Task-unrelated thought depends on the phonological short-term memory system more than the visual short-term memory system.

This study examined which type of short-term memory (STM), phonological or visual, is involved in and more important for representing contents of task-unrelated thoughts (TUTs). Three experiments consistently showed that TUTs were less likely to be reported during phonological STM tasks than either visual STM or control tasks. In contrast, the number of TUT responses did not considerably differ between visual STM tasks and control tasks even for TUTs with many visual images. This difference cannot be explained by the differential involvement of executive control processes because task difficulty was controlled for in the multi-level logistic regression analysis. These results, together with the finding that most TUT responses contained verbal images, suggest that phonological STM plays an important role in representing verbal images in TUTs, while visual STM is less or not involved in representing TUTs, even for those with many visual images.