NSERC
2007-2011

Training and transfer in simple skills

Cousineau, D. (Université de Montréal)

A challenge that remains unresolved is one of selecting candidates that will become highly-skilled operators (such as pilots). The Sureté du Québec provincial police and the Montréal's police service (SPVM) have this problem where they hire new officers, train them through very expensive courses just to discover that a certain proportion of them are unable to master their car in extreme situations.

The aim of this project is to device diagnostic tools that will help to decide whether an individual has the potential to become a highly-skilled operator. We will mainly focus on the characteristics of the learning curve obtained in simple tasks (amplitude, asymptote and rate of learning). A pilote study (Nguyen and Cousineau, in preparation) showed that knowledge of the learning curve amplitude in a simple task was a very reliable predictor of the learning curve obtained in a different task for a given person. Can such a finding generalize to the learning curve obtained in a complex task such as learning to drive a car? Beyond the empirical data, we will also examine two theories of skill learning: (1) the framework of Haider and Frensch (1996, 1999) and Lewandowsky et al. (2002) where stimulus learning are partitioned within more general components. (2) A selection theory of learning where improvement is achieved through the elimination of bad strategies. This last theory suggests that the mean performance has no special status; only the variation is reduced with the elimination of wrong and slow strategies.