The Training and Transfer of Dynamic Decision Making Performance

Abstract

This study investigated the relationship between the cognitive load experienced during training, cognitive ability and previous electronic gaming experience, and the transfer of performance to a novel situation. The training and transfer of dynamic decision making (DDM) performance from one decision making environment to another is critical for the development of effective performance under operational DDM conditions.

Cognitive ability was assessed with the standardised measures, Raven’s Standard Progressive Matrices and WOMBAT™ version 4.0, respectively. Electronic gaming experience was measured with a Likert Scale. Next, the computer game called Fire Department 3 was used to simulate various DDM environments. Participants’ practiced completing fire fighting objectives under either low or high conditions of cognitive load and then performed a (transfer) test in an alternative fire fighting environment under an intermediate level of cognitive load. Sixty-three students completed the study. Based on previous findings, it was expected that only the performance of participants who practiced under low cognitive load would improve significantly from the practice phase to the test phase, and the degree of positive transfer would be largest for those with low cognitive ability or no recent gaming experience. It was also predicted that WOMBAT™ test scores would be more predictive of DDM performance than Raven test scores.

A repeated measures analysis of covariance (ANCOVA) revealed a significant three-way interaction between test phase (practice and transfer), cognitive load (high or low) and previous electronic gaming experience. As predicted, WOMBAT™ test scores were found to be more predictive of FD3 simulation performance than Raven test scores.

Contrary to expectation, the relationship between cognitive load and transfer performance was not significantly related to cognitive ability. Furthermore, the results showed that the mean performance of neither group improved significantly between the practice phase and the test phase. Overall, compared to practicing under high cognitive load, practicing under low cognitive load was associated with better practice phase performance. However, compared to practicing under high cognitive load, practicing under low cognitive load was also associated with a significant reduction in performance on the first transfer test for participants with recent gaming experience, but not for participants without recent gaming experience. The results from a second transfer test, although non-significant, were consistent with the results of the first transfer test.

Major implications of the present study are that training conditions which facilitate faster skill acquisition and improved performance during training are not directly related to better transfer performance. It was found that high levels of cognitive load may be used during training to improve transfer performance. However, the interaction between test phase, cognitive load and previous electronic gaming experience shows that, to maximise transfer performance, cognitive load during training must be tailored to the characteristics of different individuals (e.g., level of electronic gaming experience). This study also shows that the WOMBAT™ test appears to have specificity for discriminating between different levels of performance in laboratory and real-world DDM environments.