(Mis)understanding and learning of feedback relations in a simple dynamic system

Abstract: The work presented in this dissertation has focused on laypeople's understanding of a simple dynamic system. The system in question consisted of a predator-and-prey ecology. A computer simulation was used to portray foxes feeding on rabbits. In the simulated game, the participants were allowed to change the number of foxes once per year, where the task was to establish equilibrium. General performance, reasoning and control behavior was investigated in Study 1. The task was found to be difficult, despite its structural simplicity. Low ability to apply indirect reasoning seemed to be the major obstacle. To reach a constant fox population, the rabbit population had to be adjusted to the appropriate size, and the means for obtaining that is the fox population. This was highlighted in Study 2 by the addition of pictorial hints, describing how the foxes were affected by the rabbit supply. Performance, however, was not influenced by the pictorial aids. The participants were able to understand and use the pictures, if explicitly demanded to, but otherwise they treated the pictures as redundant information. If the foxes are, for example, reduced, they lack sufficient food. To allow for growth of the rabbit population, the foxes might need to, temporarily, be reduced even further. In Study 3, many participants understood how the rabbit supply affected the foxes, yet they failed to resist an "urge" to replace lost foxes (and thereby further reduce the rabbit population). In a second predator-and-prey task, participants previously assisted in their completion of the rabbits-and-foxes task outperformed controls who only performed this second task. The performance superiority, however seemed to stem from learning for only half of the participants who were administered the rabbits-and-foxes task. Learning in about half the participants has proven a stable finding.