Intergroup conflict is a potent evolutionary force across taxa, but little research has investigated how social animals pre-emptively change their behaviour in scenarios where contests with rivals are more likely. Moreover, the few studies examining this aspect of intergroup conflict fail to consider rival characteristics despite them determining contest outcomes and interaction costs, which define the threat level different groups pose. Here we show how non-human animals can tailor their anticipatory behaviour to the specific threat posed by rivals, using 10 years of detailed behavioural observations and GPS data. We demonstrate that dwarf mongooses (Helogale parvula) adjust their space use, information provisioning and resource defence dependent on the relative group size of neighbours to help them mitigate the threat from both well-matched competitors and more dangerous larger rivals. By contrast, behavioural differences between the core and edge of home ranges were more equivocal, highlighting the importance of considering rival characteristics rather than just spatial location as an indicator of threat level. Our results showcase how animals have evolved to be best prepared for a key part of their ecology, potential future contests, indicating abilities that allow them to survive and thrive in a landscape of intergroup conflict.
Previous studies have found that androstadienone (AND) influences women's mate preferences and their attractiveness ratings of men. This study examined whether AND affects women's engagement in intrasexual competition (i.e., the likelihood of gossip) and intrasexual competition-related perceptions (i.e., perceptions of flirtatiousness from romantic rivals). Overall, 52 women participated in a double-blind, placebo-controlled, and within-subject experiment, where they were randomly assigned to receive the AND or placebo on two consecutive experimental days. After the AND or placebo was administered, participants completed an information-sharing task and a first-impression task in the context of intrasexual competition to assess women's self-reported likelihood of gossip and perceptions of flirtatiousness of romantic rivals, respectively. The results indicated that AND increased the self-reported likelihood of negative gossip in women. However, the influence of AND on the self-reported likelihood of neutral and positive gossip in women was not significant. In addition, AND increased women's perceptions of flirtatiousness from romantic rivals. These findings indicate that AND may intensify women's intrasexual competition and perceived romantic threat in the context of intrasexual competition.
In the vanishing ball illusion (VBI), a magician feigns throwing a ball, and spectators report seeing it rise into the air, even though it never leaves the performer's hand. Although many studies have investigated factors that can modulate this illusion, none have examined the extent to which participants can detect this anticipatory error, or whether susceptibility to the illusion and this potential detection depend on available executive resources. In the present study, error detection was assessed by comparing the confidence in the perceptual experience of participants susceptible to the VBI (N = 64) with that of participants exposed to a real throw (N = 64). Moreover, working memory resources were manipulated by asking participants to memorize either two (low load) or four (high load) cross positions presented in a 3 × 3 grid. Contrary to our hypothesis, working memory resource availability did not significantly affect susceptibility to the VBI. Moreover, participants' confidence in detecting an error did not differ between those experiencing the VBI and those observing a real throw, and this lack of error detection was independent of available working memory resources. Our findings suggest that the illusory ball is interpreted and trusted as if it were genuine bottom-up information (i.e., a real throw). We discuss the mechanisms underlying this illusion that may account for the absence of error detection, focusing in particular on the roles of source monitoring and the automatization of perceptual simplification processes.
Compelling epidemiological evidence suggests that exercise and smoking are modifiable risk factors that are linked to a reduced risk of Parkinson's disease. These two risk factors represent opposite ends of a spectrum: exercise is universally embraced, while smoking is rightly eschewed for its established adverse health effects. Yet, intriguingly, preclinical evidence suggests that at their biological cores, exercise and some of the many components of tobacco may share strikingly similar working mechanisms that may favorably modify PD risk or disease course, for which definitive evidence is still lacking. Here, we deconstruct these overlapping and putative neuroprotective mechanisms. Our aim is to transform this unexpected overlap into an actionable perspective toward identifying novel targets for disease-modifying therapies that can slow the progression of Parkinson's disease, and to inspire novel translational efforts in disease modification in PD. We stress that while both factors may theoretically inform disease-modifying strategies for PD, in practice, only exercise should be promoted for its health benefits, whereas smoking remains firmly contraindicated due to its known detrimental health effects.
The human extrastriate visual cortex contains fine-scale columns selectively responsive to motion, disparity, and color. However, the developmental interplay between these functional modules remains poorly understood. Using high-resolution functional MRI, we compared the mesoscale organization of the extrastriate cortex in 16 individuals with normal vision and 15 participants with amblyopia (PwA) caused by strabismus (n = 8) or anisometropia (n = 7). In controls, the cortical territory occupied by disparity-selective columns exhibited a competitive relationship with that of motion- and color-selective columns. In PwA, we witnessed a reduction in the size of disparity-selective columns accompanied by expansion of the cortical territory allocated to motion- and color-selective columns, while the interdigitated organization of these sites remained unchanged. At the macroscale, this phenomenon simply manifested as weaker disparity- plus stronger motion- and color-selective responses in PwA than controls. Our results show that the mesoscale modules are rivals in development allowing intact functions to usurp those that are compromised.
Achieving renewable, high-performance elastomers remains a key goal of green materials science. Here, we report a high-performance and recyclable elastomer produced directly from industrial Kraft lignin through a one-pot in situ graft copolymerization strategy. A deep eutectic solvent composed of oxalic acid and 1,6-hexanediol simultaneously dissolves lignin, provides flexible chains, and drives catalyst-free esterification at 110°C, constructing an interpenetrating rigid-flexible network that incorporates 50-75 wt.% lignin. The optimal elastomer delivers a tensile strength of 12.0 MPa, 878% elongation, and 85.1 MJ m- 3 fracture energy, rivaling or exceeding petroleum-derived nitrile butadiene rubber (3.1 MPa, 750% elongation and 11.0 MJ m- 3). It also offers a low dielectric constant, high electrical insulation, superior oil and abrasion resistance, efficient photothermal conversion, and infrared-induced self-healing. The material can be repeatedly reprocessed, enabling closed-loop recycling. Converting an abundant lignin by-product into a value-added elastomer thus provides a scalable route to eco-materials with broad application potential.
暂无摘要（点击查看详情）
暂无摘要（点击查看详情）
Clinical audit and quality improvement are both central to improving healthcare quality; however, there is often a lack of conceptual clarity about their similarities and differences. They are frequently misunderstood, conflated, or applied in ways that do not match their primary purpose. In the United Kingdom, this lack of conceptual clarity is evident across clinical practice, healthcare improvement literature, and professional curricula. The misapplication of methods may contribute to inefficient use of resources and missed opportunities to improve experiences and outcomes for patients, families, and staff. This practice-informed perspective article clarifies the distinct purposes and roles of clinical audit and quality improvement before demonstrating how they can be used in practice. Drawing on published literature, professional guidance, and the authors' practical experience, we examine the key similarities, differences, and common points of confusion between clinical audit and quality improvement. We highlight recurring challenges faced by UK healthcare professionals when selecting and applying these approaches in real-world settings, as a call for international reflection. We present a decision aid to support intentional and effective method selection and purposeful transition between approaches. By providing clarity on how clinical audit and quality improvement are distinct yet complementary, this article aims to support more rigorous, contextually appropriate, and impactful improvement efforts across all healthcare systems.
Visible and infrared image fusion (VIF) has gained significant attention in recent years due to its wide application in tasks such as scene segmentation and object detection. VIF methods can be broadly classified into traditional VIF methods and application-oriented VIF methods. Traditional methods focus solely on improving the quality of fused images, while application-oriented VIF methods additionally consider the performance of downstream tasks on fused images by introducing task-specific loss terms during training. However, compared to traditional methods, application-oriented VIF methods require datasets labeled for downstream tasks (e.g., semantic segmentation or object detection), making data acquisition labor-intensive and time-consuming. To address this issue, we propose a self-supervised training framework for segmentation-oriented VIF methods (SSVIF). Leveraging the consistency between feature-level fusion-based segmentation and pixel-level fusion-based segmentation, we introduce a novel self-supervised task, i.e., cross-segmentation consistency, that enables the fusion model to learn high-level semantic features without the supervision of segmentation labels. Additionally, we design a two-stage training strategy and a dynamic weight adjustment method for effective joint learning within our self-supervised framework. Extensive experiments on public datasets demonstrate the effectiveness of our proposed SSVIF. Remarkably, although trained only on unlabeled visible-infrared image pairs, our SSVIF outperforms traditional VIF methods and rivals supervised segmentation-oriented ones. Our code is publicly available at https://github.com/CnoyZ/SSVIF.
Theory-based and model-based reasoning shape macroecology: one derives models from general principles, the other builds models around data. We argue these are not rival traditions but interacting positions along a continuum of formalization. This pluralism is macroecology's strength: by allowing theories and models to exchange roles as sources of constraint, synthesis and innovation, the field can confront the complexity of natural systems and generate explanations that are cumulative and predictive.
The efficient conversion of lignin-derived oxygenates into fuels and high-value chemicals is pivotal for the development of renewable energy systems. A central challenge lies in the selective cleavage of the inherently recalcitrant C-O bonds in bio-oil model compounds under mild conditions. In this study, a multi-interface Co-Ce catalyst supported on nitrogen-doped carbon was successfully constructed via a bimetallic synergy and metal-organic framework (MOF)-derived strategy. The introduction of Ce exerted a multifunctional synergistic effect beyond that of a conventional promoter, realizing multi-scale modulation of the catalyst structure and performance. It modulated the electronic structure of Co active sites, thereby optimizing reactant adsorption and H2 activation. Also, it suppressed the agglomeration of Co nanoparticles, thereby enhancing metal dispersion. Concurrently, it significantly increased the concentration of oxygen vacancies on the support surface, which collaboratively provided abundant active sites. Furthermore, it guided the formation of a hierarchical pore architecture conducive to mass transfer. Through this precise design and regulation of the active site microenvironment, the optimal catalyst achieved 100 % conversion of the model compound vanillin with 100 % selectivity toward 2-methoxy-4-methylphenol (MMP) under mild conditions, while also exhibiting excellent cycling stability and broad substrate applicability. This work not only demonstrates catalytic performance rivaling that of noble metals through the efficient synergy between non-noble metals (Ce and Co) but, more importantly, elucidates the underlying mechanism for the selective cleavage of C-O bonds.
Large language models (LLMs) have demonstrated expert-level performance on medical licensing examinations, but most benchmarks focus on final accuracy, obscuring model-specific behaviors. Critical gaps remain in understanding model efficiency (latency), the efficacy of tiered "rescue" protocols for error correction, and the systematic correlation between performance and human-rated question difficulty. The German M2 exam, paired with the AMBOSS platform's user-data-driven difficulty ratings, provides a unique opportunity to map AI performance directly against human cognitive load. This study aimed to move beyond singular accuracy scores by (1) evaluating and comparing the baseline (Tier 1) accuracy and response latency of next-generation rapid-response LLMs; (2) analyzing the efficacy of a two-tiered rescue (Tier 2) protocol in correcting initial errors; and (3) correlating model performance with the user-data-driven Amboss difficulty rating. We evaluated four LLMs (Gemini 2.5 Flash/Pro and ChatGPT 5 Instant/Thinking) on the complete 316-item German M2 (Fall 2024) medical exam, including all multimodal (image-based) questions. A zero-shot copy-paste prompting strategy was utilized, and outputs were evaluated against ground-truth answers using a strict exact-match criterion. A two-tiered protocol was used: Tier 1 (Flash/Instant) provided baseline responses. If incorrect, a Tier 2 (Pro/Thinking) model was deployed as a "rescue." Performance was analyzed using McNemar's test, Wilcoxon signed-rank test, Fisher's exact test, and logistic regression. Baseline (Tier 1) accuracy was identical at 91.46% (95% CI 87.85-94.06; n = 289/316) for both Gemini 2.5 Flash and ChatGPT 5 Instant, with 27 errors each. However, Gemini Flash (Mean=1.57s) was significantly faster than ChatGPT Instant (Mean = 2.07s; P < .001). Additionally, ChatGPT Instant expended significantly more time on incorrect answers compared to correct ones (P = .002), whereas Gemini Flash showed no such hesitation (P = .814). The Tier 2 rescue rate for ChatGPT 5 Thinking (48.15%, 13/27; 95% CI 30.74-66.01) was higher, though not statistically significant (P = .406), than for Gemini 2.5 Pro (33.33%, 9/27; 95% CI 18.64-52.18). This rescue protocol elevated final accuracy to 94.30% (95% CI 91.18-96.37) for the Gemini system and 95.57% (95% CI 92.70-97.34) for the ChatGPT system (P = .481). A strong, inverse relationship with difficulty was found: for every one-point difficulty increase, the odds of a correct Tier 1 response decreased by 42.1% (OR 0.579, 95% CI 0.425-0.788; P < .001) for Gemini Flash and 47.7% (OR 0.523, 95% CI 0.379-0.720; P < .001) for ChatGPT Instant. This negative correlation persisted even after the rescue (P = .013 and P = .006, respectively). Expert-level LLM performance on the German M2 exam masks a critical, systematic vulnerability: a significant decrease in accuracy directly correlated with increased question difficulty. A two-tiered "rescue" system is an effective strategy to mitigate these difficulty-based failures and achieve >95% accuracy, rivaling the best-performing, full-capacity models. We conclude that a simple reliance on a single model is insufficient; hierarchical systems that manage query difficulty are essential for safe and effective integration into medical education.
Psychological research increasingly relies on computational methods to track emotion in naturalistic text. However, standard lexicon-based tools often miss semantic nuance, while powerful commercial Large Language Models (LLMs) raise privacy concerns for sensitive data. This study evaluates the efficacy of 24 open-weight LLMs (1B-120B parameters) for zero-shot sentiment analysis of spoken language, running entirely on local hardware. In this paradigm, models classify text relying solely on natural language instructions rather than labeled training examples. We compared model performance against three increasingly difficult baselines (naïve, standard, and human) across transcripts from two datasets: 193 autobiographical narratives from community participants (N=49) and 292 longitudinal audio journals from psychiatric outpatients (N=64). Results demonstrate that open-weight models significantly outperform standard lexicon-based sentiment tools and frequently surpass individual human raters. Crucially, certain mid-sized models rival the performance of much larger systems, making state-of-the-art analysis accessible via consumer hardware. Additionally, we validate a fully automated privacy-preserving pipeline, finding that transcription errors from automatic speech recognition did not significantly degrade downstream sentiment accuracy. Despite these strengths, a multidimensional fairness audit revealed several demographic disparities: our best-performing models exhibited miscalibration in specific subgroups, lower sensitivity for female speakers in the community dataset, and lower predictive precision for Black speakers in the clinical dataset. Taken together, these findings demonstrate that recent open-weight LLMs can match or surpass human-level performance in sentiment analysis of naturalistic speech while running efficiently and securely on workstation hardware. These advances open opportunities for studying emotional dynamics in daily life and developing privacy-preserving clinical tools.
Neural modeling and large language models (LLMs) has led to a significant improvement in the quality of machine translation (MT) output. While MT increasingly rivals human output, "translationese"-systematic linguistic fingerprints left by the translation process-has long been studied qualitatively, yet its quantitative boundaries remain unclear. We present an interpretable machine-learning framework that classifies Chinese-to-English human, Google-Translate, and ChatGPT outputs across news, novel, and technology genres using a dataset of 450 texts. From 308 candidate linguistic indicators which were normalized for text length, Elastic-Net logistic regression selected 14 robust predictors strictly from the training set to prevent data leakage. The partial least squares discriminant analysis (PLS-DA) model stood out across 10 algorithms and achieved an F1-score of 0.90 and an AUC (0.958 ± 0.022) after bootstrap validation. Global, conditional, and local SHapley Additive exPlanations (SHAP) reveal that normalized discourse-level cohesion (PIN: Present-participial clause density), morphological richness (PRMD: Predictive modal density), and adposition density (d_adp) are the strongest, genre-stable discriminators. Functional analysis suggests that these features serve as proxies for informational density and stancetaking, respectively, allowing the model to distinguish human stylistic sensitivity from machine normalization. These attributions align with theoretical constructs of shining-through and normalization found in Corpus-Based Translation Studies literature. By integrating interpretable modeling, this study addresses the "black box" problem in text classification, offering a potential methodological template for differentiating human from machine translations.
Levine et al. understate the range of folk moral judgments that cannot be easily explained by their contractualist theory of moral cognition (including judgments about future generations and nonhuman animals). On the flipside, Levine et al. also overstate the range of moral judgments that cannot be easily explained by rival deontological or consequentialist theories of moral cognition.
This study investigates the factors influencing copulation latency (CL) in male Drosophila melanogaster, a crucial determinant of reproductive fitness. We explore the interplay between genetic, environmental, and social factors that shape mating strategies, emphasizing the role of behavioral plasticity in CL. Our findings reveal that sensory perception, particularly visual acuity, is critical for normal CL, as demonstrated by delayed mating in visual mutants. Contrary to expectations, metabolic state and prior experiences with rivals do not significantly affect CL. We identify specific circadian clock genes, period (per) and cycle (cyc), that independently regulate CL through mechanisms that are functionally separable from their canonical roles in circadian rhythm regulation. Additionally, we highlight the essential roles of ITP (Ion transport peptide) and sNPF (short neuropeptide F) in modulating CL through ITP-LNd neurons. ITP-LNd neurons integrate various signaling inputs, including octopamine (OA) and glutamate (Glu), which are crucial for generating normal CL. This research provides novel insights into the genetic and neural mechanisms underlying behavioral plasticity in mating strategies, suggesting that the integration of clock gene function within neuropeptidergic signaling is essential for optimal reproductive success in Drosophila.
Root exudates comprise a diverse mixture of non-polar and amphiphilic compounds that are only partially recovered by aqueous extraction methods, yet can rival polar metabolites as carbon sources for microorganisms. Because many quorum-sensing (QS) signals are fatty-acyl derivatives, lipid-rich microhabitats at the root-soil interface are likely to influence signal partitioning, persistence, and local QS thresholds. We propose a lipid-mediated framework in which plant-derived lipids modulate QS through four nodes: receptor mimicry/antagonism, quorum quenching, membrane/microenvironment regulation, and lipid-dependent resource gating. These mechanisms operate across two lipid-rich interfaces - the rhizosphere and the arbuscular mycorrhizal fungi (AM fungi) hyphosphere - where host lipid fluxes may restructure microbial community composition. Combined with QS-sensitive changes in root exudation, this spatially structured lipid circuit could generate feedbacks influencing microbiome composition and function, with potential implications for microbiome engineering.
The development of sustainable hydrogen evolution reaction (HER) based on earth-abundant molecular electrocatalysts requires strategies that enable controlled access to reactive low-valent intermediates while overcoming conventional activity-overpotential scaling relationships. Although transition metal complexes have been extensively explored, the potential of main-group systems remains largely untapped due to challenges in stabilizing reactive low-valent states and the lack of predictive catalyst design frameworks integrating electronics, geometry, and secondary sphere effects. Herein, we report a series of NCN pincer-supported Bi(III) organometallic catalysts [(L1)BiCl2] (1), [(L2)BiCl2] (2), [(L3)BiCl2] (3), [(L4)BiCl2] (4), [(L5)BiCl2] (5), [(L6)BiCl2] (6), [(L7)BiCl2] (7)) that promote HER through in situ electrochemical access of active Bi(I) intermediates. Mechanistic investigations combining potential-pKa analysis, Tafel slope measurements, in situ electrochemical studies, detailed electrokinetic study exploring foot-of-the-wave analysis (FOWA), controlled potential electrolysis, and density functional theory calculations support a BiI/BiIII redox cycle-mediated proton-coupled electron transfer (PCET) pathway involving transient bismuth hydride intermediates that are challenging to isolate under conventional chemical conditions. Systematic modulation of ligand electronics, geometry, and secondary coordination sphere features reveals a prominent role in modulating reactivity and further highlights that incorporation of a pendant -NH functionality acts as an intramolecular proton-assistance that significantly enhances catalytic activity, relative to analogues lacking pendant -NH proton functionality in the ligand backbone. Notably, secondary-sphere interactions from the ligand backbone enable these catalysts to circumvent traditional activity-overpotential scaling relationships, exhibiting enhanced performance at low overpotential that rivals state-of-the-art transition metal systems. These findings establish fundamental design principles for redox-active main-group electrocatalysis and expand the accessible redox space of the periodic table by leveraging p-block redox chemistry toward practical applications.
Trimer-based metal clusters are among the most compositionally versatile secondary building units in metal-organic frameworks (MOFs), but systematic control of heterometallic composition and framework charge within a structurally invariant platform remains challenging. Here, pore-space-partitioned acs (pacs) frameworks are employed as a model system to investigate compositional tunability at both the cluster and framework levels. Using 3,4-dimethylthieno[2,3-b]thiophene-2,5-dicarboxylate (dtc), a family of homo- and heterometallic pacs-MOFs based on Mg2+, Co2+, In3+, and V3+ trimers was synthesized. In the charge-complementary Mg/In system, variation of metal precursors and solvents enables broad tuning of the Mg/In ratio, while the large atomic-number contrast permits reliable occupancy refinement by single-crystal X-ray diffraction. The trimeric platform also supports an isostructural Co-, Co/V-, and V-based series in which framework charge changes from anionic to neutral and cationic without altering topology. Gas adsorption studies reveal charge-dependent C2H2/CO2 and C2H6/C2H4 separation behavior. Notably, neutral CoV-dtc-tpt exhibits a high C2H6 uptake of 164.8 cm3 g-1 and excellent inverse-selective C2H6/C2H4 separation, rivaling benchmark pacs materials. These results establish pacs-MOFs as a versatile platform for studying heterometallic assembly, framework-charge modulation, and structure-composition-property relationships.