Next Issue

2026, Volume 34 Issue 11 Previous Issue   
---
Conceptual Framework
Meta-Analysis
Regular Articles
Research Method


For Selected:

Download Citations EndNote Ris BibTeX

Toggle Thumbnails

Conceptual Framework

Select

A core unifying biomarker underlying heterogeneous non-pharmacological interventions that attenuate age-related cognitive decline and its causal validation JIN Xinhu, TANG Wei, LI Juan 2026, 34 (11):  1899-1907.  doi: 10.3724/SP.J.1042.2026.1899 Abstract ( 92 )   PDF (1779KB) ( 104 )   Peer Review Comments Population aging has led to an increasing prevalence of cognitive decline and neurodegenerative disorders, creating an urgent need for effective intervention strategies. Non-pharmacological interventions, such as cognitive training and physical exercise, have consistently shown beneficial effects on cognitive functioning in older adults. However, existing studies largely focus on intervention-specific neural changes and lack a unified framework explaining whether different intervention modalities share common neurobiological mechanisms. To address this gap, the present work proposes the hypothesis that diverse non-pharmacological interventions may converge on a common large-scale brain network that mediates intervention-induced neuroplasticity and cognitive benefits. To identify and causally validate such a network, three complementary studies are proposed. The first study aims to identify a core unifying biomarker underlying the beneficial effects of different interventions. Multimodal datasets collected from cognitively normal older adults who participated in cognitive training, exercise training, and multimodal intervention programs will be integrated into a unified intervention database. Resting-state functional MRI data will be harmonized using standardized preprocessing and site-effect correction procedures. To characterize large-scale functional organization, functional gradients will be derived from whole-brain connectivity matrices, while intrinsic timescales will be calculated to quantify hierarchical temporal processing across brain networks. Changes in these measures before and after intervention in the training group will be compared with those observed in control participants. Computational virtual-lesion analyses will then be employed to determine which large-scale network contributes most critically to intervention-related maintenance of youthful brain organization. Furthermore, structural and functional asymmetry measures will be used to distinguish whether intervention-related plasticity reflects compensatory recruitment or optimization of neural efficiency. This multidimensional framework integrates spatial hierarchy, temporal dynamics, and hemispheric specialization to identify convergent neural mechanisms shared across intervention modalities. The study is expected to reveal the dorsal attention network (DAN) as a candidate common brain network whose enhancement is associated with improved cognitive performance following intervention. The second study provides causal validation of the identified network in cognitively normal older adults. Participants will be randomly assigned to active or sham stimulation groups and receive a two-week intervention using repetitive transcranial magnetic stimulation (rTMS). Based on findings from Study 1, stimulation will target a core hub of the DAN located in the superior parietal lobule. Behavioral performance will be assessed using the mnemonic similarity task, a sensitive measure of age-related memory decline. In addition, task-based functional magnetic resonance imaging will be conducted before and after intervention to examine neural activity and functional connectivity changes associated with memory discrimination. By directly modulating a network hub identified through large-scale neuroimaging analyses, this study tests whether enhancing DAN function improves cognitive performance and alters memory-related brain activity. Demonstrating such effects would provide causal evidence that the DAN represents a key neural substrate underlying the effectiveness of diverse non-pharmacological interventions. The third study extends causal validation to older adults with subjective cognitive decline, a population widely considered to represent a preclinical stage of cognitive impairment. Using the same randomized sham-controlled rTMS protocol, participants will undergo stimulation targeting the DAN and complete standardized neuropsychological assessments together with the mnemonic similarity task before and after intervention. Unlike Study 2, the primary focus in Study 3 is behavioral improvement in a population at elevated risk for future cognitive decline. If stimulation-induced enhancement of DAN function produces measurable cognitive benefits in this group, the findings would demonstrate that the identified network contributes not only to healthy cognitive aging but also to resilience against pathological aging processes. The present work contains two major innovations. The first is a theoretical innovation: it moves beyond intervention-specific explanations and introduces the concept of a common brain network shared across heterogeneous intervention modalities. This framework provides a unified account of how different interventions may promote cognitive resilience through convergent neuroplastic mechanisms. The second is a methodological innovation: by integrating functional gradients, intrinsic timescales, brain asymmetry measures, computational virtual-lesion analyses, task-based fMRI, and noninvasive brain stimulation, the study establishes a multidimensional and multimodal approach for investigating aging-related neuroplasticity. Importantly, the combination of large-scale neuroimaging analyses and randomized rTMS experiments creates a complete research pathway from mechanism discovery to causal verification across both healthy and pathological aging populations. Overall, this work aims to establish a core unifying biomarker for evaluating intervention effectiveness, provide mechanistic insight into how non-pharmacological interventions delay cognitive decline, and identify potential neuromodulation targets for future precision interventions. These findings are expected to advance theoretical understanding of neuroplasticity in aging and contribute to the development of evidence-based strategies for promoting healthy aging and preventing dementia. References | Related Articles | Metrics

Select

The dynamics of human trust in AI from the instrumental and value perspectives SONG Yu, HU Xiaoran 2026, 34 (11):  1908-1932.  doi: 10.3724/SP.J.1042.2026.1908 Abstract ( 91 )   PDF (812KB) ( 141 )   Peer Review Comments With the rapid development of artificial intelligence (AI) technology, human-AI relationships have become increasingly prevalent and consequential in organizations. Human trust in AI lies at the core of human-AI relationships and is critical to the effectiveness of human-AI interactions. Key challenges in research on human trust in AI include how to conceptualize trust, understand dynamic patterns of human-AI relationships, and achieve complementary advantages through human-AI interactions. This research addresses these issues by focusing on the dyadic interaction between humans and AI to explore the dynamic processes of human trust in AI over time in the workplace. Specifically, our research is structured around three progressive studies—theoretical reconceptualization, dynamic patterns, and impact effects—to systematically answer the fundamental questions of what human trust in AI is, how it develops, and how it can be effectively leveraged. First, drawing on the perspective of technological ethics, this study conceptualizes human trust in AI as a two-dimensional construct comprising instrumental trust and value trust, and further develops a corresponding measurement scale. Second, adopting a dynamic development perspective, the study explores the temporal characteristics and dynamic patterns of human trust in AI, thereby opening the “black box” of trust dynamics in human-AI relationships. Finally, from the perspective of human-AI collaboration, the study investigates the effect of different forms of human trust in AI on employee creativity, offering a nuanced understanding of human-AI relationship development and providing insights into how trust in AI shapes employees’ core competencies in the digital intelligence era. Our work contributes to the literature on human trust in AI in three ways. First, we advance a conceptual reconstruction of human trust in AI by proposing an instrumental-value trust framework. Moving beyond the cognitive-affective trust framework, and grounded in the classical definition of trust as well as the established analytical distinction between trust beliefs and trust intentions, we propose a dual-dimensional structural model of human trust in AI based on the distinct value orientations embedded in trust beliefs: instrumental trust and value trust. This framework unpacks the dual logic underlying human acceptance of AI, namely, technological reliance and value consensus, and overcomes the theoretical and measurement limitations of prior research that directly transplanted the categories of cognitive trust and affective trust from interpersonal trust studies into the human-AI context, thereby offering a more contextually grounded and explanatorily powerful theoretical framework for human trust in AI. Second, we uncover the dynamic patterns of human trust in AI. Adopting the information processing perspective and integrating three core elements of trust dynamics—structure, intensity, and time—we delineate the nonlinear developmental trajectories of human trust in AI. Specifically, we identify the threshold effects associated with the development of instrumental trust and value trust, and clarify the relative importance of different trust cues throughout the trust development process. These findings address the limitations of existing dynamic research on human trust in AI, which has largely focused on variations in trust intensity alone, and further specify the temporal boundaries of the effects of different trust cues. Finally, we elucidate the mechanisms through which human trust in AI influences employees’ creativity. From the perspective of human-AI collaboration, we explore how instrumental trust and value trust differentially shape modes of human-AI collaboration and, in turn, differentially enhance individual creativity. We further identify the contextual roles of individual mindsets and algorithmic management in this process. By doing so, we not only extend the general proposition that “human trust in AI promotes AI acceptance and use” into the more nuanced insight that “different forms of human trust in AI foster differentiated patterns of AI use”, but also provide concrete pathways for cultivating creativity through human-AI collaboration, alleviate concerns regarding cognitive degradation resulting from overreliance on AI, and offer a dialectical perspective on the development of human-AI relationships. References | Related Articles | Metrics

Select

How algorithmic management affects the work-family interface: A dual-process and dissipative structure perspective CHEN Long, ZOU Xing, YAO Feisi, GAO Xinyu 2026, 34 (11):  1933-1948.  doi: 10.3724/SP.J.1042.2026.1933 Abstract ( 60 )   PDF (674KB) ( 55 )   Peer Review Comments While improving the management efficiency of gig platforms, algorithmic management has also brought challenges to gig workers. An increasing number of gig workers endure long working hours, tight schedules, and blurred work-family boundaries. How to balance work and family has become a common concern of both platform managers and gig workers. Current literature lacks an integrated framework to analyze the impact of algorithmic management on the work-family interface. Moreover, previous research on algorithmic management has overlooked the dissipative structural characteristics of the human brain and the individual. Grounded in the dual-process model and dissipative structure theory, this study constructs moderated mediation models of the impact of algorithmic management on work-family facilitation and work-family conflict, providing an integrated theoretical perspective for analyzing the impact of algorithmic management on the work-family interface. Specifically, Study 1 relies on the dissipative structural characteristics of the human brain and constructs a linear impact model of algorithmic management on work-family facilitation from the perspective of System 1; Study 2 draws on the dissipative structural characteristics of the individual and constructs a nonlinear impact model of algorithmic management on work-family conflict from the perspective of System 2. This study has three innovative points. First, this study combines the dual-process model and dissipative structure theory, enriching the theoretical perspective of algorithmic management, expanding the outcome variable network of algorithmic management as well as the antecedent variable network of the work-family interface. Existing research on algorithmic management mainly analyzes its impact on gig workers from the perspectives of the job demands-resources model, self-determination theory, and cognitive appraisal theory. It has not yet incorporated the theoretical perspective of the dual-process model and dissipative structure theory. This study identifies the dissipative structural characteristics of the human brain and the individual, and regards algorithmic management as a key factor triggering fluctuations within the individual cognitive system. Integrating the processing pathways of System 1 and System 2 in the dual-process model, this study analyzes the complex impact of algorithmic management on work-family facilitation and work-family conflict, further enriching the theoretical perspective of algorithmic management. In addition, previous studies have paid limited attention to the relationship between algorithmic management and the work-family interface. This study introduces the work-family interface into the outcome network of algorithmic management, thereby expanding the research boundaries in this field. Second, this study uses the dissipative structural characteristics of the human brain to construct a moderated mediation model of the linear impact of algorithmic management on work-family facilitation. Previous studies have lacked an exploration of the mechanism through which algorithmic management affects work-family facilitation, as well as an examination of variables related to individual attention. Based on dissipative structure theory, this study considers the dissipative structural characteristics of the human brain, analyzes the pathway through which algorithmic management enhances work-family facilitation by improving gig workers’ work concentration and work goal progress, and uses gig workers’ familiarity with platform algorithms as a moderator to identify the boundary conditions under which algorithmic management exerts a positive effect on work-family facilitation among gig workers. Third, this study uses the individual’s own dissipative structural characteristics to construct a moderated mediation model of the nonlinear impact of algorithmic management on work-family conflict. Previous research on algorithmic management has not paid attention to the dissipative structural characteristics of gig workers themselves. Based on the dual-process model, this study uses the individual’s own dissipative structural characteristics to analyze the pathway through which algorithmic management nonlinearly affects work-family conflict via psychological entropy and uncertainty perception, and combines gig workers’ self-management level to explain the conditions under which algorithmic management’s nonlinear impact on work-family conflict operates. This study helps to reveal the complex impact of algorithmic management on gig workers and can also help to fill the gap left by previous studies that neglected the dissipative structural characteristics of gig workers themselves. In addition to the above three innovations, this study applies the dual-process model to research on algorithmic management, thereby enriching the application scenarios of the theory in the digital economy era. The findings also offer practical implications for society, enterprises, and gig workers to help them adapt to the algorithmic society and better balance the relationship between work and family. References | Related Articles | Metrics

Select

The mechanisms of human-AI collaboration on hybrid problem-solving in social entrepreneurship: Based on the attention-based view QIU Zhenyu, TAO Qin, ZHAO Chenfang, LIU Zhiyang 2026, 34 (11):  1949-1971.  doi: 10.3724/SP.J.1042.2026.1949 Abstract ( 54 )   PDF (1003KB) ( 37 )   Peer Review Comments Social entrepreneurship is increasingly recognized for its dual mission of achieving social value and economic sustainability. However, this dual mission inherently generates "hybrid problems" stemming from the deep-seated conflicts between social and economic logics. These hybrid problems manifest in three primary contradictions: hybrid value propositions, hybrid resource utilization, and hybrid governance. Consequently, social entrepreneurs face significant cognitive tension and a severe scarcity of attentional resources. While human-AI collaboration has emerged as a revolutionary paradigm for complex problem-solving, its specific mechanisms within the highly contextualized and hybrid environment of social entrepreneurship remain underexplored. Drawing on the Attention-Based View (ABV) and cognitive psychology, this research constructs an integrated theoretical framework of "human-AI collaboration—attentional engagement—hybrid problem-solving" to systematically investigate how different modes of human-AI collaboration reshape entrepreneurs' attentional allocation and subsequently influence hybrid problem-solving outcomes. We posit that human-AI collaboration fundamentally alters the distribution of cognitive load. By offloading specific cognitive tasks to artificial intelligence, entrepreneurs can release and reallocate their limited attentional resources toward higher-order strategic and ethical judgments. We conceptualize attentional engagement into two distinct dimensions: sustained attention engagement and alternating attention engagement. Based on the division of labor, we identify three distinct modes of human-AI collaboration—AI-dominated, human-dominated, and interactive—each employing different problem-solving strategies and generating unique attentional perspectives. First, the AI-dominated collaboration mode employs a solution-focused strategy, wherein AI autonomously defines the problem and generates solutions, leaving the entrepreneur to filter and optimize the results. This mode triggers a bottom-up attentional perspective. By assuming the heavy cognitive load of information processing and solution generation, AI significantly enhances the entrepreneur's sustained attention engagement. This sustained focus allows the entrepreneur to concentrate on long-term strategic alignment and ethical evaluation, thereby optimizing data-driven resource utilization efficiency and facilitating the resolution of highly structured hybrid problems. The effectiveness of this mode is moderated by human-AI factors, such as the entrepreneur's domain expertise, AI explainability, and human-AI trust, as well as the problem's degree of hybridity and structuredness. Second, the human-dominated collaboration mode utilizes a problem-focused strategy. Here, AI provides a preliminary structural framework for the complex social problem, and the entrepreneur subsequently searches for and formulates solutions based on this framework. This mode cultivates a top-down attentional perspective. The AI-generated structural framework effectively reduces the entrepreneur's short-term information processing load, thereby strengthening their alternating attention engagement. Enhanced alternating attention enables the entrepreneur to flexibly switch between competing economic and social goals, facilitating a dynamic balance among multiple stakeholder demands. The efficacy of this mode is contingent upon the entrepreneur's cognitive style, emotional state, and the degree of human-AI cognitive alignment, along with the specific characteristics of the hybrid problem. Third, the interactive human-AI collaboration mode adopts a problem-solution pair strategy, characterized by continuous, dynamic interaction and mutual feedback between the human and AI throughout the problem-solving process. This mode fosters an interactive attentional perspective, promoting dual attentional synergy that activates both sustained and alternating attention engagement. Through continuous cognitive alignment and adaptive feedback, this interactive mode empowers social entrepreneurs to manage high uncertainty and complex value conflicts, ultimately leading to the creative integration of social and economic values. The success of this interactive mechanism is moderated by the depth of AI participation, the level of human-AI trust, and the frequency of interaction. Theoretically, this research advances the Attention-Based View by extending it from the macro-organizational level to the micro-cognitive level of individual entrepreneurs, integrating it with cognitive psychology concepts of vigilance and task-switching. Furthermore, it deepens the understanding of hybrid problem-solving by illuminating the specific cognitive mechanisms through which AI interventions alleviate attentional scarcity in multi-objective conflict scenarios. Practically, the findings offer social entrepreneurs cognitively empowering pathways for effective human-AI collaboration, providing actionable guidelines for the strategic adoption of AI technologies to navigate the complex landscape of social entrepreneurship. References | Related Articles | Metrics

Select

“AI flavor” in advertising: The measurement dimensions, judgment cues, and double-edged sword effect ZHOU Zhimin, Aisihaer Nadilai, ZHANG Shaojie 2026, 34 (11):  1972-1985.  doi: 10.3724/SP.J.1042.2026.1972 Abstract ( 75 )   PDF (643KB) ( 81 )   Peer Review Comments Generative AI has revolutionized advertising production, yet a pervasive consumer complaint—“AI flavor”—has emerged. Consumers spontaneously perceive ads as AI-generated based on stylistic features such as unnaturally smooth visuals, mechanical language, rigid movements, and emotional shallowness, even when those ads were human-made. Existing research, however, relies almost exclusively on explicit AI disclosure paradigms and overlooks this natural inference process, while also emphasizing negative reactions without considering potential positive outcomes. To address these gaps, we introduce AI Flavor in Advertising (AIFIA), defined as the degree to which consumers perceive an ad as exhibiting prototypical AI-generated characteristics, regardless of its actual source. We develop an integrative framework with three studies on conceptualization, judgment cues, and double-edged brand effects. Study 1 formalizes AIFIA as a multidimensional construct comprising stiffness and fantasticality, both grounded in perceived deviation from consumers’ cognitive representations of advertising. Stiffness reflects downward deviation from naturalness—visual artifacts, mechanical language, rigid actions, and emotional void—that violates expectations of human-crafted authenticity. Fantasticality denotes upward deviation, wherein hyper-realistic or surreal visuals exceed everyday experience and appear achievable only through AI, thus tapping into consumers’ beliefs about AI’s “magical” creative power. Either dimension can independently trigger AIFIA. Following scale development protocols, we generate items via literature review, netnography, interviews, and expert evaluation, then establish validity through exploratory and confirmatory factor analyses. AIFIA is distinguished from advertising falsity, AI hallucination, and AI salience, clarifying its unique status as a consumer-centric perceptual variable. Study 2 identifies judgment cues and moderators. Drawing on cue diagnosticity and consistency theories, we propose objective cues—smooth visuals, mechanical language, rigid movements, detail errors, and saturated colors—and subjective cues—illogical structure, insufficient creativity, and emotional absence. These cues signal AI involvement when they diverge from expected naturalness. Their effects are moderated by AI literacy and critical thinking: high levels amplify cue-AIFIA relationships by enhancing sensitivity and analytical processing, whereas low levels attenuate them. Beyond isolated cue effects, we propose using fsQCA to examine configurational patterns, recognizing that real-world AIFIA judgments often arise from specific combinations of cues rather than any single attribute, an approach that can reveal multiple equifinal pathways to strong AIFIA perceptions. Study 3 articulates the double-edged effects on brand perceptions. High AIFIA reduces brand authenticity, perceived brand engagement, and increases homogenization—negative effects driven by cue inconsistency with consumers’ expectations of humanized, emotionally invested brand narratives. Simultaneously, high AIFIA enhances brand innovativeness, technological image, and rejuvenation—positive effects particularly when fantasticality dominates, as technical sophistication signals modernity and advanced capabilities. Two moderators shape these effects. AI disclosure interacts with AIFIA: when disclosure is present but AIFIA is low, inconsistency arises, mitigating negatives and amplifying positives; when no disclosure meets high AIFIA, inconsistency suggests deception, intensifying negatives and weakening positives. Brand personality also moderates: for sincere brands, high AIFIA (especially stiffness) conflicts with warmth, impairing perceptions; for exciting brands, high AIFIA (especially fantasticality) aligns with novelty, enhancing evaluations. Finally, advertising legitimacy—comprising intentional, strategic, and practical dimensions—mediates these moderated effects, as consumers assess whether the ad's purpose is credible, its strategy is coherent, and its execution is socially appropriate based on congruence between AIFIA and contextual signals; this legitimacy appraisal ultimately drives favorable or unfavorable brand inferences. Theoretically, this research offers three innovations. It introduces and validates AIFIA, shifting from disclosure-based to perception-based research. It systematically delineates objective and subjective cues along with moderators, providing the first comprehensive account of AIFIA formation. It reveals the double-edged impact on brand perceptions, demonstrating divergent pathways and contextual dependencies, with legitimacy as the explanatory mechanism. These contributions open new research avenues at the intersection of AI technology, advertising, and consumer psychology, while offering actionable guidance for brand managers to strategically calibrate AI flavor—whether to amplify its modernizing appeal or attenuate its alienating effects—depending on brand positioning and disclosure decisions. References | Related Articles | Metrics

Meta-Analysis

Select

Bidirectional associations between peer exclusion and depressive symptoms in children and adolescents: A meta-analysis of longitudinal studies YAN Yajing, ZHANG Yali 2026, 34 (11):  1986-2002.  doi: 10.3724/SP.J.1042.2026.1986 Abstract ( 90 )   PDF (776KB) ( 73 )   Peer Review Comments The causal relationship between peer exclusion and depressive symptoms in children and adolescents remains controversial. Three competing theoretical models offer divergent predictions. The stress process model views peer exclusion as a chronic interpersonal stressor that triggers depressive symptoms. The symptom-driven model argues that depressive symptoms actively elicit peer exclusion. The transactional model proposes a bidirectional vicious cycle. Despite the theoretical importance of clarifying this relationship, existing empirical findings remain inconsistent. Moreover, no prior meta-analysis has systematically synthesized longitudinal evidence using meta-analytic structural equation modeling (MASEM). To address these gaps, a systematic literature search was conducted across eight Chinese and English databases: Web of Science Core Collection, Medline, Elsevier, EBSCO-ERIC, PsycINFO, PsycArticles, ProQuest Dissertations & Theses Global, and China National Knowledge Infrastructure. The search had no start date restriction and a cutoff date of October 18, 2025. A total of 23 longitudinal studies (26 independent samples, N = 20,676) met the inclusion criteria. Two independent reviewers performed study screening, data coding, and quality assessment using the Joanna Briggs Institute Critical Appraisal Checklist, with inter-rater agreement exceeding 95%. One-stage MASEM was conducted via the webMASEM platform using full information maximum likelihood estimation with random-effects models. Publication bias analyses indicated minimal bias. Egger‘s regression was not significant (p = 0.96). The p-curve analysis exhibited significant right-skewness. Trim-and-fill correction yielded negligible changes in the pooled effect size. The main effect analysis revealed significant bidirectional associations. Peer exclusion at T1 significantly predicted subsequent depressive symptoms at Tn (β = 0.06, p = 0.04). Depressive symptoms at T1 significantly predicted subsequent peer exclusion at Tn (β = 0.09, p = 0.02). These findings support the transactional model. Moderator analyses revealed several notable findings. Cultural background significantly moderated the depressive symptoms → peer exclusion path (β = -0.03, p = 0.09, marginal) and the autoregressive path of depressive symptoms (β = -0.06, p = 0.02). Higher individualism scores were associated with weaker predictive effects and lower stability of depressive symptoms. Peer exclusion reporting method significantly moderated the depressive symptoms → peer exclusion path (β = -0.21, p = 0.003). Self-report measures yielded stronger effects than other-report measures. Publication year showed a marginally significant positive moderating effect (β = 0.06, p = 0.07), suggesting that this effect may have strengthened over time. Time interval negatively moderated both autoregressive paths. In contrast, mean age and sex ratio showed no significant moderation. This study preliminarily reconciles the theoretical debates among the stress process model, the symptom-driven model, and the transactional model. It confirms that peer exclusion and depressive symptoms in children and adolescents have a bidirectional predictive relationship. The two constructs form a vicious cycle, providing meta-analytic evidence for the transactional model. The findings suggest that mental health interventions should shift from single-path approaches to dual-track strategies. Practitioners should reduce the occurrence and psychological harm of peer exclusion while simultaneously improving the social interaction patterns of depressed individuals, thereby interrupting the mutual reinforcement between the two. Meanwhile, cultural background significantly moderates the effect of depressive symptoms on subsequent peer exclusion and the stability of depressive symptoms. In collectivistic cultures, interventions should emphasize group support, family involvement, and destigmatization education. In individualistic cultures, greater emphasis can be placed on promoting active help-seeking and emotional self-regulation. Moreover, reporting method significantly affects the effect size of depressive symptoms predicting peer exclusion, suggesting that future research should combine self-report and other-report measures. Finally, the marginal positive moderating effect of publication year on the depressive symptoms → peer exclusion path suggests that the digital social era may have amplified the interpersonal consequences of depressive symptoms. Future efforts should guide adolescents to use the internet wisely, balance online and offline social activities, and prevent depressed individuals from falling further into social isolation due to excessive reliance on digital interaction. References | Related Articles | Metrics

Select

Antecedents and consequences of consumers' perception of product naturalness: A meta-analysis LIU Wumei, LI Yue, LI Lingbo, ZHANG Haitao 2026, 34 (11):  2003-2018.  doi: 10.3724/SP.J.1042.2026.2003 Abstract ( 65 )   PDF (726KB) ( 32 )   Peer Review Comments In the post-pandemic consumer market, heightened attention to environmental quality, personal health, and product safety has fueled rapid global growth in demand for natural products. However, existing research on Perceived Product Naturalness (PPN)—consumers’ subjective judgment that a product originates from nature, contains no non-natural ingredients, and has undergone minimal processing—still has notable shortcomings. The antecedents of PPN are highly fragmented and lack a unified, theoretically grounded classification framework. Divergences remain regarding the relative effectiveness of different naturalness cues, such as explicit claims, sensory signals, and production methods. Moreover, the strength of PPN’s impact on consumer responses and its boundary conditions have yet to be systematically elucidated. To address these gaps, this study conducts a meta-analysis of 229 independent effect sizes from 59 empirical studies published over the past two decades, aiming to construct an integrated theoretical framework encompassing the antecedents, consequences, and moderators of PPN. Drawing on dual-system theory and cue diagnosticity theory, all naturalness cues are classified into two types: heuristic cues and systematic cues. Heuristic cues are simple peripheral signals that consumers can process quickly, including sensory information, explicit natural claims, and natural ambiance cues in the consumption environment. Systematic cues are more diagnostic, product-intrinsic information requiring deeper cognitive processing, including ingredient composition, degree of processing, and raw material traceability. This classification offers a new analytical framework for understanding how different cues shape consumers’ perceptions of naturalness. The results indicate that both heuristic and systematic cues significantly enhance PPN, yet the overall effect of systematic cues is stronger, suggesting that consumers are persuaded not only by superficial symbolic signals but more so by verifiable, product-based evidence. Notably, within heuristic cues, sensory information and consumption environment cues outperform explicit brand natural claims, a finding that reconciles previous disagreements. Against a backdrop of frequent greenwashing scandals and declining trust in corporate communication, consumers have become skeptical of unsubstantiated natural claims, whereas embodied sensory experiences and natural settings provide more immediate and credible perceptions. Further analysis shows that PPN positively influences consumer attitudes, choices, and behavioral intentions, but its effect on attitudinal outcomes is significantly stronger than on behavioral outcomes. This indicates that perceived naturalness first operates at the evaluative level, with its translation into actual purchase behavior constrained by factors such as price, availability, and efficacy expectations. The study also identifies boundary conditions at the consumer, product, and situational levels: the effect of PPN is stronger for food and utilitarian products, for samples with a higher proportion of women, and under conditions of high environmental threat; the effect intensifies significantly after major public health events and is more pronounced among younger consumer groups. This research makes several contributions. Theoretically, it resolves the fragmentation of PPN antecedents by establishing a clear dual-cue classification framework, reconciles prior divergences by quantitatively comparing the relative effectiveness of different cues, and extends boundary condition research from individual and product levels to the macro-situational level by introducing environmental threat as a situational moderator. Practically, the findings offer guidance for natural marketing strategies: companies should prioritize investment in verifiable systematic cues (e.g., short ingredient lists, transparent production processes) over vague natural slogans, strategically allocate resources to health-related product categories, and strengthen communication on safety and ingredient transparency when environmental or health threats intensify. References | Related Articles | Metrics

Regular Articles

Select

Computational mechanisms of attention in value-based decision making: Dynamic regulation of additive and multiplicative models ZHOU Yuxuan, YANG Yilin, HUANG Jianping 2026, 34 (11):  2019-2031.  doi: 10.3724/SP.J.1042.2026.2019 Abstract ( 71 )   PDF (1160KB) ( 76 )   Peer Review Comments In value-based decision making, individuals must sample, compare, and integrate information from different options and their attributes under constraints of limited time and cognitive resources. As a critical mechanism linking information processing to choice formation, attention has become central to understanding how evidence is accumulated during decision making. Existing accounts have proposed two major computational explanations. The multiplicative model suggests that attention enhances the weight of attended information, thereby amplifying its contribution to evidence accumulation. In contrast, the additive model proposes that attention provides an additional directional bias toward the currently attended option, exerting an influence that is relatively independent of its subjective value. However, these two mechanisms have often been conceptualized as competing explanations, making it difficult to account for inconsistent attentional effects observed across different experimental contexts. Here, we argue that attentional processing in value-based decision making is inherently dynamic and hierarchical. Rather than exerting a fixed computational influence throughout the decision process, attention may serve different computational functions depending on the progression of decision formation and the level of information representation being processed. By systematically integrating evidence from behavioral studies, computational modeling, and neuroscience, we propose a “time × information hierarchy” framework to explain how attention dynamically regulates evidence accumulation through two complementary computational principles: value amplification and directional bias. Specifically, we propose that attentional influences on value-based decisions do not rely on a single mechanism, but instead exhibit a gradual transition from multiplicative to additive processes over the course of decision formation. During the early stage of decision making, individuals face substantial uncertainty and engage in broad information sampling and value comparison across multiple attributes and options. At this stage, attention primarily operates on lower-level information representations, including attribute-level features (e.g., price, reward, and health value) and option-level subjective value representations. By increasing the relative weight of relevant value signals entering the evidence accumulation process, attention enhances the impact of high-value or goal-relevant information, consistent with a multiplicative value-gain mechanism. This account aligns with the attentional drift diffusion model (aDDM), which proposes that attention influences choice by modulating the relative contribution of attended and unattended information during evidence accumulation. As decision formation progresses, uncertainty within the decision system gradually decreases, and information processing shifts from broad exploration toward focused comparison among dominant options. We propose an intermediate filtering stage to characterize this transition from open-ended information sampling to choice convergence. Importantly, this stage does not represent a fixed temporal window or an abrupt switch between two mechanisms. Instead, it reflects a dynamic computational state in which both mechanisms may coexist. During this period, multiplicative value amplification continues to contribute to evidence accumulation, while attentional effects increasingly involve directional reinforcement toward currently favored options. During the late stage of decision making, when accumulated evidence approaches the decision threshold, processing priorities shift from value comparison toward evidence integration and choice commitment. At this stage, attentional effects are more likely to exhibit an additive pattern. Rather than continuously amplifying the strength of value signals, attention may provide an additional directional bias that facilitates the convergence of evidence toward a particular choice and supports final decision commitment. Importantly, we further emphasize that late-stage additive effects should not be attributed exclusively to attention itself. Such effects may also reflect the joint contribution of attentional allocation, decision commitment, action preparation, and urgency-related processes. Therefore, late-stage attentional influences should be understood as the outcome of interactions among attentional mechanisms, evolving decision states, and behavioral output processes. The theoretical contribution of this framework does not lie in proposing the first computational integration of multiplicative and additive mechanisms. Existing models, such as the Gaze-weighted Linear Accumulator Model (GLAM), have already incorporated both value weighting and directional bias components. Instead, the present framework advances existing theories by explaining why these two mechanisms may dominate under different conditions through a temporal and hierarchical perspective. We propose that multiplicative and additive mechanisms are not mutually exclusive alternatives, but rather represent distinct computational functions emerging at different stages and levels of information processing within the same decision system. By establishing the “time × information hierarchy” framework, this review provides a unified account of existing behavioral, computational, and neural findings and offers a new perspective for understanding how attention dynamically shapes value representation, evidence accumulation, and choice formation. References | Related Articles | Metrics

Select

Construction of an integrated model for the influence mechanisms of emotional flexibility WANG Xuan, QIU Xiaodan, YOU Xuqun 2026, 34 (11):  2032-2048.  doi: 10.3724/SP.J.1042.2026.2032 Abstract ( 88 )   PDF (810KB) ( 360 )   Peer Review Comments Emotional flexibility is defined as individuals’ core adaptive capacity to dynamically adjust emotional responses in accordance with changing situational demands. Although existing studies have extensively explored the beneficial outcomes of diverse emotion regulation strategies and their underlying proactive processing mechanisms, they have largely neglected the independent and interactive influences of multiple predictive factors on emotional flexibility across pre-processing and online processing stages. Accordingly, the current literature presents three prominent research limitations. First, prior studies primarily focus on in-processing influencing factors while ignoring the vital role of pre-processing antecedent variables in shaping emotional flexibility. Second, most research regards emotion regulation strategies as the core of emotional flexibility and underestimates the essential function of contextual factors, despite the fact that context serves as a fundamental driving condition for flexible emotional adjustment based on the intrinsic connotation of emotional flexibility. Third, previous theoretical frameworks fail to integrate influencing factors across multi-dimensional temporal and spatial scales. Specifically, individual emotional flexibility is shaped not only by immediate situational experiences and different developmental stages throughout life but also by broad familial and social environments, as well as key interpersonal influences such as parental upbringing. To address these research gaps, the present study incorporates both pre-processing and in-processing factors and adopts a comprehensive temporal and spatial perspective, covering macro-level lifespan development and micro-level momentary stimulus processing in temporal dimension, as well as social, familial, and situational environments in spatial dimension, to construct a systematic integrated model of the mechanisms underlying emotional flexibility. This study innovatively establishes a unified theoretical model of emotional flexibility by synthesizing antecedent factors (individual and environmental factors) and in-processing factors (cognitive components, situational characteristics, and stimulus characteristics). A series of sub-models are separately constructed to clarify the influencing pathways of personality traits, psychological capital, family factors, social factors, cognitive components, situational features, and stimulus characteristics, which are further integrated into a holistic theoretical system. The model reveals that the dynamic interaction between individual and environmental factors acts as a critical antecedent driver of emotional flexibility. Family and social environments shape individual personality traits and psychological capital, which further promote individuals’ effective perception and comprehension of complex situational information; such cyclical reciprocal interactions continuously facilitate the development and improvement of emotional flexibility. In experimental processing of emotional flexibility, cognitive components, situational features, and emotional stimulus characteristics jointly dominate real-time emotional adjustment processes. Affective representation, contextual and facial information processing, and cognitive empathy constitute the basic cognitive components of emotional flexibility. When shifts in stimulus emotional valence or valence conflicts between stimuli and situational contexts occur, additional error monitoring processing will be activated, significantly increasing the cognitive load of emotional flexible adjustment. Furthermore, the time interval between contextual priming stimuli and emotional target stimuli produces preparatory effects, while the presentation order of contextual cues and situational response modes also substantially regulate individual emotional flexibility levels. This study provides significant theoretical and practical contributions. Theoretically, it refines and complements the theoretical framework of emotional flexibility’s influencing mechanisms, revealing the dynamic shaping rules of emotional flexibility under broad temporal and spatial dimensions. Practically, it offers reliable theoretical support for educational cultivation and clinical intervention. In educational scenarios, it provides systematic strategies for the coordinated cultivation of individual emotional flexibility among families, schools, and society. In clinical practice, it offers empirical references for the prevention and auxiliary treatment of psychological disorders and mental illnesses caused by deficient emotional flexibility. References | Related Articles | Metrics

Select

The intervention effects and mechanisms of exergaming on executive function in children and adolescents LIU Hanhui, WU Ji, GE Xuejing, LI Hui-Jie 2026, 34 (11):  2049-2065.  doi: 10.3724/SP.J.1042.2026.2049 Abstract ( 96 )   PDF (3148KB) ( 182 )   Peer Review Comments Exergaming-a form of interactive digital game requiring physical movement—has emerged as a promising tool for improving executive functions (EFs) in children and adolescents. While previous meta-analyses have quantitatively synthesized the overall effectiveness of exergaming interventions, they have paid limited attention to how exergaming works (i.e., underlying mechanisms), for whom it works best, and under what conditions (i.e., moderators). This paper provides a comprehensive review and proposes a novel integrative framework to address these gaps. A key innovation of this review is the proposal of a Triple‑Path Synergistic Hypothesis Model, which posits that exergaming enhances EFs through the dynamic interaction of three mechanistic pathways: (1) physiological arousal (e.g., increased heart rate, cerebral blood flow, and release of brain‑derived neurotrophic factor), (2) cognitive engagement (e.g., dual‑task demands, rule switching, and adaptive cognitive load), and (3) neural plasticity (e.g., enhanced prefrontal connectivity and experience‑dependent structural reorganization). Unlike prior work that has treated these mechanisms in isolation, our model explicitly integrates them and explains previously inconsistent findings. For example, the model resolves the apparent contradiction between studies showing that acute EF improvements depend primarily on exercise intensity versus those showing a primary role of cognitive engagement: in acute interventions, neural plasticity has not yet developed and stable cognitive strategies are absent, so benefits rely largely on physiological arousal; in contrast, improvements in higher‑order functions such as cognitive flexibility require activation of specific frontoparietal networks, making cognitive engagement indispensable. A second major contribution is the systematic articulation of a multi‑level moderator framework spanning individual characteristics (age, baseline EF, body mass index), game attributes (cognitive load, exercise intensity, difficulty adaptation, interaction mode), intervention parameters (duration, frequency, period), and environmental settings (school, home, laboratory). This framework provides a unified explanation for the substantial heterogeneity observed across studies. For instance, intervention effects are larger in children with lower baseline EF (“low‑baseline, high‑gain” pattern), competitive game formats outperform cooperative ones, and younger children may benefit more from physiological arousal whereas adolescents gain more from cognitive challenge. The review also identifies critical evidence gaps that shape future research priorities. First, most existing exergames are commercially developed for entertainment and lack systematic, adaptive cognitive loading targeting specific EF components. Second, dose-response relationships remain poorly understood; current studies vary widely in intervention parameters (1-12 weeks, 1-5 sessions/week, 10-60 min/session), and no consensus exists on optimal intensity, duration, or frequency for different age groups or EF subdomains. Third, external validity is limited due to small sample sizes, restricted sampling (mostly from specific schools or regions), and a near‑absence of studies in naturalistic home environments. Fourth, mechanistic evidence is fragmented; few studies include both pure cognitive training and pure physical exercise control groups, making it difficult to disentangle the unique contributions of cognitive versus physical components. Finally, longitudinal follow‑ups are rare, leaving the long‑term sustainability of EF gains unknown. To advance the field, we propose five future directions: (1) developing personalized, adaptive exergames based on real‑time assessment of individual cognitive and physical states; (2) establishing dose-response models for acute, short‑term, and long-term interventions; (3) integrating immersive virtual reality technologies to enhance engagement and ecological validity; (4) conducting large‑scale, multi‑site, and home-based trials with representative samples; and (5) testing the Triple‑Path Synergistic Hypothesis Model using multi‑modal neuroimaging (e.g., fMRI, fNIRS, EEG) and rigorous experimental designs (including pure cognitive and pure physical control groups). By shifting the research question from “whether exergaming works” to “how, for whom, and under what conditions it works”, this framework aims to guide the development of scientifically grounded, precisely targeted exergaming interventions for cognitive health promotion in children and adolescents. References | Related Articles | Metrics

Select

How individuals decide to cheat: Processes and mechanisms QIN Hengjie, ZHAO Li, LI Songze 2026, 34 (11):  2066-2085.  doi: 10.3724/SP.J.1042.2026.2066 Abstract ( 68 )   PDF (706KB) ( 84 )   Peer Review Comments Cheating, broadly defined as the deliberate violation of established rules or norms to secure undue advantage or personal gain, poses a profound threat to social equity and public trust. Understanding the cognitive architecture that underpins cheating decisions is therefore essential for devising targeted and effective integrity-promoting interventions. Synthesizing insights from extant theoretical accounts and recent empirical findings, this article proposes a comprehensive process model that decomposes the cheating decision into four sequential yet interactive components: benefit appraisal, risk assessment, moral justification, and behavioral execution. This framework specifies the temporal ordering of cost-benefit calculations, perceived controllability of negative outcomes, rational moral reasoning, and final action selection, thereby delineating how individuals successively address four pivotal questions during the decision cascade: (1) Is the prospective gain from cheating sufficiently compelling? (2) Are the likelihood and consequences of detection acceptable and manageable? (3) Can the contemplated transgression be cognitively and morally rationalized to preserve a positive self-image? (4) On the basis of the preceding evaluations, does the individual ultimately commit to the cheating act? Critically, these four phases are not merely linear but exert reciprocal influences on one another, jointly capturing the full trajectory from motivational inception to overt behavioral manifestation, while each phase contributes unique variance to the overall decisional outcome. Building on this theoretical foundation, the present article further integrates neurophysiological mechanisms and modulatory factors into the model, mapping each decision stage onto distinct but overlapping brain regions. Such neural grounding not only provides independent convergent evidence for the stage-wise partition at the biological level but also lays the groundwork for future neuromodulation techniques aimed at specific neural substrates to curb cheating propensity. Moreover, by systematically reviewing how diverse contextual, dispositional, and social regulatory variables differentially operate across the four stages, we argue that effective integrity governance must move beyond generic moral exhortation toward stage-specific, precision-oriented strategies that address the unique vulnerabilities inherent in each phase of the decision process. Finally, we outline three promising avenues for future research that will refine and extend the current model. First, at the neural level, we advocate for the adoption of dual-brain neuroscientific approaches—such as hyperscanning and inter-brain synchrony analyses—to capture the dynamic interpersonal and interactive neural processes that characterize cheating in social contexts, thereby moving beyond isolated intra-individual accounts. Second, regarding developmental trajectories, systematic investigations spanning from childhood to adulthood are urgently needed to map age-related shifts in the weighting and integration of the four decision components, with particular attention to critical developmental turning points. Such evidence would furnish an empirical basis for age-appropriate integrity education and early intervention programs. Third, in the era of artificial intelligence (AI), emerging challenges—including the increasing sophistication of AI-assisted cheating, the erosion of personal accountability due to algorithmic mediation, and the diffusion of responsibility in human-AI collaborative environments—demand novel theoretical and regulatory responses. We therefore call for adaptive updates to the model that accommodate these unprecedented contexts, ensuring its relevance and applicability in digitally transformed societies. In sum, this article offers an integrative and neurocognitively grounded framework for understanding cheating decisions as a multistage process, while simultaneously providing actionable guidelines for policy makers, educators, and technologists. By pinpointing the distinct cognitive and neural junctures at which interventions can be most effective, and by anticipating future challenges posed by technological advancement and developmental diversity, our framework aspires to inform a new generation of integrity-building initiatives that are both forward-looking and precisely calibrated. Ultimately, we contend that such a refined, evidence-based approach will not only enrich theoretical discourse on moral decision-making but also contribute substantively to the design of resilient integrity systems capable of adapting to the complexities of the modern digital age. References | Related Articles | Metrics

Select

The promoting effect of shared attention on interpersonal relationships and its underlying mechanisms YUAN Shengfeng, LIU Juntong, XIA Yi, ZHANG Yarui, LEI Yi 2026, 34 (11):  2086-2108.  doi: 10.3724/SP.J.1042.2026.2086 Abstract ( 84 )   PDF (689KB) ( 70 )   Peer Review Comments Shared attention refers to a state in which individuals attend to an object while perceiving that others are simultaneously attending to it, thereby generating a momentary attentional experience from a first-person-plural perspective. Previous research has shown that shared attention influences cognitive processing, affective experience, and motivational states. However, its promoting effect on interpersonal relationships and the underlying mechanisms have not yet been systematically integrated. Focusing on the core question of how shared attention promotes interpersonal relationship improvement, the present article systematically reviews the relevant empirical and theoretical literature and, on this basis, proposes an Interpersonal Effects Model of Shared Attention. Existing evidence suggests that the promoting effect of shared attention on interpersonal relationships is manifested across multiple levels, including cognitive, affective, and motivational-behavioral domains. Rather than influencing only a single type of social judgment, shared attention may systematically promote improvement across multiple components of interpersonal relationships. The present article integrates the mechanisms through which shared attention promotes interpersonal relationship improvement into three interrelated pathways. The cognitive pathway emphasizes that shared attention supports the formation of common knowledge and common ground, thereby reducing recursive uncertainty about others’ states of knowing and providing a cognitive basis for perceived understanding, trust, self-other overlap, and reduced social distance. The affective pathway emphasizes that shared attention promotes interpersonal affective experiences such as intimacy, emotional bonding, and emotional closeness by optimizing emotional experience, enhancing interpersonal attraction, and reducing the psychological cost of empathy. The motivational-behavioral pathway emphasizes that shared attention strengthens approach motivation, relationship commitment, and relationship-maintenance tendencies by supporting the structural cognitive conditions required for coordination and cooperation, and further enhances the attractiveness of interaction partners through the immediate satisfaction of relational needs. Relevant neuroscientific research provides neural-level support for these psychological pathways. On this basis, the present article proposes an Interpersonal Effects Model of Shared Attention. The model makes three core contributions. First, it extends the explanatory target of shared attention from immediate psychological effects to interpersonal relationships as structured and cumulative outcomes. Second, it emphasizes that the cognitive, affective, and motivational-behavioral pathways are not independent of one another under shared attention, but instead jointly promote interpersonal relationship improvement through cross-pathway coordination. Third, it proposes a dynamic cycle of relational gains, whereby shared attention can promote interpersonal relationship improvement, while improved relationships may in turn facilitate the subsequent formation and maintenance of shared attention and amplify its effects, thereby fostering long-term relational development. The present article further argues that the promoting effect of shared attention on interpersonal relationship improvement does not occur unconditionally, but is constrained by a series of boundary conditions. First, whether shared attention can be formed at all depends on whether the interacting parties genuinely enter a shared attentional state of “we are attending” rather than merely achieving objective attentional alignment. Second, whether the shared content can be jointly integrated also influences whether shared attention can be further translated into positive interpersonal consequences; when the shared content is difficult to understand or accommodate jointly, or elicits strong psychological defensiveness, its relationship-promoting effect may be weakened. Finally, whether the positive effects of shared attention can consolidate into relatively stable interpersonal relationship improvement also depends on whether the interaction is sufficiently repeated and stable over time. On the basis of these boundary conditions, the Interpersonal Effects Model of Shared Attention proposed in this article has strong testability and can be empirically examined across different contexts, relationship types, and time scales. In practical terms, the Interpersonal Effects Model of Shared Attention has important implications for promoting interpersonal relationships in domains such as education, virtual reality, parent-child interaction, older adults’ social engagement, and clinical intervention. It not only helps explain how cognitive benefits and relational improvement may emerge simultaneously within the same interactional structure, but also provides a new theoretical perspective for examining the links among co-presence, synchronous co-attention, and relational change. Overall, by systematically integrating previously scattered findings, the present article reveals how shared attention may gradually develop from a short-term interpersonal interaction experience into deeper interpersonal relationship improvement, and provides a theoretical framework for future research and practical applications in related fields. References | Related Articles | Metrics

Research Method

Select

Methods for testing and learning attribute hierarchical structures from a latent variable modeling perspective MAO Xiuzhen, HUANG Linchao, XU Binghui, ZHENG Qiaoyu 2026, 34 (11):  2109-2120.  doi: 10.3724/SP.J.1042.2026.2109 Abstract ( 53 )   PDF (718KB) ( 172 )   Peer Review Comments Attribute hierarchical structure (AHS) characterizes the prerequisite relationships among attributes. It plays an important role in diagnostic test design, instructional design and assessment, learning path planning, and personalized learning recommendation. Latent variable models serve as the foundation for parameterized approaches to AHS analysis, as they determine the objectives, analytical perspectives, and implementation strategies of these methods. From the perspective of latent variable modeling, this study systematically reviews methods for estimating and testing AHS according to three methodological perspectives: hypothesis testing, Bayesian structure learning and parameter estimation. Specifically, two major frameworks have been developed for testing AHS. The first evaluates AHS by comparing the model fit of nested hierarchical structures, including the likelihood ratio test (LRT), parametric bootstrap LRT, and non-parametric bootstrap LRT. The second tests AHS by examining whether the structural parameters corresponding to all impossible attribute mastery patterns under a prespecified AHS are significantly equal to zero, primarily using the Wald test. The empirical distribution of the LRT statistic often deviates substantially from its asymptotic distribution. Although parametric and non-parametric bootstrap LRTs improve statistical inference, they are computationally intensive in large-scale applications. In contrast, the Wald test is computationally simpler, provides better control of Type I error rates, achieves higher statistical power, and performs better than the LRT under small-sample conditions. Three major strategies have been proposed for estimating AHS. The first is based on structural parameter testing, including the Z test and the iterative Z test. The second relies on Bayesian network structure learning algorithms, such as the K2 algorithm (Cooper & Herskovits, 1992), Hill Climbing (HC) (Scutari & Denis, 2014), and Max-Min Hill Climbing (MMHC) (Tsamardinos et al., 2006). The third is based on parameter estimation and includes methods based on penalized marginal maximum likelihood estimation (MMLE), such as Regularized Latent Class Modeling (RLCM) (Wang & Lu, 2021), Latent Variable Selection (LVS) (Wang & Lu, 2021), and the Penalized Likelihood Approach (PLA) (Ma et al., 2022), as well as Bayesian joint estimation methods, including the General Bayesian Estimation Method (GBEM) (Chen & Wang, 2023) and the Novel Bayesian Estimation Method (NBEM) (Wang et al., 2026). Taken together, these methods each have their own strengths and limitations. These methods also differ in terms of analytical strategies, evaluation criteria, and classification accuracy across different types of AHS, reflecting their distinctive characteristics. It is worth noting that recent studies have further extended AHS analysis to the joint estimation or refinement of the Q-matrix and AHS (Lee & Gu, 2024; Ma et al., 2022; Wang et al., 2026; Wang & Sun, 2025). In particular, the Latent Class Bayesian Network (LCBN) proposed by Lee and Gu models structural parameters under AHS constraints, substantially reducing the number of structural parameters and providing a promising solution for high-dimensional attribute settings. In addition, the Attribute Correlation Intensity Matrix (ACIM) proposed by Yan (2022), as a representative non-parametric approach, performs well without imposing stringent sample-size requirements. Collectively, these studies represent important directions for the future development of AHS analysis. Overall, research on AHS has evolved along several dimensions. In terms of research objectives, it has progressed from AHS validation to AHS estimation, joint estimation of the number of attributes and AHS, and joint estimation of the Q-matrix and AHS. In terms of research content, the focus has shifted from describing the external forms of AHS to characterizing probabilistic relationships within its internal structure. Methodologically, the underlying modeling framework has evolved from single latent variable modeling frameworks based on Latent Class Models (LCMs), Diagnostic Classification Models (DCMs), or Bayesian Network Models (BNMs) toward integrated modeling frameworks. In practice, appropriate methods should be selected based on prior information, including sample size, attribute specification, the number of attributes, the Q-matrix, the measurement model and AHS. Future research should continue to optimize key methodological components, enrich experimental conditions, expand application scenarios, and develop innovative research perspectives. For example, promising directions include extending AHS analysis to polytomous items, polytomous attributes, multiple-strategy responses, longitudinal assessment, and computerized adaptive testing; integrating the strengths of existing methods to develop more effective analytical approaches; exploring new strategies for AHS analysis from the perspective of latent variable modeling; and expanding modeling frameworks through machine learning algorithms. These all represent important avenues for advancing methodological research on AHS analysis. References | Related Articles | Metrics