Skip to main content

How prior brain states govern access to working memory

Periodic Reporting for period 1 - ACCESS2WM (How prior brain states govern access to working memory)

Reporting period: 2016-09-01 to 2018-08-31

This Marie Curie action (acronym ACCESS2WM) aimed to understand the mechanisms by which information is prioritised for access to human working memory. In addition, during the project, we have also investigated the related question of how information is accessed from working memory once this information becomes relevant for guiding behaviour. Working memory is a fundamental building block of adaptive cognition that allows us to hold available past sensory information that is anticipated to guide ensuing behaviour. Because it has limited capacity, it is key that only those pieces of information enter working memory that are (expected to be) of most importance to the agent. It is hence important to understand the cognitive and neural mechanisms that govern access to it. I set out to investigate the role of different kinds of prior expectations and brain states in governing this access.

Understanding these mechanisms is not only relevant for the advancement of basic discovery science, but may ultimately also have implications that are relevant for society. The breakdown of working memory is often ascribed a core problem cognitive disorders that see increasing prevalence in modern day society (such as ADHD). It is highly conceivable that at least part of these deficits originate from a non-adaptive regulation of what information accesses working memory (such as a reduced efficiency in filtering memory-targets from sensory distractors).
During the two years of the project, I ran three experiments (of which 1 currently published in Nature Communications, the others submitted) for which I took the lead in all aspects (experimental design, piloting, data collection, analysis, and dissemination). In addition, I contributed 3 review articles as first or last author (all of which published, including in the top-ranking journals Nature Reviews Neuroscience, and Trends in Neurosciences). I also supervised many related research project in the host laboratory (of which several have already resulted in shared first/last author publications), and acquired several new techniques as set-out in the personal learning aims of the project (including multivariate decoding analyses, hidden markov modelling to extract dynamic brain states, and eye-tracking analysis). I also presented my work at several international conferences and other universities.
In a relatively short period of time, we have made several ground-breaking discoveries within the study of working memory. First, we have uncovered novel mechanisms of how anticipatory (pre-stimulus) brain states prioritise the processing of (to-be-remembered) targets over temporally-competing distractors – namely by boosting target processing as well as delaying the interference on the targets caused by temporally-adjacent distractors (van Ede et al., Nature Communications, 2018). This project has also pushed a methodological advance by revealing how multivariate analyses can be used to simultaneously track target and distractor stimuli, even when these are presented in close temporal proximity. Second, we have developed a novel working memory task in which we have uniquely associated memory representation to their associated actions. In doing so, we were able to investigate the mechanisms by which both visual representations and their associated actions are accessed from working memory when individual memory representations become relevant for behaviour. This revealed the surprising observation that action plans are not selected after the relevant visual representation has been selected. Instead they are selected in parallel with the relevant visual representation, suggesting that parallel action plans are held available jointly with visual working memories – a hitherto unexplored concept (van Ede et al., submitted). Third, we have discovered that that accessing of visual information from working memory biases gaze in the direction of the attended memory item – even when there is nothing to look at in external space (van Ede et al., submitted). This uniquely reveals that the human oculomotor system is not only recruited for covert shifts of attention in the external space of perception (which has been known for a long time), but also within the internal space of memory.

These results contribute primarily to a better and more comprehensive understanding of the basic cognitive and neural mechanisms that support access to and from working memory. Ultimately these insights and developed methods and tasks may be useful for also for gaining a better understanding of the potential breakdown of such basic mechanisms in increasingly prevalent cognitive disorders, such as ADHD.