![]() |
Niels Taatgen |
HomeResearchOverview Skill Acquisition Multi-tasking Time perception Learning from instructionsTransfer: ActransferPublicationsPersonal |
Skill AcquisitionWhen I started working with ACT-R it lacked a plausible mechanism for learning new production rules, i.e. task specific strategies. During a half-year sabbatical in 1999, I developed the mechanism of production compilation as a plausible account of learning task-specific skills. In production compilation, general problem-solving strategies (e.g., analogy) are specialized by a single mechanism into task-specific strategies that in some models leads to qualitatively different behavior. As a first demonstration of the viability of production compilation, I modeled learning the English past tense (Taatgen & Anderson, 2002), which grew into a contender of connectionist models of the phenomenon. We later extended this model to the German plural (Taatgen, 2001), and to phonetic representations (Taatgen & Dijkstra, 2003). Production compilation has proved to be a useful mechanism in other developmental domains, for example the balanced-scale task (van Rijn, van Someren & van der Maas, 2003). Another component of skill acquisition is the ability to learn from instructions. Our idea is that new task knowledge initially has a literal format (examples or explicit instructions), and that production compilation gradually transforms these instructions into task-specific rules. I have demonstrated this idea in a model of the Kanfer-Ackerman Air Traffic Controller task (KA-ATC, Taatgen, 2002; Taatgen & Lee, 2003) where the model accurately predicted both aggregate data, and explained individual differences between individuals. The control aspect in learning dual tasking and complex dynamic tasksMy recent work has focused on a problem beyond production compilation: how can we explain that during skill acquisition people not only become faster, but also more robust and flexible? The solution I finally found was to abandon the traditional hierarchical representation of tasks (Taatgen, 2005; 2007). Instead of tying the instructions in a strict order, instructions now contain the preconditions in which they can be carried out, and are organized in instructions sets (I have called this a weak hierarchy). This new representation produces models with a nice balance between top-down and bottom-up control, and led me to formulating the Minimal Control Principle. This principle states that people organize their knowledge for a task to minimize the amount of top-down control they have to exert. In a cognitive model, the number of control states corresponds to the amount of top-down control. I applied this principle in several studies (Taatgen, 2005), and have used it as a guiding principle for subsequent models. Key referencesTaatgen, N. A., Huss, D., Dickison, D. & Anderson, J. R. (2008). The acquisition of robust and flexible cognitive skills. Journal of Experimental Psychology: General, 137(3), 548-565. (pdf) Taatgen, N.A. (2005). Modeling parallelization and flexibility improvements in skill acquisition: from dual tasks to complex dynamic skills. Cognitive Science, 29(3), 421-455. (pdf) Taatgen, N.A. & Lee, F.J. (2003). Production Compilation: A simple mechanism to model Complex Skill Acquisition. Human Factors, 45(1), 61-76. (pdf) Taatgen, N.A. & Anderson, J.R. (2002). Why do children learn to say "broke"? A model of learning the past tense without feedback. Cognition, 86(2), 123-155. (pdf) (models)
|