Model Of Episodic Memory Research Articles

True and false recognition in MINERVA2: Integrating fuzzy-trace theory and computational memory modeling.

Previous research suggests that the MINERVA2 model can capture basic Deese/Roediger/McDermott (DRM) false recognition findings with either randomized representations or distributional semantic representations. In the current article, we extended this line of research by showing that MINERVA2 can accommodate not only basic DRM recognition findings but also the effects of various theory-driven manipulations. Importantly, we incorporated two assumptions of fuzzy-trace theory into MINERVA2: the verbatim-gist distinction and hierarchies of gist. To implement the verbatim-gist distinction, we represented local gist traces with distributional semantic vectors and verbatim traces with holographic word-form vectors. With separate representations incorporated, MINERVA2 successfully simulated a wide range of empirical effects in the DRM illusion, as well as remember/know and source judgments. To incorporate hierarchies of gist into the framework, we added an assumption that an item's storage quality depends on its semantic similarity to the preceding item. This accommodated the effect of global gist beyond that of local gist and solved the problem of storage independence in multitrace models of episodic memory. Our findings provided extensive evidence that MINERVA2 is a viable candidate for scalable modeling of the DRM illusion and strengthened the connection between computational modeling and substantive theories of false memory. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

An embedded computational framework of memory: The critical role of representations in veridical and false recall predictions.

Human memory is reconstructive and thus fundamentally imperfect. One of its critical flaws is false recall-the erroneous recollection of unstudied items. Despite its significant implications, false recall poses a challenge for existing computational models of serial recall, which struggle to provide item-specific predictions. Across six experiments, each involving 100 young adults, we address this issue using the Embedded Computational Framework of Memory (eCFM) that integrates existing accounts of semantic and episodic memory. While the framework provides a comprehensive account of memory processing, its innovation lies in the inclusion of a comprehensive lexicon of word knowledge derived from distributional semantic models. By integrating a lexicon that captures orthographic, phonological, and semantic relationships within an episodic memory model, the eCFM successfully accounts for patterns of veridical serial recall (e.g., proportion correct, intralist errors, omissions) while also capturing false recall (e.g., extralist errors including both critical lures and non-critical lures). We demonstrate the model's capabilities through simulations applied to six experiments, with lists of words (Experiments 1A, 1B, 2A, and 2B) and non-words (Experiments 3A and 3B) that are either related or unrelated semantically (Experiments 1A and 1B), phonologically (Experiments 2A and 2B), or orthographically (Experiments 3A and 3B). This approach fills a computational gap in modelling serial recall and underscores the importance of integrating traditionally separate areas of semantic and episodic memory to provide more precise predictions and holistic memory models.

A neural index reflecting the amount of cognitive resources available during memory encoding: A model-based approach.

Humans have a limited amount of cognitive resources to process various cognitive operations at a given moment. The Source of Activation Confusion model of episodic memory proposes that resources are consumed during each processing, and once depleted, they need time to recover gradually. This has been supported by a series of behavioral findings in the past. However, the neural substrate of the resources is not known. In the present study, over an existing electroencephalogram data set of a free recall task (Kahana et al., 2022), we provided a neural index reflecting the amount of cognitive resources available for forming new memory traces. Unique to our approach, we obtained the neural index not through correlating neural patterns with behavior outcomes or experimental conditions, but by demonstrating its alignment with a latent quantity of cognitive resources inferred from the Source of Activation Confusion model. In addition, we showed that the identified neural index can be used to propose novel hypothesis regarding other long-term memory phenomena. Specifically, we found that according to the neural index, neural encoding patterns for subsequently recalled items correspond to greater available cognitive resources compared with those for subsequently unrecalled items. This provides a mechanistic account for the long-established subsequent memory effects (i.e., differential neural encoding patterns between subsequently recalled vs. subsequently unrecalled items), which has been previously associated with attention, fatigue, and properties of the stimuli. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

A non-Hebbian code for episodic memory.

Hebbian plasticity has long dominated neurobiological models of memory formation. Yet, plasticity rules operating on one-shot episodic memory timescales rarely depend on both pre- and postsynaptic spiking, challenging Hebbian theory in this crucial regime. Here, we present an episodic memory model governed by a simpler rule depending only on presynaptic activity. We show that this rule, capitalizing on high-dimensional neural activity with restricted transitions, naturally stores episodes as paths through complex state spaces like those underlying a world model. The resulting memory traces, which we term path vectors, are highly expressive and decodable with an odor-tracking algorithm. We show that path vectors are robust alternatives to Hebbian traces, support one-shot sequential and associative recall, along with policy learning, and shed light on specific hippocampal plasticity rules. Thus, non-Hebbian plasticity is sufficient for flexible memory and learning and well-suited to encode episodes and policies as paths through a world model.

Disentangling emotional source memory: a mega-analysis on the effects of emotion on item-context binding in episodic long-term memory.

It has long been known that highly arousing emotional single items are better recollected than low arousing neutral items. Despite the robustness of this memory advantage, emotional arousing events may not always promote the retrieval of source details (i.e., source memory) or associated neutral information. To shed more light on these effects, we pooled data from seven different studies (N = 333) to investigate the role of emotion on item-context binding in episodic memory, as well as potential interacting factors (e.g., encoding instructions, type of retrieval task, or acute stress). In all studies, participants incidentally encoded common neutral objects (i.e., items), overlaid on different pleasant, neutral, or unpleasant background scenes (i.e., contexts). One week later, the encoded objects were presented intermixed with new ones and memory for item and source contextual details was tested, also considering the contribution of recollection and familiarity-based processes. Linear mixed models revealed a recollection-based retrieval advantage for unpleasant and pleasant source contextual details compared to neutral ones. Bayes hypothesis-testing analysis further indicated decisive evidence in favor of a relevant role of emotional arousal and recollection in source contextual memory. Regarding item memory, linear mixed models revealed enhanced recollection-based memory for items encoded in pleasant contexts compared to their neutral and unpleasant counterparts. However, Bayes analysis revealed strong to moderate evidence for models without affective category (or its interactions), indicating that the affective category of contexts in which objects were paired during encoding had little influence on item memory performance. The present results are discussed in relation to existing evidence and current neurobiological models of emotional episodic memory by also emphasizing the role of predictive processing as a useful conceptual framework to understand the effects of emotion on memory for source details and associated neutral information.

Pattern integration and differentiation: dual process model of episodic memory

Abstract The role of precise timing in episodic memory remains obscure. We showed 139 participants episodes consisting of objects, and tested subsequent memory for the precise timing and order of the objects and episodes. Temporal compression of the episode enhanced memory for relative but not absolute timing of the objects’ presentation and their order. Conversely, temporal expansion between neighboring episodes was associated with successful memory for episode order. fMRI in 36 participants revealed that temporal compression of the episode was associated with more similar activation patterns within episodes in several brain regions including the posterior hippocampus. On the other hand, the activation pattern in the anterior hippocampus and other brain regions became more unique with temporal expansion between neighboring episodes. We propose that human episodic memory relies on two fundamentally opposite processes; pattern integration helps strengthen the relationship between the items that make up an episode, and pattern differentiation keeps different episodes apart.

The virtual habits underlying the behavioural hallmarks of alloanimal episodic memory: a Peircean model

Abstract Experimental studies of animal episodic memory test the behavioural expression of a what-where-when ‘memory content’. This is to say, autobiographical memory in some species (mainly birds, rodents, and great apes) has been understood on the basis of how the subject’s actions demonstrate their long-term awareness of already enacted and yet-to-be enacted scenarios. However, such behavioural focus has come at the expense of discussing the interdependent role of two equally important episodic hallmarks: ‘memory structure’ (mental representations), and ‘memory flexibility’ (adaptive choice-making). This paper provides a semiotic characterization of those less discussed hallmarks, in terms of the Peircean concept of virtual habit or the mental enactment of non-present episodes. It is argued that the behavioural expression of a ‘memory content’ (as object in a process of semiosis) cannot be accounted for without understanding the phenomenological or experiential support of memory structure (as a representamen), and the teleological or purposeful self-control of memory flexibility (as interpretant). Based on this, a new cognitive model of animal episodic memory is proposed, consisting of interrelated modalities called Virtualiter, Habitualiter, and Actualiter. Capitalizing on multispecies evidence, possible applications and new directions of the model are suggested.

Elements of episodic memory: insights from artificial agents.

Many recent artificial intelligence (AI) systems take inspiration from biological episodic memory. Here, we ask how these 'episodic-inspired' AI systems might inform our understanding of biological episodic memory. We discuss work showing that these systems implement some key features of episodic memory while differing in important respects and appear to enjoy behavioural advantages in the domains of strategic decision-making, fast learning, navigation, exploration and acting over temporal distance. We propose that these systems could be used to evaluate competing theories of episodic memory's operations and function. However, further work is needed to validate them as models of episodic memory and isolate the contributions of their memory systems to their behaviour. More immediately, we propose that these systems have a role to play in directing episodic memory research by highlighting novel or neglected hypotheses as pursuit-worthy. In this vein, we propose that the evidence reviewed here highlights two pursuit-worthy hypotheses about episodic memory's function: that it plays a role in planning that is independent of future-oriented simulation, and that it is adaptive in virtue of its contributions to fast learning in novel, sparse-reward environments. This article is part of the theme issue 'Elements of episodic memory: lessons from 40 years of research'.

Synthesizing the temporal self: robotic models of episodic and autobiographical memory.

Episodic memories are experienced as belonging to a self that persists in time. We review evidence concerning the nature of human episodic memory and of the sense of self and how these emerge during development, proposing that the younger child experiences a persistent self that supports a subjective experience of remembering. We then explore recent research in cognitive architectures for robotics that has investigated the possibility of forms of synthetic episodic and autobiographical memory. We show that recent advances in generative modeling can support an understanding of the emergence of self and of episodic memory, and that cognitive architectures which include a language capacity are showing progress towards the construction of a narrative self with autobiographical memory capabilities for robots. We conclude by considering the prospects for a more complete model of mental time travel in robotics and the implications of this modeling work for understanding human episodic memory and the self in time. This article is part of the theme issue 'Elements of episodic memory: lessons from 40 years of research'.

Scene construction processes in the anterior hippocampus during temporal episodic memory retrieval.

Although the hippocampus has been implicated in both the temporal organization of memories and association of scene elements, some theoretical accountsposit that the role of the hippocampus in episodic memory is largely atemporal. In this study, we set out to explore this discrepancy by identifying hippocampal activity patterns related to scene construction while participants performed a temporal order memory task. Participants in the fMRI scanner were shown a sequence of photographs, each consisting of a central object and a contextual background scene. On each retrieval trial, participants were shown a pair of the original photographs (FULL), objects from the scenes without the background (OBJ), or background contexts without the main foreground object (BACK). In the temporal order judgment (TOJ) task, participants judged the temporal order of the pair of scenes; in the Viewing trials, two identical scenes were shown without any task. First, we found that the anterior hippocampus-particularly the CA1 and subiculum-showed similar patterns of activation between the BACK and OBJ conditions, suggesting that scene construction occurred spontaneously during both TOJ and Viewing. Furthermore, neural markers of scene construction in the anterior hippocampus did not apply to incorrect trials, showing that successful temporal memory retrieval was functionally linked to scene construction. In the cortex, time-processing areas, such as the supplementary motor area and the precuneus, and scene-processing areas, such as the parahippocampal cortex, were activated and functionally connected with the hippocampus. Together, these results support the view that the hippocampus is concurrently involved in scene construction and temporal organization of memory and propose a model of hippocampal episodic memory that takes both processes into account.

Music Therapy Enhances Episodic Memory in Alzheimer's and Mixed Dementia: A Double-Blind Randomized Controlled Trial.

Objective: This study aimed to assess whether a music therapy (MT) intervention could improve memory decline in older adults with and without cognitive impairment. A dual-retrieval model of episodic memory was employed to estimate memory processes. Methods: Forty-three older adults with a mean age of 76.49 years (n = 25 with Alzheimer's disease (AD) and mixed dementia, and n = 18 healthy older adults) were randomly selected for the experimental and control groups. The study design was a double-blind randomized controlled clinical trial and a certified music therapist delivered the MT. The primary outcomes were measures of working memory, episodic memory, and autobiographical memory. Results: In the primary outcome measures, MT effects were restricted to episodic memory measures like the Figure Memory Test of the BCSB and the Speech and Sing Memory Test. In both tests, the experimental group improved from pre-test to post-test in delayed recall, but in the Speech and Sing Memory Test, the improvement was restricted to the AD and mixed dementia group. MT had no effects on the secondary outcome measures. Conclusions: These findings suggest that a structured MT intervention can be promising for rehabilitating episodic memory in older adults with dementia.

30 Exploring the Differential Importance of Modifiable Fitness Variables on Cognitive Performance in Older Adults

Objective:To identify the relative contributions and importance of modifiable fitness and demographic variables to cognitive performance in a cohort of healthy older adults.Participants and Methods:Metrics of modifiable fitness (gait speed, respiratory function, grip strength, and body mass index (BMI)) and cognition (executive function, episodic memory, and processing speed) were assessed in 619 older adults from the Health and Retirement Study 2016 wave (mean age = 74.9, sd = 6.9; mean education = 13.4 years, sd = 2.6; 42% female). General linear models were employed to assess the contribution of modifiable fitness variables in predicting three domains of cognition: executive function, episodic memory, and processing speed. Demographics (age, sex, education, time between appointments, and a chronic disease score) were entered as covariates for each model. Relative importance metrics were computed for all variables in each model using Lindeman, Merenda, and Gold (lmg) analysis, a technique which decomposes a given model’s explained variance to describe the average contribution of each predictor variable, independent of its position in the linear model.Results:When all variables were entered into the general linear model, demographic and modifiable fitness variables explained 35%, 24%, and 26% of the variance in executive function, episodic memory, and processing speed, respectively (all three models were significant, p <0.001). Age, education, respiratory function, and walking speed had higher relative importance values (all lmgs > 1.8) compared to BMI, grip strength, and other covariates in all three models (all lmgs < 1.3). Gender was also relatively important in the executive function (lmg = 4.2) and episodic memory models (lmg = 5.0). Of the modifiable fitness variables, walking speed and respiratory function had the greatest lmg values (5.8 and 6.4 respectively) in the executive function model, similar to demographic variables age (lmg = 6.0) and education (lmg = 8.9). When demographic variables were entered as covariates, modifiable fitness variables collectively accounted for an additional 9.7%, 6.3%, and 6.0% variance in the executive function, episodic memory, and processing speed models respectively (all three models were significant, p <0.001).Conclusions:Our findings indicate that walking speed and respiratory function are of similar importance compared to “traditional” demographic variables such as age and education in predicting cognitive performance in a cohort of healthy older adults. Moreover, modifiable fitness variables accounted for unique variance in executive function, episodic memory, and processing speed after accounting for age and education. Modifiable fitness variables explained the most unique variance in executive function. These results extend the current literature by demonstrating that modifiable fitness variables, even when assessed with brief and relatively coarse measures of physical performance, may be useful in predicting cognitive function. Moreover, the results highlight the need to assess metrics of cognitive reserve, such as education, as well as modifiable fitness variables and their respective roles in accounting for cognitive performance. The data further suggest that relative contributions of physical performance metrics may vary by cognitive domain in healthy older adults.

Spiking neural network-based computational modeling of episodic memory

In this research article, a spiking neural network-based simulation of the hippocampus is performed to model the functionalities of episodic memory. The purpose of the simulation is to find a computational model through the biological architecture of the hippocampus and correct values for their architectural biological parameters to support the episodic memory functionalities. The episodic store of the model is represented by the collection of events, where each event is further subdivided into coactive activities of experience. The model has tried to mimic the three functionalities of episodic memory, which are pattern separation, pattern association, and their recallings. In pattern separation model used the dentate biological connectivity to generate almost different output patterns corresponding to similar input patterns to reduce interference between two similar memory traces so that ambiguity can be reduced during recalling. In pattern association, an STDP based event encoding and forgetting mechanism are used to mimic the encoding function of the CA3 region in which the coactive activities get associated with each other. A decoder is proposed based on CA1, which can answer the stored event related queries. Along with these functionalities model also supports recalling and encoding based forgetting. Experimental work is performed on the model for the given set of events to check for the pattern separation efficiency, pattern completion efficiency and to check the capability of decoding the answer. An empirical analysis of the results is done and compared with the SMRITI model of episodic memory.

Neural temporal context reinstatement of event structure during memory recall.

The transformation of experiences into meaningful events and memories is intertwined with the notion of time. Temporal perception can influence, and be influenced by, segmenting continuous experience into meaningful events. Episodic memories formed from these events become associated with temporal information as well. However, it is less clear how temporal perception contributes to structuring events and organizing memory: whether it plays a more active or passive role, and whether this temporal information is encoded initially during perception or influenced by retrieval processes. To address these questions, we examined how event segmentation influences temporal representations during initial perception and memory retrieval, without testing temporal information explicitly. Using a neural measure of temporal context extracted from scalp electroencephalography in human participants (N = 170), we found reduced temporal context similarity between studied items separated by an event boundary when compared to items from the same event. Furthermore, while participants freely recalled list items, neural activity reflected reinstatement of temporal context representations from the study phase, including temporal disruption. A computational model of episodic memory, the context maintenance and retrieval (CMR) model, predicted these results, and made novel predictions regarding the influence of temporal disruption on recall order. These findings implicate the impact of event structure on memory organization via temporal representations, underscoring the role of temporal information in event segmentation and episodic memory. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Amyloid and tau deposition influences cognitive and functional decline in Down Syndrome

Abstract BackgroundDown Syndrome (DS) is the most common neurodevelopmental disorder caused by the presence of trisomy 21 (1:800 to 1:1000 live births worldwide). The extra copy of chromosome 21 is associated with a 4‐5 fold overexpression of the amyloid precursor protein (APP) gene, increased accumulation of cerebral beta‐amyloid (Aβ) in the brain, subsequent neurofibrillary tangle (NFT) formation and neurodegeneration. This study investigates whether tau has (i) an independent effect from amyloid on changes in cognitive and functional performance and (ii) a synergistic relationship with amyloid in the exacerbation of decline in aging DS.Method105 participants with DS underwent baseline PET AV‐1451 and PET PiB scans to quantify tau deposition in Braak regions II‐VI and Striatum and amyloid‐β status respectively. A longitudinal cognitive testing battery was performed consistent with previous research in the same study cohort, with all participants completing at least two timepoints, and 26 completing three timepoints, with an average of 17 months between study visits. Linear Mixed Effects models were implemented to assess how tau and amyloid deposition are related to change over three time points.ResultThree cognitive outcome measures showed a significant effect of time: the Free + Cued Recall (t = ‐6.09, p < 0.01), Cued Intrusions (t = 3.19, p < 0.01) and NTG (z = 5.33, p = 0) variables. Tau SUVR was a significant independent predictor of cognitive and functional change. The three‐way interaction between time, PiB status and tau was significant in the models of episodic memory and visuospatial cognition (Figures 1‐3).ConclusionOur results contribute to an emerging framework in which the elevated risk of developing dementia involves mechanisms associated with both amyloid‐β and tau aggregation. This is supported by our finding that when including the two‐way interactions between amyloid status and time and regional tau and time in the same model, tau seems to be a more significant predictor of cognitive change over time than amyloid status. Our results suggest a synergistic relationship between PiB status and tau as predictors of change in memory and visuospatial cognition.

Impacts of cerebral small vessel disease on global and domain‐specific cognition

Abstract BackgroundThe 2021 Alzheimer’s Association International Conference press release forecasted global dementia cases to triple by 2050, due to the major contribution of increases in vascular risk factors. The Global Vascular Risk Score (GVRS) was developed to predict vascular events using a point‐based index for risk factor variables, though has demonstrated to be a significant predictor of global and domain‐specific cognitive performance in a multi‐ethnic cohort (Rundek, T. et al. Global Vascular Risk Score and CAIDE Dementia Risk Score Predict Cognitive Function in the Northern Manhattan Study. J Alzheimers Dis 73, 1221‐1231 (2020)). However, more than half of the variance in global and domain‐specific cognitive performance is still unexplained by GVRS variables. We sought to explain additional variance in cognitive performance by addition of MRI biomarkers of cerebral small vessel disease to the GVRS model.MethodWe included 1290 subjects from the multiethnic Northern Manhattan Study population (age 64+/‐8 years, 40% women, 66% Hispanic, 17% non‐Hispanic Black, 15% non‐Hispanic White, education 10+/‐5 years). Multivariate linear regression models were employed to assess the added value of the cerebral small vessel disease MRI measures (white matter hyperintensity volume, silent brain infarcts, cerebral microbleeds, and perivascular spaces) beyond GVRS variables in explaining more variance in global and domain‐specific (episodic memory, executive function, semantic memory, and processing speed) cognitive performance. The explained variance (R2) of the GVRS model was compared to the proposed model with added imaging markers to determine their predictive performance. Statistical significance was determined using F‐statistic difference test between the models.ResultThe explained variance (R2) and F‐statistic for models of global and domain‐specific cognition are in the table. Addition of MRI measures significantly improved the model for global cognition (F(4, 1108) = 2.68, p < 0.05) and domain‐specific cognitive performance models of episodic memory (F(4, 1096) = 3.34, p < 0.01) and processing speed (F(4, 1096) = 2.40, p < 0.05).ConclusionMRI measures of cerebral small vessel disease explain more variance in global cognition, episodic memory, and processing speed than GVRS factors alone. However, most of the variance in cognitive performance remains unexplained.

Visual Episodic Memory-based Exploration

In humans, intrinsic motivation is an important mechanism for open-ended cognitive development; in robots, it has been shown to be valuable for exploration. An important aspect of human cognitive development is episodic memory which enables both the recollection of events from the past and the projection of subjective future. This paper explores the use of visual episodic memory as a source of intrinsic motivation for robotic exploration problems. Using a convolutional recurrent neural network autoencoder, the agent learns an efficient representation for spatiotemporal features such that accurate sequence prediction can only happen once spatiotemporal features have been learned. Structural similarity between ground truth and autoencoder generated images is used as an intrinsic motivation signal to guide exploration. Our proposed episodic memory model also implicitly accounts for the agent's actions, motivating the robot to seek new interactive experiences rather than just areas that are visually dissimilar. When guiding robotic exploration, our proposed method outperforms the Curiosity-driven Variational Autoencoder (CVAE) at finding dynamic anomalies.

The role of rehearsal and reminding in the recall of categorized word lists

Most theories of free recall emphasize the importance of retrieval in explaining temporal and semantic regularities in recall; rehearsal mechanisms are often absent or limit rehearsal to a subset of what was last rehearsed. However, in three experiments using the overt rehearsal method, we show clear evidence that just-presented items act as retrieval cues during encoding (study-phase retrieval) with prior related items rehearsed despite well over a dozen intervening items. Experiment 1 examined free recall of categorized and uncategorized lists of 32 words. In Experiments 2 and 3, we presented categorized lists of 24, 48, and 64 words for free recall or cued recall, with the category exemplars blocked in successive list positions (Experiment 2) or randomized throughout the list (Experiment 3). The probability of rehearsing a prior word was affected by its semantic similarity to the just-presented item, and the frequency and recency of its prior rehearsals. These rehearsal data suggest alternative interpretations to well-known recall phenomena. With randomized designs, the serial position curves were reinterpreted by when words were last rehearsed (which contributed to the list length effects), and semantic clustering and temporal contiguity effects at output were reinterpreted by whether words were co-rehearsed during study. The contrast with the blocked designs suggests that recall is sensitive to the relative (not absolute) recency of targeted list items. We discuss the benefits of incorporating rehearsal machinery into computational models of episodic memory, and suggest that the same retrieval processes that generate the recalls are used to generate the rehearsals.

Assessing episodic memory in rodents using spontaneous object recognition tasks.

Models of episodic memory are successfully established using spontaneous object recognition tasks in rodents. In this review, we present behavioral techniques devised to investigate this type of memory, emphasizing methods based on associations of places and temporal order of items explored by rats and mice. We also provide a review on the areas and circuitry of the medial temporal lobe underlying episodic-like memory, considering that a large number of neurobiology data derived from these protocols. Although spontaneous recognition tasks are commonplace in this field, there is need for careful evaluation of factors affecting animal performance. Such as the ongoing development of tools for investigating the neural basis of memory, efforts should be put in the refinement of experimental designs, in order to provide reliable behavioral evidence of this complex mnemonic system.

Toward a statistical foundation for brain signatures of behavioral outcome

Abstract BackgroundThe “brain signature of cognition” concept has garnered interest as a data‐driven, exploratory approach to better understand key brain regions involved in specific cognitive functions, with the potential to maximally characterize brain substrates of clinical outcomes. However, to be a robust brain phenotype, the signature approach requires a statistical foundation showing that model performance replicates across a variety of cohorts. Here, we outline a procedure that provides this foundation for a signature models of episodic memory and extend this to a measure of functional everyday memory.MethodUniversity of California, Davis (UCD) and ADNI provided imaging and memory data. For cognitive and ECog memory, we derived regional brain gray matter (GM) cortical thickness (CT) measures, testing for replicability (Fig. 1). Signature model generation used two steps in each cohort: 1) Randomly select 40 discovery subsets of size N = 400. In each subset, use voxelwise regression to compute the CT signature regions for outcome at three t‐values of association. 2) Compile frequency maps of region selection across all 40 discovery sets and designate regions with selection frequency >70% as “consensus” signature masks for each cohort. Model validation used regressions of outcomes on consensus masks: 1) Compare adjusted R2 fits of both consensus signature models in 40 discovery sets in each cohort (thus, symmetrically compare same and oppositely generated models). 2) In each full cohort, compare signature model fits with competing models.ResultSpatial replications produced strongly convergent consensus signature regions derived from UCD and ADNI (Fig. 2). Consensus model fits were highly correlated in 40 random subsets of each cohort (Fig. 3 top) indicating high replicability, with tight (∼80%) ratios of within/outside cohort model fits (Fig. 3 bottom), indicating low bias. In comparisons over each full cohort, signature models outperformed other models with one exception (Fig. 4).ConclusionMultiple random model generations, followed by consensus selection of regional brain substrates, produced signature models that both consistently replicated model fits to memory and outperformed other commonly used measures. Results support consensus signature models as robust biomarkers of cognition and everyday function across imaging cohorts with differing demographic composition (see Table).