BACKGROUND:Impaired memory function is a frequent yet understudied symptom in kidney transplant recipients. In part, this knowledge gap reflects the lack of scalable, sensitive, and low-burden tools for quantifying memory function in large clinical populations. The aim of this study is to evaluate a brief, remote memory assessment and to examine memory function and its clinical correlates in kidney transplant recipients compared with healthy controls. METHODS:In this cross-sectional study, we demonstrate a remote, minimally burdensome memory screener, the Seattle-Groningen Memory Assessment, which estimates a patient's speed of forgetting from paired-associate learning. Participants aged 22-86 years from a large transplant cohort completed an eight-minute online memory test. Memory performance was compared between kidney transplant recipients (n = 556) and kidney donors (n = 408). Associations with demographic, clinical, and physiological variables were examined using regression analyses. RESULTS:Here, we show that the memory score derived from an eight-minute session is a reliable and accurate measure of an individual's ability for long-term retention. Kidney transplant recipients show more forgetting than donors. Memory scores are sensitive to demographic factors, including age and education level, and are associated with self-reported sleep quality, fatigue, and health-related quality of life. On the physiological level, more forgetting in recipients is linked to higher monocyte, neutrophil, reticulocyte, and white blood cell counts, as well as lower ferritin and greater iron deficiency. CONCLUSIONS:This work highlights the potential of computational memory assessment as a minimally burdensome and reliable tool for detecting cognitive impairment in complex clinical populations. Such tools may enable scalable monitoring of cognitive health and improve the detection of subtle cognitive changes relevant for disease progression and treatment evaluation.

Performance on memory tasks varies systematically between individuals and within individuals over time. However, computational models of memory are often fitted at the group level or summarise performance in a single parameter, potentially obscuring the distinct cognitive processes that give rise to this variation. Here we present a likelihood-based procedure for fitting the multi-parameter ACT-R declarative memory model at the level of the individual. Using this method, we jointly estimate five participant-level or idiographic parameters, each capturing a distinct component of memory performance, together with item-level variation in memorability. A parameter recovery study on synthetic data shows that these participant- and item-level parameters are jointly identifiable from realistic amounts of data on a typical spaced repetition memory task, requiring only about 40-90 observations of response accuracy and speed per participant. We demonstrate the method on a dataset of individuals with mild cognitive impairment and age-matched healthy controls, finding that mild cognitive impairment is marked by differences in three parameters signifying accelerated forgetting, less efficient memory retrieval, and general psychomotor slowing. These results show that the method can successfully decompose observed performance on a memory task into separable, individually interpretable components. More broadly, this work contributes to the development of fine-grained idiographic memory models that can capture dynamics in memory performance both between and within individuals, with applications in experimental research and applied settings, such as computational phenotyping for clinical diagnosis.

Objective: Extensive evidence shows that performance during sustained attention tasks declines over time and mind-wandering (MW) frequency increases with time on task. It is not clear, however, what types of interventions can address these issues and if those interventions can simultaneously mitigate both increased MW frequency and performance decline. Method: This study examined the effects of two common interventions, taking a break (rest-break) vs. changing to a different task (task-switch) during the Sustained Attention to Response Task (SART). Traditional reaction time and accuracy measures were computed, as well as, parameters from a drift-diffusion model (DDM), which allowed for a closer look at potential cognitive mechanisms. In addition, the effectiveness of the interventions was evaluated across different mental states (on-task vs. off-task). Results: Behaviorally, both interventions preserved performance compared to the no-break group. For the DDM parameters, both interventions maintained stable no-go drift rates, and the task-switch intervention additionally showed an increase in boundary separation. However, neither intervention reduced rates of MW. Additionally, the effects of interventions were more pronounced when participants reported being on-task. Conclusions: These findings suggest that short breaks can help sustain performance, yet do not necessarily halt MW.

In this commentary we extend the Shiffrin et al. (2026) analysis of illusions in understanding via a focus on the general class of extrapolatory illusions. This provides a generalization across multiple illusions they identified while also expanding the analysis to additional illusions worth considering.

Transformer-based models of human behavior (e.g., the Centaur model by Binz, et al., 2025) posit to be domain general computational models of human behavior. The claim of domain-generality is by virtue of the supposed capability to predict and simulate human behavior across a vast range of cognitive and perceptual domains. Further, it is argued that this degree of performance places such models on a path toward general, unified theories of cognition (Newell, 1990). We contest this characterization. We propose the Domain-Generality Thesis: A computational model is domain-general if and only if it performs well across a structurally distinct set of tasks. While transformer-based models of human behavior achieve impressive statistical breadth, we demonstrate that, by example, they fail this structural criterion, conflating parametric variations of a single task with genuine cognitive diversity. We construct an argument that denies the domain-generality of transformer-based models of human behavior and thus denies the purported status as a start on the path towards general, unified theories of cognition.

We propose a unified approach to integrating comprehensive metacognition into the Common Model of Cognition (CMC), an abstract characterization of the structure and processing required by a cognitive architecture for human-like minds. Our central claim is that metacognition requires no dedicated metacognitive reasoning modules: it reuses the CMC’s existing cognitive capabilities, adding only explicit working-memory representations of an agent’s cognitive state, capabilities, and processes, together with episodic memory for historical self-knowledge. We show how this approach supports all phases of metacognition and the diverse types and sources of knowledge that contribute to metareasoning. Notably, these mechanisms already exist, in whole or in part, in Soar, Sigma, and ACT-R.

Despite the importance of memories in everyday life and the progress made in understanding how they are encoded and retrieved, the neural processes by which declarative memories are maintained or forgotten remain elusive. Part of the problem is that it is empirically difficult to measure the rate at which memories fade, even between repeated presentations of the source of the memory. Without such a ground-truth measure, it is hard to identify the corresponding neural correlates. This study addresses this problem by comparing individual patterns of functional connectivity against behavioral differences in forgetting speed derived from computational phenotyping. Specifically, the individual-specific values of the speed of forgetting in long-term memory (LTM) were estimated for 33 participants using a formal model fit to accuracy and response time data from an adaptive paired-associate learning task. Individual speeds of forgetting were then used to examine participant-specific patterns of resting-state fMRI connectivity, using machine learning techniques to identify the most predictive and generalizable features. Our results show that individual speeds of forgetting are associated with resting-state connectivity within the default mode network (DMN) as well as between the DMN and cortical sensory areas. Cross-validation showed that individual speeds of forgetting were predicted with high accuracy (r = .77) from these connectivity patterns alone. These results support the view that DMN activity and the associated sensory regions are actively involved in maintaining memories and preventing their decline, a view that can be seen as evidence for the hypothesis that forgetting is a result of storage degradation, rather than of retrieval failure.

We offer a comment on the Centaur (Binz et al., 2025) transformer-based model of human behavior. In particular, Centaur was cast as a path towards unified theories of cognition. We offer a counter claim with supporting argument: Centaur is a path divergent from unified theories of cognition, one that moves towards a unified model of behavior sans cognition.

A beginning is made at mapping four neural theories of consciousness onto the Common Model of Cognition. This highlights how the four jointly depend on recurrent local modules plus a cognitive cycle operating on a global working memory with complex states, and reveals how an existing integrative view of consciousness from a neural perspective aligns with the Com-mon Model.

The Common Model of Cognition (CMC) provides an abstract characterization of the structure and processing required by a cognitive architecture for human-like minds. We propose a unified approach to integrating metacognition within the CMC. We propose that metacognition involves reasoning over explicit representations of an agent's cognitive capabilities and processes in working memory. Our proposal exploits the existing cognitive capabilities of the CMC, making minimal extensions in the structure and information available within working memory. We provide examples of metacognition within our proposal.

Researchers agree input from the basal ganglia (BG) to the prefrontal cortex (PFC) plays an important role in cognition, but they disagree on its computational properties. Theoretical models characterize the majority of BG input as either direct (directly transmitting information to the PFC), or modulatory (indirectly influencing PFC activity through the gating of signals from other cortical areas). To determine the computational nature of these BG-PFC inputs in cognition, we tested three alternative connectivity configurations (Direct, Modulatory, and Mixed) within a large-scale cognitive architecture. This architecture, the Common Model of Cognition, has been independently validated using fMRI data from the Human Connectome Project (HCP). The Direct model reflected the standard CMC configuration, featuring a bidirectional, direct connection between the BG and PFC modules. The Modulatory model removed this direct link, instead incorporating a unidirectional connection from the PFC to the BG and two modulatory pathways from the BG to the PFC that passed through other cortical modules. The Mixed model included both direct and modulatory connections. Using fMRI data from 200 HCP participants performing six cognitive tasks and one resting-state session, we applied Dynamic Causal Modeling (DCM) to estimate and compare the influence of these different BG–PFC connectivity patterns. Here, we show that for each of the six cognitive tasks and resting state, the Mixed model consistently outperformed the Direct and Modulatory models. The next best model depended on the specific cognitive task, suggesting the ability for the BG to flexibly adapt to various task demands. Taken together, the current data provide evidence for a likely set of core computations that the BG uses to differentially regulate cortical activity. Author Summary This study aimed to understand the nature of regulatory input from the basal ganglia to the prefrontal cortex in cognition using a neuro-computational approach. Three model architectures of basal ganglia connectivity were tested using fMRI data from 200 individuals during cognitive tasks and resting state sessions. Our findings indicate that the Mixed architecture, in which the basal ganglia can affect the prefrontal cortex both directly and indirectly, outperformed the Direct and Modulatory architectures across various cognitive tasks and resting state sessions. The results suggest that the basal ganglia likely uses a set of core computations to regulate cortical activity, which is present during task and task-free behavior. ### Competing Interest Statement The authors have declared no competing interest.

Posttraumatic stress is about memory; distressing and intrusive memories of a traumatic event. Although considerable focus has been on initial encoding, memory processes after the traumatic event are likely as important if not more important for facilitating resilience, natural recovery, and therapeutic recovery. In this proposed dynamic social retrieval theory (DSRT) of posttraumatic stress, ongoing retrieval and related forgetting processes shape the changing nature of the trauma memory. Retrieval events number in the hundreds, thousands, or even more, strengthening some memory traces and associations and weakening others. These retrieval events take many forms from conversations with friends and loved ones, social media interactions, intentional recall, spontaneous thoughts, cued- and un-cued reexperiencing, and avoidance of trauma reminders. We argue that the initial days, weeks, and months are important; through systems consolidation, the memory shifts to a more general, gist-like representation, incorporating social/cultural schemas and one's view of self. Better orthogonalized, separation from other related memories can further result in the reduction of trauma-related psychopathology. Clinical implications highlight the importance of adaptive retrieval in daily life, in social interactions and cultural messaging, and within the therapeutic relationship to shape the long-term nature of the traumatic memory and help the trauma survivor flourish.

Impaired memory function is a frequent yet understudied symptom in kidney transplant recipients. Here, we demonstrate a remote, minimally burdensome memory screener, the Seattle-Groningen Memory Assessment (SGMA), which estimates a patient’s speed of forgetting from paired-associate learning. We show that the memory score derived from an 8-minute session is a reliable and accurate measure of an individual’s ability for long-term retention. Kidney transplant recipients (n=556) showed more forgetting than (potential) donors (n=408). Memory scores were sensitive to demographics (age and education level), and related to self-reported sleep quality, fatigue, and health-related quality of life. On the physiological level, we were able to link more forgetting in recipients to more monocytes, neutrophils, reticulocytes, and a higher white blood cell count, as well as lower ferritin and more iron deficiency. Overall, this work highlights the potential of computational memory assessment as a minimally burdensome and reliable tool for detecting cognitive impairments in complex multimorbidity populations. The approach may be particularly valuable in research settings where detecting subtle changes in cognitive health—often missed by existing assessments—is crucial for understanding disease progression and treatment effects.

Complex skill learning depends on the joint contribution of multiple interacting systems: working memory (WM), declarative long-term memory (LTM) and reinforcement learning (RL). The present study aims to understand individual differences in the relative contributions of these systems during learning. We built four idiographic, ACT-R models of performance on the stimulus-response learning, Reinforcement Learning Working Memory task. The task consisted of short 3-image, and long 6-image, feedback-based learning blocks. A no-feedback test phase was administered after learning, with an interfering task inserted between learning and test. Our four models included two single-mechanism RL and LTM models, and two integrated RL-LTM models: (a) RL-based meta-learning, which selects RL or LTM to learn based on recent success, and (b) a parameterized RL-LTM selection model at fixed proportions independent of learning success. Each model was the best fit for some proportion of our learners (LTM: 68.7%, RL: 4.8%, Meta-RL: 13.25%, bias-RL:13.25% of participants), suggesting fundamental differences in the way individuals deploy basic learning mechanisms, even for a simple stimulus-response task. Finally, long-term declarative memory seems to be the preferred learning strategy for this task regardless of block length (3- vs 6-image blocks), as determined by the large number of subjects whose learning characteristics were best captured by the LTM only model, and a preference for LTM over RL in both of our integrated-models, owing to the strength of our idiographic approach. Individuals rely on different strategies-combination of declarative memory and procedural memory-to learn new associations. Idiographic computational models were used to capture these differences.

The detection and tracking of progressive memory impairments, particularly in the context of neurodegenerative disorders, relies predominantly on traditional neuropsychological assessment and short cognitive screening tools. These methods, however, are resource-intensive and lack the accessibility and/or the repeatability necessary for effective early identification and tracking interventions. This study addresses the critical need for reliable and efficient diagnostic tools to track and predict memory decline in clinical settings. We demonstrate that an online, remote model-based memory assessment, can identify individuals with Mild Cognitive Impairment (MCI) with an accuracy rate exceeding 84% in a single 8-minute session. Furthermore, the test can be repeated multiple times with increasing accuracy over multiple assessments. The system's ability to monitor individual memory function inexpensively and longitudinally across various materials offers a robust and repeatable alternative to the static measures currently employed. Our findings show that traditional methods to assess memory decline could be replaced by adaptive, precise, and patient-friendly online tools based on computational modeling techniques. Moreover, our findings also open avenues for the proactive management of Alzheimer's disease and other dementias, as well as sensitively tracking the effect of interventions in early disease. ### Competing Interest Statement The authors declare competing interests as follows: H.S.H., M.v.d.V., H.v.R., and A.S. are named on a preliminary patent related to the subject matter discussed in this article; M.v.d.V. and H.v.R. are employed by SlimStampen B.V., the company responsible for the development and commercialization of the AFLS (Adaptive Fact-Learning System). ### Funding Statement This study was funded by a pilot award from the Garvey Institute for Brain Health Solutions ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: IRB of the University of Washington gave ethical approval for this work I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes All data produced are available online at our osf repository