Flexible updating of visual working memory - The joint roles of attention and action
Visual working memory allows us to retain information over short periods of time, thereby enabling the comparison of objects separated in time or space. This ability is critical for various tasks, but it is highly limited in capacity (e.g., Luck & Vogel, 2013). As visual information constantly g...
|PDF Full Text
No Tags, Be the first to tag this record!
|Visual working memory allows us to retain information over short periods of time, thereby enabling the comparison of objects separated in time or space. This ability is critical for various tasks, but it is highly limited in capacity (e.g., Luck & Vogel, 2013). As visual information constantly gains or loses relevance as we interact with our environment, there is a need to update the contents of visual working memory in a flexible manner to ensure that its limited capacity is used efficiently. In five studies, this dissertation examined how this updating is accomplished.
The first part of this dissertation (Studies I-III) investigated updating following so-called retrocues presented during visual working memory maintenance, indicating some memorized items as more task-relevant than others and thereby inducing a strategic internal orienting of attention (e.g., Griffin & Nobre, 2003). Results showed that this attentional selection of representations yields a benefit (i.e., better memory) for task-relevant information (Studies I – III), with the magnitude of this benefit being related to the attentional efficiency of an individual (Study I). The consequence of this attentional selection for the other, unselected representations is sensitive to task context: When there are graded differences in the relevance of maintained information, the contents of visual working memory can be weighted to reflect these differences. While the most important information is robustly maintained inside the focus of attention, less important information can be maintained in a more vulnerable state outside the focus of attention, from where it can be accessed to be refocused and retrieved if need be (Study I). Studies II and III established that different visual properties (e.g., location or colour) can be used to guide the selection of relevant representations. A basic distinction can be drawn between mechanisms of spatial and feature-based attentional selection, which can be dissociated in terms of behavioural signatures (Study II) and involved cortical areas (Study III).
The second part of the dissertation looked at the effects of more natural indicators of the relevance of specific aspects of our visual surroundings: actions and action-intentions. Instead of presenting retrocues during visual working memory maintenance, Studies IV and V used dual-task paradigms, in which an action was to be executed or prepared. This action rendered some items in a concurrently performed memory task more potentially relevant than others. Results revealed that selective action-related processing continues to influence visual processing beyond the perceptual stage, inducing an updating of visual working memory that reflects differences in the action-relevance of representations. Representations that hold potential action-relevance because they spatially correspond to the location of an action goal (Study IV) or because they contain information that is coded on a feature-dimension that is critical for a particular type of action being prepared (Study V) are preferentially maintained and recalled with higher accuracy than information that is action-irrelevant. The effect of actions on maintenance was found to be particularly pronounced when memory load corresponded to the average visual working memory capacity (Study IV), suggesting that information holding potential relevance for an action is prioritized when demand on the system is high. Furthermore, Study IV provided evidence that action-related enhancement at the mnemonic level is spatially not as precise as it has been shown to be for perception (e.g., Baldauf et al., 2006). Instead, results indicated a graded enhancement spreading out from the representation corresponding to the action goal location.
In sum, the present dissertation demonstrates that updating of visual working memory is remarkably flexible. Maintained information can be weighted to reflect graded differences in relevance (Studies I-V), irrespective of whether this relevance is explicitly indicated by external cues (Studies I-III) or more implicitly indicated by action intentions (Studies IV and V). Different representational characteristics can guide the selection of relevant memory contents: Updating is induced when some representations are more important than others because they correspond to relevant locations (Studies I-IV) or because they contain a feature, which is more relevant than other features of the same dimension (Studies II and III) or coded on a feature dimension that is more relevant than other dimensions (Study V). This flexibility highlights the versatile nature of visual working memory, which allows for an efficient use of its highly limited capacity in any given situation.