The finding that irrelevant sound that changes from one token to the next disrupts working memory more strongly than steady-state sound is well-established as the changing-state effect. The two dominant accounts differ in whether they expect the changing-state effect to be process-sensitive: According to the duplex-mechanism account, changing-state effects are restricted to tasks in which participants process order information. According to an attentional capture account, they should occur independent of the role of order information. Most studies manipulating the task observed a stronger changing-state effect in tasks that necessitate order information than in tasks that do not, in line with the duplex-mechanism account. However, contrary to this account, a changing-state effect can also be found in tasks without order information. Moreover, a single task (i.e. the missing-item task) is almost exclusively used as the task without order information and the focus is on tasks demands instead of on what participants actually do.
In our experiment, we used four different tasks (two based on order information, two not), recorded participants’ strategy choice and used Bayesian modeling to evaluate if the changing-state effect is actually equal to zero when no order information is processed. Surprisingly, a changing-state effect was observed in only one of the order-based tasks and only in this task was the changing-state effect larger when participants applied an order-based strategy. The findings thus only partially support the duplex-mechanism account. In an ongoing conceptual replication, we aim to increase the probability of observing a changing-state effect in the first place.