Eren Günseli
Assistant Professor
Research Interests
Make your next move.
Prepare your application
Review drafts before you send them.
Explore related searches
Articles (15)
Working memory (WM) has long been viewed as a store that must be engaged to maintain task-relevant information. Yet growing evidence shows that the brain often avoids activating WM, relying instead on perception or long-term memory (LTM). What determines whether and when WM is engaged? Rather than assuming obligatory maintenance, this review documents systematic patterns of WM avoidance and reactivation across perceptual, mnemonic, and control domains. I propose that these patterns reflect a tradeoff between metabolic and cognitive costs and behavioral flexibility. When perception or LTM can support behavior with low effort and stable content–context bindings, WM is bypassed through external or internal offloading. WM engagement emerges when behavior requires verifying or updating the match between memory content and the current context. From this perspective, WM activation is not the default state but a strategic investment: recruited only when low-cost perceptual or LTM reliance cannot guarantee correct content–context binding. Integrating behavioral, electrophysiological, and computational approaches, this framework unifies disparate findings under a single principle: WM engagement reflects a conditional control policy governing not only what is stored, but when active maintenance is recruited.
Year:
2026
Most daily tasks require frequent information exchange between working memory (WM) and long-term memory (LTM). However, the factors that modulate the reactivation of LTMs in WM remain to be explored. Here, we tested the effects of anticipated perceptual interference (in Experiment 1) and anticipated dual task interference (in Experiment 2) on reactivation of LTMs in WM using contralateral delay activity (CDA) in the EEG. On each trial, participants saw one or three previously studied or novel objects that were tested after a brief retention interval. In half of the blocks, the retention contained perceptual distractors (Experiment 1) or dual task interference (Experiment 2). In both experiments, the CDA prior to the interference period was similar regardless of the anticipated interference. Additionally, we measured contralateral alpha power suppression as an index of spatial attention. We found that spatial attention allocated to the memorized objects’ locations also was not modulated with anticipated interference. These findings indicate that anticipating interference does not influence the extent to which WM is engaged in representing information stored in LTM.
Year:
2026
Adaptive cognition requires determining when previously stored information must be re-engaged to guide behavior. Prevailing accounts of working memory and cognitive control propose that contextual change cues the retrieval or reactivation of relevant information, reinstating stored representations into working memory when the task model changes. Here, I show that reactivation depends not on change per se, but on the stability of the change. In an EEG experiment, participants performed visual memory tasks using repeatedly presented items. The task rule was either repeating or consistently varying across trials. While contralateral delay activity (CDA) declined with memory repetition, reflecting reduced working memory engagement, this reduction was unaffected by whether the task rule consistently changed or repeated. Thus, when variability itself was consistent, it was learned as a stable context, no longer prompting reactivation. Reactivation, therefore, is not triggered by change itself but by the failure of contextual stability, revealing that consistent variability is learned as a stable context.
Year:
2026
Gating processes that regulate sensory input into visual working memory (WM) and the execution of planned actions share neural mechanisms, suggesting a mutual interaction. In a preregistered study (Open Science Framework), we examined how this interaction may result in sensory interference during WM storage using a delayed match-to-sample task. Participants (12 males, 20 females) memorized the color of a target stimulus for later report on a color wheel. The shape of the target indicated which hand they would adjust the color wheel with. During the retention interval, an interference task was presented, requiring a response with either the same or different hand as the main task. In half of the interference trials, the interfering task cue was also colored to introduce visual interference. EEG results showed early motor planning during sensory encoding, evidenced by mu/beta suppression contralateral to the responding hand. The interference task only impaired WM performance when it included an irrelevant color, indicating that the interference effect was primarily driven by the irrelevant sensory information. In addition, color reporting in the WM task was biased toward the irrelevant color. This was more pronounced when both tasks were performed with the same hand, suggesting a selective gating mechanism dependent on motor control processes. This effect was mitigated by a control mechanism, which was evident in frontal theta activity, where higher power predicted lower bias on the single-trial level. Our findings thus reveal that sensory WM updating can be induced by interfering motor actions, which can be compensated by a reactive control mechanism.
Year:
2025
Event segmentation is the cognitive process of dividing continuous experiences into meaningful units. Although working memory (WM) is believed to play an instrumental role in event segmentation, its specific contribution remains under debate. Here, we used two electroencephalography (EEG) indices of WM load, alpha-band suppression and contralateral delay activity (CDA) amplitude, to test whether WM supports event segmentation by gradually accumulating information during unfolding events or by reactivating information at the transitions between events (i.e., event boundaries). Participants viewed object images paired with sounds that both remained in the same category for 6 consecutive items to form a stable event, followed by 2 items from another category, creating an event boundary. Temporal order and sequential memory were used to assess event segmentation. We found increased alpha-band suppression following event transitions, supporting the reactivation account. Evidence for the accumulation was mixed, with no evidence for increased alpha-band suppression and moderate evidence for increased CDA during events. These results indicate that WM contributes to event segmentation primarily through boundary-triggered reactivation, with only limited evidence for within-event accumulation.
Year:
2025
Results regarding the effect of divided attention during LTM retrieval are inconclusive. This may be due to most prior work relying on categorical recall or recognition measures, which lack sufficient precision to detect subtle effects. Here, we provide the first test of divided attention effects on LTM retrieval using a continuous report paradigm, allowing retrieval errors to be measured with greater precision. In Experiment 1, participants (N = 44) encoded object–orientation associations and later retrieved object orientations under either full or divided attention while performing a visual search task. Dividing attention during retrieval resulted in increased errors. To characterize the nature of these errors, we analyzed error distributions using both a standard mixture model (SMM) and the Target Confusability Competition (TCC) model. The SMM showed that divided attention reduced memory accessibility, while the precision of retrieved memories remained unaffected. The TCC model provided better fits and similarly indicated reduced memory strength under divided attention. Because Experiment 1 included an additional working memory load due to withholding the divided attention task response throughout the main retrieval task, Experiment 2 (N=44) replicated the main findings while eliminating this load by having participants respond to the secondary task before the retrieval response. Together, these findings show that continuous report reveals robust costs of divided attention on LTM retrieval.
Year:
2025
Results regarding the effect of divided attention during LTM retrieval are inconclusive. This may be due to most prior work relying on categorical recall or recognition measures, which lack sufficient precision to detect subtle effects. Here, we provide the first test of divided attention effects on LTM retrieval using a continuous report paradigm, allowing retrieval errors to be measured with greater precision. In Experiment 1, participants (N = 44) encoded object–orientation associations and later retrieved object orientations under either full or divided attention while performing a visual search task. Dividing attention during retrieval resulted in increased errors. To characterize the nature of these errors, we analyzed error distributions using both a standard mixture model (SMM) and the Target Confusability Competition (TCC) model. The SMM showed that divided attention reduced memory accessibility, while the precision of retrieved memories remained unaffected. The TCC model provided better fits and similarly indicated reduced memory strength under divided attention. Because Experiment 1 included an additional working memory load due to withholding the divided attention task response throughout the main retrieval task, Experiment 2 (N=44) replicated the main findings while eliminating this load by having participants respond to the secondary task before the retrieval response. Together, these findings show that continuous report reveals robust costs of divided attention on LTM retrieval.
Year:
2025
Episodic memories are segmented. This study explores the dual role of prior knowledge in event segmentation, hypothesizing that knowledge leads to coarser segmentation when experiences align with it, and finer segmentation when they do not. Participants underwent a training phase to learn sequences of images, followed by a segmentation phase where they identified event boundaries, and finally, a surprise memory test. Our findings reveal that trained participants segmented less at expected transitions compared to untrained participants. Conversely, when sequences were disrupted, trained participants exhibited more segmentation. Interestingly, the segmentation differences diminished over time, suggesting that participants updated their expectations based on new experiences. Additionally, our study dissociated the roles of prediction errors and context switches in event segmentation. While prediction errors alone would predict larger differences between trained and untrained groups, we observed high segmentation rates in the trained group, even for expected transitions, underscoring the importance of contextual shifts. In conclusion, our results highlight that knowledge is a double-edged sword in event segmentation, promoting both coarser and finer segmentation. These findings provide insights into how the mind navigates and organizes experiences, emphasizing the nuanced interplay between knowledge, perception, and memory.
Year:
2025
Most daily tasks require frequent information exchange between working memory (WM) and long-term memory (LTM). However, the factors that modulate the reactivation of LTMs in WM remain to be explored. Here, we tested the effects of anticipated perceptual interference (in Experiment 1) and anticipated dual task interference (in Experiment 2) on reactivation of LTMs in WM using contralateral delay activity (CDA) in the EEG. On each trial, participants saw one or three previously studied or novel objects that were tested after a brief retention interval. In half of the blocks, the retention contained perceptual distractors (Experiment 1) or dual task interference (Experiment 2). In both experiments, the CDA prior to the interference period was similar regardless of the anticipated interference. Additionally, we measured contralateral alpha power suppression as an index of spatial attention. We found that spatial attention allocated to the memorized objects’ locations also was not modulated with anticipated interference. These findings indicate that anticipating interference does not influence the extent to which WM is engaged in representing information stored in LTM.
Year:
2025
Our everyday experiences unfold continuously, yet we naturally segment them into distinct memory units—a phenomenon known as event segmentation. While event segmentation is a well-explored topic, its underlying mechanisms have remained a subject of debate. In this study, we address this debate by comparing the two contrasting theories of event segmentation: prediction error and contextual stability. In two experiments, we manipulated contextual stability while keeping prediction error constant. Event segmentation was more prominent when there were stable contexts (Experiment 2) compared to prediction errors among unstable contexts (Experiment 1). This stronger segmentation took place despite signs of a smaller prediction error in Experiment 2. We conclude that contextual stability plays a pivotal role in driving event segmentation, highlighting its dominance over the role of prediction errors. Our study sheds new light on the mechanisms behind how our minds divide continuous experiences into meaningful memory units.
Year:
2025
Year:
2025
Year:
2025
Year:
2025
Year:
2024
Year:
2023
Collaborators (3)
Daniel Schneider
Leibniz-Institut für Arbeitsforschung an der TU Dortmund
Keisuke Fukuda
University of Toronto Mississauga
Christian Olivers
Full Professor of Visual Cognition
University of Amsterdam
Make your next move.
Prepare your application
Review drafts before you send them.