Ignoring versus updating in working memory reveal differential roles of attention and feature binding

Abstract

Ignoring distracting information and updating current contents are essential components of working memory (WM). Yet, although both require controlling irrelevant information, it is unclear whether they have the same effects on recall and produce the same level of misbinding errors (incorrectly joining the features of different memoranda). Moreover, the likelihood of misbinding may be affected by the feature similarity between the items already encoded into memory and the information that has to be filtered out (ignored) or updated into memory. Here, we investigate these questions. Participants were sequentially presented with two pairs of arrows. The first pair of arrows always had to be encoded into memory, but the second pair either had to be ignored (ignore condition) or allowed to displace the previously encoded items (update condition). To investigate the effect of similarity on recall, we also varied, in a factorial manner, whether the items that had to be ignored or updated were presented in the same or different colours and/or same or different spatial locations to the original memoranda. By applying a computational model, we were able to quantify the levels of misbinding. Ignoring, but not updating, increased overall recall error as well as misbinding rates, even when accounting for the retention period. This indicates that not all manipulations of attention in WM are equal in terms of their effects on recall and misbinding. Misbinding rates in the ignore condition were affected by the colour and spatial congruence of relevant and irrelevant information to a greater extent than in the update condition. This finding suggests that attentional templates are used to evaluate relevant and irrelevant information in different ways during ignoring and updating. Together, the results suggest that differences between the two functions might occur due to higher levels of attentional compartmentalisation – or protection – during updating compared to ignoring.

Shielding working memory from distraction is more effortful than flexible updating

“…The observation that the subjective cost of repeating ignore is higher than that of update is in line with the finding that participants perform more poorly on ignore compared with update trials. This finding concurs with previous results from studies using an analogous task with ignore and update conditions 19,35,36 . In those prior studies, however, the task-relevant delay between the to-be-remembered items and the probe was shorter in the update than the ignore condition, rendering inference about the cognitive mechanism underlying the performance difference difficult.…”

Dopamine guides competition for cognitive control: Common effects of haloperidol on working memory and response conflict

“…This analysis included drug (placebo, haloperidol), delay (short, long) and presence of irrelevant information, in other words ignoring/updating (absent, present). Consistent with the results of previous studies (Fallon, Mattiesing, Dolfen, Manohar, & Husain, 2018), there was a significant main effect of delay on recall (F(1,28) = 58.10, p < .001; Figure 3), such that error was reduced for short compared to long delays. Having to deal with irrelevant information, regardless of whether this was in the ignore or update condition, also significantly impaired recall (F(1,29) = 36.82, p <.…”

“…It could be expected that if this mechanism was responsible for coupling ignoring and overcoming response conflict, then updating should be similarly affected, as irrelevant information is also presented. However, recently, using a similar WM task, we found that the irrelevant -initially presented -information in the update condition is almost entirely purged from WM, i.e., there is very little evidence to suggest that the to-be-removed information in the update condition lingers to corrupt WM representations (Fallon et al, 2018). This suggests that the selection process at retrieval may be fundamentally different in the ignore and update conditions.…”

A new toolbox to distinguish the sources of spatial memory error

“…To illustrate how the models work with real human data, and to examine how they cope with more complex task designs, we fitted previously unpublished data collected from an Ignore/Update spatial WM task (i.e. a 2D task), and compared the pattern of recovered parameters to those found on a 1D orientation analogue of this task (Fallon, Mattiesing, Dolfen, Manohar, & Husain, 2018;Fallon, Mattiesing, Muhammed, Manohar, & Husain, 2017). New data were collected from 49 healthy older adults (mean age = 71.4 years, SD = 5.5, 18 female, 31 male).…”

“…In order to compare convergent validity between the 1D and 2D modelling approaches, we compared the fitted parameters from the 2D misbinding model to those reported on the 1D version of this task (Fallon et al, 2018). Histograms of the recovered parameters are shown in Figure 17, and show that while imprecision is approximately normally distributed, the other parameters are skewed, being bounded by 0 and 1.…”