Biology subjects

Meder, B.

Publications and source records attributed to Meder, B..

18 recordsLinked to original sources

bioRxiv2026

Damaging RBM20 E-rich domain variants are not rescued by gene replacement

The promise of precision therapeutics in genetic cardiomyopathies relies on linking specific therapies to variant mechanisms. Missense variants in the cardiac splice regulator RBM20 cause a highly penetrant and arrhythmogenic dilated cardiomyopathy. Disease-causing variants in RBM20s arginine-serine rich (RS) domain act via formation of toxic gain of function cytoplasmic granules, but this is not true for a small number of clinically adjudicated pathogenic variants in its glutamate(E)-rich domain. To better define the effects of E-rich domain variants, we developed a scalable screen based on induced pluripotent stem cell (iPSC) cardiomyocyte differentiation that identified several additional damaging variants. Several of these reduced RBM20 protein abundance and stability. We therefore hypothesized that, unlike RS domain variants, these E-rich variants might be rescued by RBM20 overexpression. To test this hypothesis, we generated induced pluripotent stem cells (iPSCs) from a patient with a pathogenic E-rich domain variant (p.E913K), and confirmed reduced RBM20 protein expression in these RBM20+/p.E913K cells after differentiation to iPSC-derived cardiomyocytes (iPSC-CM, vs. engineered isogenic RBM20+/+). These iPSC-CMs also displayed aberrant transcriptional splicing, reduced contractility, increased calcium-induced calcium release, and nuclear localization of RBM20 protein, often with more than the two expected RBM20-centric splice factories. AAV-based overexpression of RBM20 reversed some, but not all of the mis-splicing events identified in RBM20+/p.E913K iPSC-CMs, and did not improve their abnormal contractility, calcium handling or supernumerary RBM20 nuclear granules. In summary, our data indicate that pathogenic E-rich domain variants reduce RBM20 protein abundance, but that their mechanism is unlikely to be explained by haploinsufficiency alone.

genetics↗

bioRxiv2026

Beyond model-free Pavlovian responding: a two-stage Pavlovian-instrumental transfer paradigm

BackgroundPavlovian responding is a core component of behavior and can be measured via Pavlovian-instrumental transfer (PIT), where Pavlovian responses bias instrumental actions. Standard single-lever PIT paradigms, which assess responses using a single-choice option, cannot dissociate the contribution of model-free versus model-based reinforcement learning. While indirect evidence suggests a role for model-free responding in single-lever PIT, the contribution of model-based strategies is unclear. It also remains unknown whether internal cognitive states, such as mind wandering, impair specifically model-based but not model-free PIT, as is theoretically expected. MethodsWe developed a novel, trial-by-trial two-stage PIT paradigm designed to computationally dissociate model-free and model-based Pavlovian responding by leveraging probabilistic state transitions and trial-wise outcome predictions. After each two-stage Pavlovian learning trial, participants performed a single-lever PIT trial as well as a query trial of explicit value judgment. Detailed task instructions were provided to support potential model-based strategies. Computational modeling was used to quantify individual learning strategies. We assessed mind-wandering questionnaires and thought probes. ResultsAnalysis of query and PIT trials revealed trial-by-trial updating of outcome expectations based on probabilistic task structure, consistent with model-based Pavlovian responding. Behavioral responses during PIT were best explained by a computational model-based reinforcement learning model. In contrast, we found little evidence for model-free Pavlovian responding. Higher levels of mind wandering were associated with reduced model-based control but did not impact model-free indices. ConclusionWe introduce a novel single-lever PIT paradigm that enables fine-grained dissociation of model-free versus model-based Pavlovian response systems. Our findings provide evidence that single-lever PIT can operate through model-based mechanisms, challenging the assumption that single-lever PIT is predominantly model-free. Our findings also indicate that internal attentional states selectively modulate model-based PIT. Given the involvement of Pavlovian responding in numerous psychiatric conditions, our paradigm offers new avenues for understanding maladaptive behavior. Author SummaryOur daily actions are often influenced by cues like the smell of food or the sound of phone notifications that signal potential rewards or losses. These Pavlovian cues can shape our instrumental behavior even though their outcomes do not depend on what we do - a process known as Pavlovian-instrumental transfer (PIT). Here we study the computational learning mechanisms that underlie such PIT effects. While it is often assumed that Pavlovian responding follows simple, automatic rules without a cognitive model of cue consequences (i.e., model-free), evidence also shows a role for cognitive anticipations in Pavlovian responding (i.e., model-based). In this study, we extend this evidence by showing that PIT responding can be driven by flexible model-based learning. We designed a task to test whether participants use model-free versus model-based strategies to guide PIT, providing detailed task instructions. Using reinforcement learning models, we found that most participants used model-based learning when forming cue-outcome associations. Importantly, peoples attention mattered: when they were more distracted and doing mind wandering, they relied less on model-based strategies. Our findings suggest that Pavlovian learning is complex, flexible, and influenced by internal mental states, opening new windows to understand decision-making problems in mental health conditions like addiction.

neuroscience↗

bioRxiv2026

Multicellular signaling and partial recovery define reverse cardiac remodeling

Heart failure results from maladaptive multicellular remodeling triggered by sustained biomechanical stress. Although mechanical unloading can promote reverse remodeling, recovery is frequently incomplete and its mechanistic basis remains unclear. Using a reversible murine pressure-overload model combined with bulk and single-nucleus transcriptomics, we demonstrate that reverse remodeling represents an actively maintained yet constrained multicellular state. Unloading improved cardiac function and partially restored extracellular matrix and metabolic programs, whereas inflammatory and mitochondrial dysfunction signatures persisted. Cardiomyocytes and endothelial cells largely re-established homeostatic transcriptional states, while fibroblasts retained activated programs that dominated residual pathology. Multicellular factor analysis delineated a reversible metabolic stress program and a persistent inflammatory-mitochondrial program coordinated across cardiac cell types. Cell-cell communication analysis identified lymphatic endothelial cells as key instructive regulators of recovery. Notably, lymphatic-derived Reelin directly suppressed pathological fetal gene activation in murine and human cardiomyocytes, uncovering a previously unrecognized lymphoangiocrine mechanism that constrains myocardial recovery in chronic heart failure.

molecular biology↗

bioRxiv2025

Sarcomeric Remodelling in Human Heart Failure unraveled by single molecule long read sequencing

Dysregulation of alternative splicing - mediated by factors such as RBM20 or SLM2 - can affect proper gene isoform control disrupts gene isoform homeostasis and underpins severe cardiomyopathy in both animal models and patients. Although innovative therapies target various sarcomeric components, the impact of isoform switching in cardiac disease remains poorly understood. Here, we applied nanopore long-read sequencing to map the full-length transcriptome of left ventricular tissue from thirteen nonfailing controls, ten patients with dilated cardiomyopathy (DCM), and ten with ischemic cardiomyopathy (ICM). Our analysis identified 78,520 transcripts, 31% of which represent novel isoforms of known genes. Notably, the transcriptomes of DCM and ICM were largely indistinguishable, indicating that end-stage heart failure is characterized by a convergent isoform landscape, irrespective of disease etiology. Among 11 prototypical sarcomere genes, 10 displayed highly significant isoform shifts (p= 5.23x10-45 - 2.89x10-200). Focusing on tropomyosin, we observed that while the predominant cardiac gene TPM1 showed moderate up-regulation of its transcript isoforms, transcripts derived from TPM3--typically expressed at lower levels in the healthy heart--were markedly increased in heart failure.

genetics↗

bioRxiv2025

RBM20-variants induce distinct calcium handling and metabolic phenotypes in patient-specific stem cell models of dilated and non-compaction cardiomyopathy

Background and aimMutations in the splice regulator RBM20 account for [~]3 % of genetic cardiomyopathies. In particular, the highly conserved RS domain is a hotspot for disease-associated mutations. Previously, mutations at same amino acid position 634 in the hotspot RS-domain were found to cause dilated cardiomyopathy (DCM) with left ventricular non-compaction (R634L) or without (R634W), but the pathophysiological mechanisms that govern the heterogeneity in phenotype presentation remained unknown. Here, we identify the molecular events caused by the distinct RBM20 mutations from DCM and left-ventricular non-compaction (LVNC) using patient-specific stem cell models. MethodsWe generated induced pluripotent stem cell-derived cardiomyocytes (iPSC-CM) of one LVNC- and two DCM-patients harboring the RBM20-mutations R634L (LVNC) or R634W (DCM). We investigated alternative splicing activity, RBM20 localization, sarcomeric regularity, cAMP level, kinase-specific phosphorylation of key Ca2+ handling enzymes, physiological cardiac functions as Ca2+ homeostasis, and metabolic activity on a patient-specific cardiomyocyte level. Force generation was analyzed in patient-specific engineered myocardial tissues. Isogenic rescue and mutation insertion lines were generated by CRISPR/Cas9 technology to analyze the direct impact of the RBM20 mutations on the cardiac phenotype. ResultsWe observed common splicing aberrations for LVNC- and DCM-CM in TTN and RYR2, RBM20 cytoplasmatic accumulation and irregular sarcomeric structure. LVNC-CM harboring the RBM20-p.R634L variant show distinct molecular, cellular and functional impairments that manifest in CAMK2D, TRDN and IMMT mis-splicing. Splicing defects in LVNC-CM correlate with elevated systolic Ca2+ and faster Ca2+ kinetics with elevated cAMP levels and PLN-hyperphosphorylation. An increased metabolic activity and mitochondrial membrane potential support the hyperactive LVNC-CM. By contrast, DCM-CM (RBM20-p.R634W) distinctly present with decreased systolic Ca2+ and increased SR Ca2+leak but unchanged Ca2+ kinetics and metabolic activity. Both mutations lead to severely reduced force of contraction in engineered myocardium. CRISPR/Cas9 gene-edited isogenic control lines of both described RBM20 mutations in LVNC and DCM demonstrated the causative nature of the two mutations and their diverging effects. Further, L-type Ca2+ channel blockade by verapamil ameliorates the Ca2+ cycling and leakage phenotypes in LVNC- and DCM-CM. ConclusionWe show the first comparative iPSC-model of splice-defect-associated RBM20-dependent LVNC-p.R634L and DCM-p.R634W. We found shared and variant-specific phenotypes on a patient-specific level. Our data demonstrate that the different RBM20 mutations manifest in distinct molecular aberrations in alternative splicing and RBM20 cytoplasmic accumulation that convey various physiological impairments in structure, Ca2+ handling, metabolism and contractile force.

molecular biology↗

bioRxiv2023

Visual-spatial dynamics drive adaptive social learning in immersive environments

Human cognition is distinguished by our ability to adapt to different environments and circumstances. Yet the mechanisms driving adaptive behavior have predominantly been studied in separate asocial and social contexts, with an integrated framework remaining elusive. Here, we use a collective foraging task in a virtual Minecraft environment to integrate these two fields, by leveraging automated transcriptions of visual field data combined with high-resolution spatial trajectories. Our behavioral analyses capture both the structure and temporal dynamics of social interactions, which are then directly tested using computational models sequentially predicting each foraging decision. These results reveal that adaptation mechanisms of both asocial foraging and selective social learning are driven by individual foraging success (rather than social factors). Furthermore, it is the degree of adaptivity--of both asocial and social learning--that best predicts individual performance. These findings not only integrate theories across asocial and social domains, but also provide key insights into the adaptability of human decision-making in complex and dynamic social landscapes.

animal behavior and cognition↗

bioRxiv2021

Deep assessment of human disease-associated ribosomal RNA modifications using Nanopore direct RNA sequencing

The catalytically active component of ribosomes, rRNA, is long studied and heavily modified. However, little is known about functional and pathological consequences of changes in human rRNA modification status. Direct RNA sequencing on the Nanopore platform enables the direct assessment of rRNA modifications. We established a targeted Nanopore direct rRNA sequencing approach and applied it to CRISPR-Cas9 engineered HCT116 cells, lacking specific enzymatic activities required to establish defined rRNA base modifications. We analyzed these sequencing data along with wild type samples and in vitro transcribed reference sequences to specifically detect changes in modification status. We show for the first time that direct RNA-sequencing is feasible on smaller, i.e. Flongle, flow cells. Our targeted approach reduces RNA input requirements, making it accessible to the analysis of limited samples such as patient derived material. The analysis of rRNA modifications during cardiomyocyte differentiation of human induced pluripotent stem cells, and of heart biopsies from cardiomyopathy patients revealed altered modifications of specific sites, among them pseudouridines, 2-O-methylation of riboses and acetylation of cytidines. Targeted direct rRNA-seq analysis with JACUSA2 opens up the possibility to analyze dynamic changes in rRNA modifications in a wide range of biological and clinical samples.

molecular biology↗

bioRxiv2021

Single-molecule, full-length transcript isoform sequencing reveals disease mutation-associated RNA isoforms in cardiomyocytes

Alternative splicing generates differing RNA isoforms that govern phenotypic complexity of eukaryotes. Its malfunction underlies many diseases, including cancer and cardiovascular diseases. Comparative analysis of RNA isoforms at the genome-wide scale has been difficult. Here, we established an experimental and computational pipeline that performs de novo transcript annotation and accurately quantifies transcript isoforms from cDNA sequences with a full-length isoform detection accuracy of 97.6%. We generated a searchable, quantitative human transcriptome annotation with 31,025 known and 5,740 novel transcript isoforms (http://steinmetzlab.embl.de/iBrowser/). By analyzing the isoforms in the presence of RNA Binding Motif Protein 20 (RBM20) mutations associated with aggressive dilated cardiomyopathy (DCM), we identified 121 differentially expressed transcript isoforms in 107 cardiac genes. Our approach enables quantitative dissection of complex transcript architecture instead of mere identification of inclusion or exclusion of individual exons, as exemplified by novel IMMT isoforms. Thereby we achieve a path to direct differential expression testing independent of an existing annotation of transcript isoforms as the functional unit, instead of genes or exons, providing more immediate biological interpretation and higher resolution transcriptome comparisons.

genetics↗

bioRxiv2021

Specialization and selective social attention establishes the balance between individual and social learning

A key question individuals face in any social learning environment is when to innovate alone and when to imitate others. Previous simulation results have found that the best performing groups exhibit an intermediate balance, yet it is still largely unknown how individuals collectively negotiate this balance. We use an immersive collective foraging experiment, implemented in the Minecraft game engine, facilitating unprecedented access to spatial trajectories and visual field data. The virtual environment imposes a limited field of view, creating a natural trade-off between allocating visual attention towards individual search or to look towards peers for social imitation. By analyzing foraging patterns, social interactions (visual and spatial), and social influence, we shine new light on how groups collectively adapt to the fluctuating demands of the environment through specialization and selective imitation, rather than homogeneity and indiscriminate copying of others.

animal behavior and cognition↗

bioRxiv2021

A genetic variant alters the secondary structure of the lncRNA H19 and is associated with Dilated Cardiomyopathy

lncRNAs are at the core of many regulatory processes and have also been recognized to be involved in various complex diseases. They affect gene regulation through direct interactions with RNA, DNA or proteins. Accordingly, lncRNAs structure is likely to be essential for their regulatory function. Point mutations, which manifest as SNPs (single nucleotide polymorphisms) in genome screens, can substantially alter their function and, subsequently, the expression of their down-stream regulated genes. To test the effect of SNPs on structure, we investigated lncRNAs associated with dilated cardiomyopathy. Among 322 human candidate lncRNAs we demonstrate first the significant association of a SNP located in lncRNA H19 using data from 1084 diseased and 751 control patients. H19 is generally highly expressed in the heart, with a complex expression pattern during heart development. Next, we used MFE (minimum free energy) folding to demonstrate a significant refolding in the secondary structure of this 861 nt long lncRNA. Since MFE folding may overlook the importance of sub-optimal structures, we showed that this refolding also manifests in the overall Boltzmann structure ensemble. There, the composition of structures is tremendously affected in their thermodynamic probabilities through the genetic variant. Finally, we confirmed these results experimentally, using SHAPE-Seq, corroborating that SNPs affecting such structures may explain hidden genetic variance not accounted for through genome wide association studies. Our results suggest that structural changes in lncRNAs, and lncRNA H19 in particular, affect regulatory processes and represent optimal targets for further in-depth studies probing their molecular interactions.

genetics↗

bioRxiv2020

Doxorubicin induces cardiotoxicity in a pluripotent stem cell model of aggressive B cell lymphoma cancer patients

Cancer therapies have been shown to induce cardiovascular complications. The aims of this study were to establish an in vitro induced pluripotent stem cell model (iPSC) of anthracycline-induced cardiotoxicity (ACT) from patients with an aggressive form of cancer. ACT-iPSC-CM generated from individuals with CD20+ B-cell lymphoma cancer who had received high doses of DOX and suffered cardiac dysfunction were observed to be persistently more susceptible to DOX toxicity compared to control-iPSC-CM. ACT-iPSC-CM exhibited increased DOX-dependent disorganized myofilament structure and cell death, as well as higher reactive oxygen species (ROS) compared to controls. Importantly, analysis of engineered heart muscle (EHM) from ACT-iPSC-CM showed an impaired DOX-dependent mechanical functionality. Transcriptome profiles of EHM are in line with a disturbed adjustment to DOX-dependent alteration of Ca2+ homeostasis in ACT-iPSC-CM. Furthermore, genetic variants in different cardiac key regulators were uncovered. In conclusion, we developed the first human iPSC-CM and EHM model of DOX-induced cardiac dysfunction in patients with B-cell lymphoma. Our results suggest that DOX-related stress resulted in decreased contractile activity and finally in heart failure in ACT patients. Brief summaryDevelopment of the first human iPSC-CM model of DOX-induced cardiac dysfunction in patients with aggressive B cell lymphoma and high-dose DOX treatment.

cell biology↗

bioRxiv2020

Genome wide association analysis in dilated cardiomyopathy reveals two new key players in systolic heart failure on chromosome 3p25.1 and 22q11.23

We present the results of the largest genome wide association study (GWAS) performed so far in dilated cardiomyopathy (DCM), a leading cause of systolic heart failure and cardiovascular death, with 2,719 cases and 4,440 controls in the discovery population. We identified and replicated two new DCM-associated loci, one on chromosome 3p25.1 (lead SNP rs62232870, p = 8.7 x 10-11 and 7.7 x 10-4 in the discovery and replication step, respectively) and the second on chromosome 22q11.23 (lead SNP rs7284877, p = 3.3 x 10-8 and 1.4 x 10-3 in the discovery and replication step, respectively) while confirming two previously identified DCM loci on chromosome 10 and 1, BAG3 and HSPB7. The genetic risk score constructed from the number of lead risk-alleles at these four DCM loci revealed that individuals with 8 risk-alleles were at a 27% increased risk of DCM compared to individuals with 5 risk alleles (median of the referral population). We estimated the genome wide heritability at 31% {+/-} 8%. In silico annotation and functional 4C-sequencing analysis on iPSC-derived cardiomyocytes strongly suggest SLC6A6 as the most likely DCM gene at the 3p25.1 locus. This gene encodes a taurine and beta-alanine transporter whose involvement in myocardial dysfunction and DCM is supported by recent observations in humans and mice. Although less easy to discriminate the better candidate at the 22q11.23 locus, SMARCB1 appears as the strongest one. This study provides both a better understanding of the genetic architecture of DCM and new knowledge on novel biological pathways underlying heart failure, with the potential for a therapeutic perspective.

genomics↗

bioRxiv2020

Similarities and differences in spatial and non-spatial cognitive maps

Learning and generalization in spatial domains is often thought to rely on a "cognitive map", representing relationships between spatial locations. Recent research suggests that this same neural machinery is also recruited for reasoning about more abstract, conceptual forms of knowledge. Yet, to what extent do spatial and conceptual reasoning share common computational principles, and what are the implications for behavior? Using a within-subject design we studied how participants used spatial or conceptual distances to generalize and search for correlated rewards in successive multi-armed bandit tasks. Participant behavior indicated sensitivity to both spatial and conceptual distance, and was best captured using a Bayesian model of generalization that formalized distance-dependent generalization and uncertainty-guided exploration as a Gaussian Process regression with a radial basis function kernel. The same Gaussian Process model best captured human search decisions and judgments in both domains, and could simulate realistic learning curves, where we found equivalent levels of generalization in spatial and conceptual tasks. At the same time, we also find characteristic differences between domains. Relative to the spatial domain, participants showed reduced levels of uncertainty-directed exploration and increased levels of random exploration in the conceptual domain. Participants also displayed a one-directional transfer effect, where experience in the spatial task boosted performance in the conceptual task, but not vice versa. While confidence judgments indicated that participants were sensitive to the uncertainty of their knowledge in both tasks, they did not or could not leverage their estimates of uncertainty to guide exploration in the conceptual task. These results support the notion that value-guided learning and generalization recruit cognitive-map dependent computational mechanisms in spatial and conceptual domains. Yet both behavioral and model-based analyses suggest domain specific differences in how these representations map onto actions. Author summaryThere is a resurgence of interest in "cognitive maps" based on recent evidence that the hippocampal-entorhinal system encodes both spatial and non-spatial relational information, with far-reaching implications for human behavior. Yet little is known about the commonalities and differences in the computational principles underlying human learning and decision making in spatial and non-spatial domains. We use a within-subject design to examine how humans search for either spatially or conceptually correlated rewards. Using a Bayesian learning model, we find evidence for the same computational mechanisms of generalization across domains. While participants were sensitive to expected rewards and uncertainty in both tasks, how they leveraged this knowledge to guide exploration was different: participants displayed less uncertainty-directed and more random exploration in the conceptual domain. Moreover, experience with the spatial task improved conceptual performance, but not vice versa. These results provide important insights about the degree of overlap between spatial and conceptual cognition.

animal behavior and cognition↗

bioRxiv2018

Searching for rewards like a child means less generalization and more directed exploration

How do children and adults differ in their search for rewards? We consider three different hypotheses that attribute developmental differences to either childrens increased random sampling, more directed exploration towards uncertain options, or narrower generalization. Using a search task in which noisy rewards are spatially correlated on a grid, we compare 55 younger children (age 7-8), 55 older children (age 9-11), and 50 adults (age 19-55) in their ability to successfully generalize about unobserved outcomes and balance the exploration-exploitation dilemma. Our results show that children explore more eagerly than adults, but obtain lower rewards. Building a predictive model of search to disentangle the unique contributions of the three hypotheses of developmental differences, we find robust and recoverable parameter estimates indicating that children generalize less and rely on directed exploration more than adults. We do not, however, find reliable differences in terms of random sampling.

animal behavior and cognition↗

bioRxiv2018

Active Function Learning

How do people actively explore to learn about functional relationships, that is, how continuous inputs map onto continuous outputs? We introduce a novel paradigm to investigate information search in continuous, multi-feature function learning scenarios. Participants either actively selected or passively observed information to learn about an underlying linear function. We develop and compare different variants of rule-based (linear regression) and non-parametric (Gaussian process regression) active learning approaches to model participants active learning behavior. Our results show that participants performance is best described by a rule-based model that attempts to efficiently learn linear functions with a focus on high and uncertain outcomes. These results advance our understanding of how people actively search for information to learn about functional relations in the environment.

animal behavior and cognition↗

bioRxiv2018

Connecting conceptual and spatial search via a model of generalization

The idea of a \"cognitive map\" was originally developed to explain planning and generalization in spatial domains through a representation of inferred relationships between experiences. Recently, new research has suggested similar principles may also govern the representation of more abstract, conceptual knowledge in the brain. We test whether the search for rewards in conceptual spaces follows similar computational principles as in spatial environments. Using a within-subject design, participants searched for both spatially and conceptually correlated rewards in multi-armed bandit tasks. We use a Gaussian Process model combining generalization with an optimistic sampling strategy to capture human search decisions and judgments in both domains, and to simulate human-level performance when specified with participant parameter estimates. In line with the notion of a domain-general generalization mechanism, parameter estimates correlate across spatial and conceptual search, yet some differences also emerged, with participants generalizing less and exploiting more in the conceptual domain.

animal behavior and cognition↗

bioRxiv2017

Exploration and generalization in vast spaces

From foraging for food to learning complex games, many aspects of human behaviour can be framed as a search problem with a vast space of possible actions. Under finite search horizons, optimal solutions are generally unobtainable. Yet how do humans navigate vast problem spaces, which require intelligent exploration of unobserved actions? Using a variety of bandit tasks with up to 121 arms, we study how humans search for rewards under limited search horizons, where the spatial correlation of rewards (in both generated and natural environments) provides traction for generalization. Across a variety of diifferent probabilistic and heuristic models, we find evidence that Gaussian Process function learning--combined with an optimistic Upper Confidence Bound sampling strategy--provides a robust account of how people use generalization to guide search. Our modelling results and parameter estimates are recoverable, and can be used to simulate human-like performance, providing insights about human behaviour in complex environments.

animal behavior and cognition↗

bioRxiv2017

Mapping the unknown: The spatially correlated multi-armed bandit

We introduce the spatially correlated multi-armed bandit as a task coupling function learning with the exploration-exploitation trade-off. Participants interacted with bi-variate reward functions on a two-dimensional grid, with the goal of either gaining the largest average score or finding the largest payoff. By providing an opportunity to learn the underlying reward function through spatial correlations, we model to what extent people form beliefs about unexplored payoffs and how that guides search behavior. Participants adapted to assigned payoff conditions, performed better in smooth than in rough environments, and--surprisingly--sometimes performed equally well in short as in long search horizons. Our modeling results indicate a preference for local search options, which when accounted for, still suggests participants were best-described as forming local inferences about unexplored regions, combined with a search strategy that directly traded off between exploiting high expected rewards and exploring to reduce uncertainty about the spatial structure of rewards.

animal behavior and cognition↗