes of product defects and fix them by regulating the press. The constraints model was developed for classic problems of modifying the state of a situation until the final desired state is achieved. It formalizes the idea that problem solving is a compromise among demands that can ultimately be contradictory: achieve the goals of the task using one's knowledge of the situation, interpretations of the objects and their properties, and the adaptation heuristics of daily life.

Constraints are rules that restrict possible actions: goals, interpretations, and heuristics are expressed as additional constraints. The accumulating restrictions can result in dead-ends, or situations where a person has no options considering the constraints, or certain constraints must be sacrificed in order to act.

The model is based on the principle of analyzing protocols to find all the rules for action that might be used by the subjects of interest: this is the space of possible rules. A protocol is a sequence that takes place in a space difined by the status of the adjustments: this is not visible but can be inferred in part by product defects and other information available in the system. The problem resolution is therefore a sequence with low visibility. An adjustment affects the state of the system and one must find rules that are potentially able to take into account the required information and the modifications that were made. Computer algorithms allow us to determine what subset of rules will provide a sequence in space that is as close as possible to the one observed in the analyzed protocol. These rules lead to an interpretation of the protocol. This group of rules will change during problem solving, leading to an evolution stemming from the fact that each regulator resolved many problems.
It was possible to effectively simulate all of the regulators studied. Each regulator had a varied level of competence, which can be  characterized in specific terms: their knowledge of the defect causes, their efficiency in taking in information, and their reading ability. In addition, the analysis of the rules in each protocol revealed three problem solving strategies corresponding to three different ways of viewing the problem.