Jelte Medema The problem state: a cognitive bottleneck in driving

The cognitive architecture ACT-R offers an integrated theory of human cognition which consists of multiple modules (Anderson, 2004). Salvucci and Taatgen (2008) proposed that ACT-R could be extended with threaded cognition, which adds the ability of parallel operations. A bottleneck can arise when multiple tasks need a module concurrently. Research of Borst and Taatgen (2007) shows that there is another potential bottleneck, called the problem state. The problem state is a temporary representation of an intermediate step of a task. The experiment of Borst and Taatgen was centered around a simple driving task. A car on a screen had to be steered by a keyboard. While adhering to given directions, participants had to type in a word. Interference occurs when both tasks require the problem state, i.e. when both tasks are difficult at the same time.
As a follow up study, the experiment of Borst and Taatgen is replicated using a driving simulator, which adds a more realistic setting. The results show that when both tasks are difficult, the performances on the tasks declines. This indicates a problem state bottleneck, confirming the results of Borst and Taatgen. The driving task and the typing task also influenced the actual driving behavior, which contributes to the real world value of this research.