Advances in Psychological Science

Self-other integration in joint action: A hierarchical task representation perspective

WANG Jun, LI Qiankai, ZHAO Haiyang, ZHAO Mengfei, PAN Tingting, ZHENG Zheng

2026, 34 (10): 1709-1721. doi: 10.3724/SP.J.1042.2026.1709

Abstract ( 5 )

PDF (1151KB) ( 6 )

Real-world joint actions inherently possess a hierarchical structure, requiring co-actors to coordinate both concrete, low-level sensorimotor mappings and abstract, high-level task rules. However, prevailing theories—such as co-representation and referential coding—conceptualize self-other integration as a static, one-shot mapping, largely confining investigations to a single hierarchical level. This static, single-level approach fails to capture the dynamic information accumulation and cross-level interactions essential for complex interpersonal coordination. To address these critical gaps, the current study adopts a “hierarchical task representation” perspective, aiming to achieve a paradigm shift from “static single-level representation” to “dynamic hierarchical prediction.” The core innovations span three dimensions: phenomenological framing, computational mechanism modeling, and the identification of asymmetric social modulation effects.
First, we introduce a hierarchically nested paradigm to reveal the dissociation and bidirectional interaction of self-other integration. Previous studies have typically examined high-level (e.g., joint statistical learning) and low-level (e.g., joint Simon) tasks in isolation, leaving their dynamic interplay unexplored. By embedding statistical regularities within a joint Simon task, we simultaneously capture both levels within a single paradigm. Across five experiments combining behavioral measures and EEG hyperscanning, we provide systematic evidence that high-level and low-level integrations exhibit distinct temporal trajectories. Crucially, we reveal bidirectional interactions: high-level regularity acquisition exerts top-down modulation on low-level spatial compatibility, while low-level conflict influences high-level rule updating. Methodologically, we dissociate their neural signatures using time-resolved ERPs (frontal-central N2 for low-level conflict vs. parietal-occipital P3 for high-level prediction) and inter-brain synchrony, thereby overcoming the signal overlap inherent in behavioral reaction times.
Second, we construct a Bayesian computational model to quantify the cognitive mechanisms underlying level specificity. Transcending the descriptive limitations of prior theories, we formalize self-other integration as a dynamic Bayesian updating process. We propose that the observed level specificity arises from distinct modes of prior information accumulation, operationalized by a weight parameter (η). For high-level abstract tasks characterized by high uncertainty, the cognitive system adopts a "data-driven fast update" strategy (η>1), amplifying new observational evidence to rapidly revise prior beliefs. Conversely, for low-level concrete tasks with stable mappings, a “prior stability maintenance” strategy (0<η<1) is employed, diminishing the impact of new input to preserve behavioral coherence. Furthermore, to model cross-level interactions, we introduce regulation parameters (α and β) that mathematically define how the posterior distribution at one level dynamically modulates the prior of the other. By manipulating task difficulty and working memory load, and by linking these computational parameters to frontoparietal theta connectivity (a neural marker of evidence accumulation), we furnish computable and biologically verifiable evidence for the model's validity. This framework also reconciles the long-standing debate between co-representation and referential coding, re-conceptualizing them as top-down “social priors” and bottom-up “sensory inputs” within a unified predictive processing stream.
Third, we identify the asymmetric modulation effects of interpersonal common ground on hierarchical integration. While prior literature assumes a uniform facilitating role for common ground, we reveal its level-specific paradoxical effects. Across four experiments manipulating action-based and trait-based common ground, we demonstrate that action constraints (e.g., restricting response hands) specifically impair low-level spatial integration while sparing high-level abstract rule integration. More strikingly, we discover a counter-intuitive effect of trait common ground: similarity in prosocial traits (e.g., extraversion, agreeableness) facilitates low-level integration by enhancing behavioral predictability, yet paradoxically hinders high-level integration. This occurs because trait similarity induces a social projection bias, leading individuals to over-rely on self-knowledge and substitute their own abstract task representations for the partner’s, thereby diminishing the distinct encoding of the other’s high-level rules. Furthermore, by testing the robustness of our Bayesian model across these varying interpersonal contexts, we validate the generalizability of the proposed computational mechanisms.
In summary, this research constructs a comprehensive “phenomenon-mechanism-modulation” framework for hierarchical self-other integration. By integrating behavioral experiments, EEG hyperscanning, and cognitive computational modeling, we deliver a computable and verifiable theoretical tool. These innovations not only fundamentally advance the psychological understanding of dynamic interpersonal coordination but also provide critical theoretical inspiration and architectural parameters for the development of adaptive, collaborative multi-agent systems and human-machine teams.

References | Related Articles | Metrics

A two-stage model of the effects of prosocial effort on reward processing

ZHENG Ya, YANG Ziyang

2026, 34 (10): 1722-1732. doi: 10.3724/SP.J.1042.2026.1722

Abstract ( 3 )

PDF (1424KB) ( 5 )

Prosocial behavior often requires people to invest their own effort so that someone else can benefit. Yet theories and empirical work have largely emphasized the value of others’ outcomes, whereas the cost of the effort one personally incurs has received far less attention. The few studies that directly address prosocial effort have concentrated on effort-based reward decisions made before any effort is exerted, such as whether people are willing to work for rewards delivered to themselves or to others. This decision-stage focus leaves a critical dynamic question unanswered: once effort has actually been performed, does it continue to devalue the outcome obtained for others, or retrospectively enhance it, as it does for self-benefiting effort?
To address this question, we propose a two-stage model of how prosocial effort shapes reward processing. The model builds on the effort paradox observed in nonsocial contexts, in which effort exerts opposite effects at different processing stages. Before action, higher effort prospectively discounts reward value because it is encoded as a cost, whereas after action the same effort can retrospectively enhance reward value through effort justification or contrast effects. Extending this framework to the social domain, we argue that the effort-reward relationship depends on the beneficiary. When people work for themselves, reward processing should follow the classic effort paradox, showing prospective devaluation before effort exertion and retrospective enhancement after effort exertion. When people work for an anonymous other, however, effort may be encoded mainly as a personal cost weighed against someone else’s gain. Prosocial effort would then produce a double-devaluation pattern, reducing subjective reward value both before and after the effort is exerted. This prediction offers a new account of prosocial apathy: people may fail to help not because they do not care about others, but because the effort required to help persistently suppresses the perceived value of the outcome.
The proposed research will test this model along two integrated lines of investigation. Study 1 examines the cognitive and neural mechanisms by comparing reward processing when participants exert cognitive effort to earn rewards for themselves versus for anonymous strangers. Experiment 1 will use behavioral tasks to measure decision preferences before effort and affective evaluations after reward feedback, together with computational modeling of effort discounting. Experiments 2a and 2b will extend this design with EEG and fMRI. EEG will resolve the temporal dynamics of reward processing at the cue and feedback stages, and fMRI will localize the cortical and subcortical reward-related regions. Together, these experiments will reveal whether prosocial effort yields neural evidence of double devaluation, and whether this pattern diverges from the effort paradox observed in self-benefiting effort. Study 2 further tests whether the proposed mechanism is domain-general or domain-specific across effort types and benefit domains. Experiment 3 contrasts cognitive with physical effort, using matched task structures to examine whether prosocial effort produces comparable neural dynamics across cognitively and physically demanding tasks. Experiment 4 compares reward gain with loss avoidance, asking whether exerting effort to prevent another person’s loss shifts prosocial valuation toward the self-benefiting pattern. Experiment 5 compares reward gain with harm avoidance by adding a pain avoidance context, thereby testing whether a morally salient benefit to another person attenuates or reverses the double-devaluation pattern.
The innovation of this work lies in reconceptualizing prosocial behavior as a continuous effort-reward integration process rather than a static cost-benefit choice. It introduces effort cost and its temporal dynamics into the study of prosocial behavior, distinguishes two competing hypotheses, and links behavioral decisions, subjective experience, electrophysiological dynamics, and neural valuation within a single framework. By clarifying how effort reshapes the value of benefits to others before and after its exertion, the research will address a key gap in current theories of prosocial behavior and provide a new perspective on how self-incurred costs and others' benefits are balanced, and on how such behavior can be promoted.

References | Related Articles | Metrics

Driving mechanisms and impacts of stretch goals: An AI-enabled perspective

CHENG Jialin, LI Jinsong

2026, 34 (10): 1733-1748. doi: 10.3724/SP.J.1042.2026.1733

Abstract ( 3 )

PDF (636KB) ( 5 )

As artificial intelligence is progressively applied and deployed in enterprises, a growing number of work teams have moved away from traditional goal-setting methods and are attempting to stretch extreme goals to stimulate employee and team creativity within AI-enabled environments. However, the formation mechanism of stretch goals under AI empowerment, as well as their impact on creativity, has yet to be systematically explored and validated. To address this research gap, this paper develops an integrated research framework that systematically elucidates the driving mechanisms and effects of stretch goal setting from an AI empowerment perspective.
The core innovation of this study is to unpack the formation mechanism of stretch goals in the digital intelligence era. Prior studies led by Sitkin identified firm resource constraints and past performance as major predictors of stretch goal setting based on pre-2011 contexts, yet such formation mechanisms have changed dramatically amid digital and intelligent transformation. Managers now recognize that AI integration drives industrial competition from incremental improvement to disruptive innovation, allowing competitors to break traditional limits via AI and gain competitive advantages. In such a competitive environment, merely setting challenging goals is insufficient for firms to cope with growing market pressure. Instead, stretch goals help organizations break traditional thinking and optimize technology and resource allocation, which is essential for organizational survival and development in the AI age. Despite its practical significance, extant literature has largely overlooked how AI technology adoption affects firms’ stretch goal decisions. Against this theoretical and practical gap, this study anchors its analysis in the digital intelligence context. We argue that team-level AI adoption significantly motivates team leaders to prioritize stretch goals over regular challenging goals, and further verify that team leaders’ innovation expectation acts as a critical mediating variable underlying this causal linkage.
Secondly, this paper is the first to propose the empowering role of AI technology in the process through which stretch goals influence individual and team creativity. Existing research on stretch goals has focused only on the boundary effects of “social” factors—such as individual characteristics, individual strategies, work climate, and communication styles—while overlooking how AI can empower employees in pursuing stretch goals. This paper argues that although stretch goals trigger employees’ willingness to explore (i.e., “want to explore”), the technological system determines the extent to which employees are able to explore (i.e., “can explore”). Once a team adopts AI technology, both the perceived usefulness of AI and the human-AI role clarity influence the progression of stretch goal pursuit. Accordingly, drawing on socio‑technical systems theory, Study 2 proposes that perceived AI usefulness plays a significant moderating role in the process through which stretch goals affect individual creativity via employees’ exploratory learning. Study 3 identifies human-AI role clarity as a key contextual variable that shapes how stretch goals facilitate human‑AI collaboration and thereby enhance team creativity.
Finally, this study adopts a multi-level research approach to investigate the consequences of stretch goals. Although existing research has conceptualized stretch goals as a team-level construct, it has only examined their impact on unethical team behavior. How and under what conditions stretch goals foster team creativity in the context of AI empowerment remains largely unexplored. After setting team-level stretch goals, leaders typically communicate the content and requirements of these goals to team members through formal task allocation, team meetings, performance feedback, and other similar channels. In this process, on the one hand, members form individual-level perceptions of stretch goals based on their own understanding and the information received; on the other hand, interactions and sense-making among team members lead to an emergent shared cognition of stretch goals at the team level. Therefore, this study examines the effects of stretch goals on individual creativity and team creativity, as well as the underlying mediating mechanisms, from both individual and team perspectives, thereby contributing to research on stretch goals and creativity.
In summary, this study makes three major theoretical contributions. First, it reveals the formation mechanism of stretch goals in the digital intelligence era, clarifying how team AI adoption and perceived AI intelligence level influence stretch goal setting. Second, it goes beyond the boundary effects of “social” factors emphasized in prior goal research, and proposes the enabling role of AI as a “technological” factor in the process through which stretch goals enhance employee and team creativity, thus advancing stretch goal research. Third, it adopts a multi-level research approach to examine the consequences of stretch goals, investigating their effects on individual and team creativity from both levels, and reveals the mediating role of human‑AI collaboration in the relationship between stretch goals and team creativity. This not only fills the research gap concerning the link between stretch goals and team creativity but also promotes empirical research on human‑AI collaboration.

References | Related Articles | Metrics

When customers are empowered by algorithms: Dynamic mechanisms and interventions in gig workers’ coping with customer mistreatment

CAO Wenrui, LIU Beini, ZHANG Shanshan, TIAN Zonglin

2026, 34 (10): 1749-1767. doi: 10.3724/SP.J.1042.2026.1749

Abstract ( 3 )

PDF (774KB) ( 6 )

With the increasing use of algorithmic management in the gig economy, customers are no longer merely service recipients but are increasingly empowered by algorithms. Through rating systems, complaint mechanisms, and platform-mediated feedback, customers can influence gig workers’ task opportunities, income, reputation, and occupational sustainability. In this context, customer mistreatment becomes embedded in platform rules and amplified by algorithm-enabled customer empowerment. Existing research has examined how platforms control gig workers through algorithms, but has paid insufficient attention to how algorithms empower customers and reshape gig workers’ coping with customer mistreatment. Drawing on cognitive appraisal theory, this study develops an integrative framework to examine the dynamic mechanisms and interventions in gig workers’ coping with customer mistreatment. It addresses three questions: the forms that customer mistreatment takes when customers are empowered by algorithms and how it can be measured; how customer mistreatment affects long-term occupational outcomes through cognitive appraisal, coping responses, and emotional labor; and whether AI-mediated responsive forgiveness intervention can support post-event adjustment and recovery.
First, this study reconceptualizes customer mistreatment in the algorithmic management context and develops a measurement instrument tailored to gig work. Prior research has focused mainly on traditional service settings, where customer mistreatment typically appears as direct verbal abuse, threats, disrespect, or harassment. However, when customers are empowered by algorithms, mistreatment may become more indirect, concealed, and instrumental. Customers may exploit gig workers’ dependence on ratings, threaten negative reviews, manipulate evaluation systems, abuse complaint channels, or impose additional service demands through platform rules. Such mistreatment may occur before, during, and after task completion, and its effects may be intensified when customer feedback is converted by algorithms into future opportunities, income, and platform status. Existing measures may therefore fail to capture the distinctive forms of mistreatment produced by algorithm-enabled customer empowerment. To address this limitation, this study identifies the dimensions of customer mistreatment through in-depth interviews, textual analysis, and topic modeling, and develops a gig-specific measurement instrument. This effort clarifies the conceptual boundaries of customer mistreatment in the gig economy and provides a foundation for research on gig workers’ coping processes.
Second, this study constructs a dynamic, cross-temporal mechanism model to explain how gig workers cope with customer mistreatment and how such coping shapes occupational outcomes. Existing studies often portray gig workers as passive objects of algorithmic control, while overlooking their active cognitive appraisal and coping responses. Based on cognitive appraisal theory, this study argues that coping depends on how gig workers evaluate the mistreatment event. When customer mistreatment is appraised as a challenge, gig workers may adopt problem-focused coping responses, such as adjusting communication strategies or managing customer expectations. When it is appraised as a hindrance, they may rely more on emotion-focused coping, avoidance, defensive responses, or withdrawal-oriented strategies. These coping responses may accumulate over time and develop into stable emotional labor patterns. In particular, this study proposes deep acting and surface acting as key mediating mechanisms linking customer mistreatment to occupational burnout, perceived employability, and exit intention. The model moves beyond short-term stress reactions and highlights how repeated exposure to customer mistreatment may influence occupational sustainability. It further examines personal, platform, and social resources as moderators, explaining why some gig workers can sustain adaptive coping whereas others become vulnerable to exhaustion and withdrawal.
Third, this study proposes AI-mediated responsive forgiveness intervention as an innovative approach for supporting gig workers’ coping with customer mistreatment. Platform technologies are often discussed as instruments of surveillance, control, and performance discipline. In contrast, this study emphasizes that AI can also empower workers and support psychological recovery. Because gig workers often lack formal organizational support, they may have limited access to psychological resources after customer mistreatment. By integrating forgiveness intervention theory with AI-mediated conversational support, this study proposes that AI can guide gig workers to reinterpret mistreatment events, reduce rumination, regulate negative emotions, and restore adaptive coping. The intervention is “responsive” because it responds to gig workers’ specific mistreatment experiences and post-event adjustment needs rather than offering generic psychological advice. This framework extends forgiveness intervention theory to the gig economy and provides a new direction for AI-enabled occupational mental health support.
In summary, this study extends customer mistreatment theory by revealing how algorithm-enabled customer empowerment generates new forms of customer mistreatment in gig work. It advances cognitive appraisal and coping research by linking customer mistreatment, cognitive appraisal, coping responses, emotional labor, and occupational outcomes in a dynamic framework. It also introduces AI-mediated responsive forgiveness intervention as a worker-centered approach, shifting the role of technology from labor control to worker empowerment. These contributions provide foundations for understanding gig workers’ coping with customer mistreatment and for building more human-centered platform governance.

References | Related Articles | Metrics

The echo effect of gratitude expression in tourism service interactions

CAO Limei, LI Fangshi, LI Yaoqi

2026, 34 (10): 1768-1778. doi: 10.3724/SP.J.1042.2026.1768

Abstract ( 3 )

PDF (690KB) ( 4 )

Gratitude expression is ubiquitous in interpersonal interactions. Prior research has primarily examined its positive effects from the perspective of a single interactions or a single actor, with limited attention to its dynamic evolution in ongoing interactions. Notably, in tourism service interactions, gratitude expression is particularly likely to be continually acknowledged, reciprocated, and amplified across multiple rounds of interaction, because tourism services are characterized by open expression norms, reciprocal and continuous interaction, and network connectivity. However, existing theories are largely based on static, linear causal logic and struggle to explain how gratitude expression continuously amplifies its influence through bidirectional feedback and role switching.
In response, this manuscript proposes the echo effect of gratitude expression, which re-conceptualizes gratitude expression from a one-time emotional event into an ongoing interaction process. The echo effect is defined as a dynamic phenomenon in which gratitude expression, grounded in bidirectional feedback within a core interaction dyad, progressively extends its influence across multiple rounds of interaction and role transitions. This concept further distinguishes between two levels: direct echo cycles (bidirectional feedback within the same interaction dyad) and cross-actor echo expansion (the transfer of influence to other interaction partners).
Building on this conceptualization, this manuscript develops two modules comprising three studies. Study 1 focuses on theoretical construction. It proposes four core propositions that elucidate a closed-loop mechanism: gratitude expression→positive responsiveness→reciprocal gratitude expression→sustained positive responsiveness. This framework integrates role switching and network connectivity as key conditions for expansion, thereby establishing an integrated theoretical model of the echo effect of gratitude expression.
Study 2 focuses on the tourist-to-employee pathway. Drawing on conservation of resources theory, this study will reveal how tourist gratitude expression alleviates employee emotional exhaustion and enhances service performance through two parallel mechanisms: cognitive reappraisal (reframing service work as meaningful contribution) and emotional energy (accumulating positive affect from gratitude encounters). These mechanisms may help employees recover from emotional depletion and sustain high-quality service delivery.
Study 3 focuses on the employee-to-tourist pathway. Grounded in construal level theory, this study will examine how identity cues in employee gratitude expression (individual identity vs. collective identity) influence tourist responses. Individual-identity expressions may reduce tourists’ psychological distance from employees, enhance their perceived closeness to the employee, and thus promote immediate reciprocal behaviors (e.g., tipping, positive reviews). Collective-identity expressions may increase psychological distance, yet they enhance brand reliability perceptions and foster long-term loyalty. Furthermore, perceived sincerity of gratitude expression may be introduced as a critical boundary condition, moderating the effects of gratitude expression on tourist responses.
The theoretical contributions of this manuscript are fourfold. First, it proposes the original concept of the echo effect of gratitude expression, shifting gratitude research from a static, single-interaction perspective to a dynamic, multi-round, bidirectional interaction perspective. This represents a fundamental reconceptualization of how gratitude operates in ongoing social exchanges. Second, it reveals the accumulation-expansion mechanism of gratitude expression in continuous interactions, moving beyond existing linear causal models to explain how seemingly minor gratitude expressions can generate progressively larger effects over time and across interactants. Third, it will identify and test two distinct pathways—tourist-to-employee and employee-to-tourist—each with different underlying mechanisms. This dual-pathway design provides a more comprehensive and integrative theoretical account of emotional expressions in service contexts. Fourth, by leveraging the unique features of the tourism service setting—including open expressive norms, interactional reciprocity and continuity, and network connectivity—this manuscript offers new theoretical insights and empirical avenues for understanding the sustained role of emotional expressions in service ecosystems. The echo effect is not merely a dyadic phenomenon but can propagate across broader service networks.
Practically, this manuscript provides actionable guidance for tourism organizations to optimize emotional interaction management, enhance service experiences, and facilitate value co-creation. By understanding how gratitude expressions can be strategically initiated and reciprocated, organizations can foster positive emotional climates, reduce employee burnout, and build lasting customer loyalty.

References | Related Articles | Metrics

A Bayesian meta-analysis of the association between AI literacy and self-efficacy

WAN Qianyi, WEN Siqing, MA Zaoming, HU Bo

2026, 34 (10): 1779-1794. doi: 10.3724/SP.J.1042.2026.1779

Abstract ( 7 )

PDF (1595KB) ( 6 )

Objective: Previous studies on the relationship between AI literacy and self-efficacy have reported inconsistent findings because of varied conceptualization, measurement approaches and sample contexts. Importantly, the relationship is theoretically complex. Students with stronger AI literacy may experience an enhanced sense of control and competence, while those with greater awareness of AI’s capabilities may potentially erode confidence in their competence. To address this issue, the present study used a Bayesian meta-analysis to systematically synthesize the available evidence and identify the moderating conditions under which the association varies.
Method: This study followed PRISMA 2020 guidelines and systematically searched Web of Science (SCI/SSCI), Scopus, and CNKI through November 2025. A total of 36 independent studies involving 25,251 university students were included. A Bayesian random-effects model with weakly informative priors [Normal (0,1) for the pooled effect; Half-Normal (0.5) for heterogeneity] was used as the primary analytical framework. Model convergence was evaluated through MCMC sampling with R-hat and ESS convergence checks. Prior sensitivity analyses (Half-Normal (1.0)) confirmed robustness. Frequentist random-effects models were also conducted for robustness checks. Moderator analyses covered self-efficacy type, AI literacy measurement approach, student discipline, country/region, and study design. Publication bias was evaluated via funnel plot inspection and Egger's regression. Methodological quality was assessed using the Newcastle-Ottawa Scale (inter-rater ICC = 0.938).
Results: The Bayesian model yielded a robust positive overall association (posterior median r = 0.495, 95% CrI [0.420, 0.570]; frequentist r = 0.518), with substantial heterogeneity (I² = 91.73%, τ = 0.32). Publication bias was negligible (Egger's p = .358), and leave-one-out sensitivity analyses confirmed stability (r range: 0.50-0.53). Four significant moderators emerged (all Qb p < .05); country/region was not significant (Qb = 0.76). Self-efficacy type was the strongest moderator: associations were highest for creative self-efficacy (r = 0.71) and entrepreneurial self-efficacy (r = 0.61), exceeding those for AI-specific self-efficacy (r = 0.53) and general academic self-efficacy (r = 0.37). Studies using self-developed or integrated self-report scales yielded larger effects (r = 0.71) than those using objective AI literacy tests (r = 0.14). Students in education, media, and language disciplines showed stronger associations (r = 0.60) than those in engineering/computer science (r = 0.45) or business (r = 0.41). Within-region heterogeneity was equally high in East Asian (I² = 78.22%) and Western samples (I² = 77.52%), indicating that meaningful variation resides within rather than between geographic regions.
Contributions: This study makes three major contributions. First, it adopts a Bayesian meta-analytic framework. Compared with traditional frequentist approaches, it more effectively captures both effect and heterogeneity parameters. Second, the moderation pattern challenges domain-matching intuitions: AI literacy's strongest links are not with AI-specific self-efficacy nor with technically oriented students (engineering and computer science), but with creative and entrepreneurial efficacy beliefs and with students from humanities-adjacent disciplines. This counterintuitive result suggests AI literacy operates less as a narrow technical competency reinforcing task-specific confidence than as a transferable psychological resource activating higher-order efficacy beliefs around creativity, career adaptation, and future agency. In the generative AI era, this reframes AI literacy as a form of human capital with broad motivational transfer rather than a domain-bound skill. Third, although no significant cross-national differences were found, substantial within-region heterogeneity remained. This suggests that micro-level instructional factors, such as curriculum design, Al integration strategies, and teacher support, may play a more important role than broader cultural or policy contexts.
Conclusions:Overall, AI literacy is positively associated with university students’ self-efficacy, although the effect strengths depend substantially on self-efficacy type, measurement approach and disciplinary context. These findings suggest that AI literacy education should move beyond technical training alone and place greater emphasis on creativity, adaptability and higher-order efficacy beliefs. Higher education institutions should develop differentiated AI literacy curricula based on students’ disciplinary backgrounds and developmental needs. Future research should emphasize longitudinal and experimental design, as well as more standardized measurement approaches, in order to better identify the causal mechanism underlying the relationship between AI literacy and self-efficacy.

References | Related Articles | Metrics

Applications of transcranial focused ultrasound stimulation in psychological research

XIA Xue, CHENG Si, ZHANG Dandan

2026, 34 (10): 1795-1812. doi: 10.3724/SP.J.1042.2026.1795

Abstract ( 3 )

PDF (1371KB) ( 3 )

Transcranial focused ultrasound stimulation (TUS) is becoming important in cognitive and affective neuroscience because it can deliver acoustic energy to cortical and subcortical targets with high spatial precision. For psychological research, TUS makes it possible to ask causal questions at two temporal scales: how a target region contributes to an ongoing psychological process at a particular moment, and how stimulation-induced changes in neural state influence subsequent behavior. We therefore organize this review around the distinction between online and offline TUS, and treat these two modes as complementary routes for causal investigation in psychology.
In human research, stimulation parameters are usually described through several dimensions, including acoustic output, temporal structure, and exposure descriptors. Acoustic output includes carrier frequency and pressure-related measures; temporal structure includes pulse duration, pulse repetition frequency, duty cycle, and total sonication duration; and exposure descriptors include spatial-peak pulse-average intensity, spatial-peak temporal-average intensity, and mechanical index. This organization is useful because no single parameter can be equated with excitation or inhibition. The similar protocol may yield different outcomes when target depth, skull transmission, focal coverage, baseline neural state, task context, and outcome measure differ.
Online TUS is delivered during or immediately around a task event or neural response. Its defining feature is temporal locking: ultrasound pulses can be aligned with sensory input, response preparation, decision competition, or the onset of inhibitory control. This makes online TUS suitable for testing when a region is causally involved in perception, action, or control. Somatosensory studies show that stimulation of primary somatosensory cortex can reduce early evoked potentials while improving tactile discrimination, suggesting that TUS may reshape sensory coding rather than simply enhance or suppress cortical activity. Thalamic studies further extend causal manipulation to deep sensory relay structures. In visual research, stimulation of primary visual cortex and frontal eye fields has been used to modulate visually evoked activity and bias spatial choice under uncertainty. In the motor system, online TUS has been used to examine motor cortex excitability, intracortical facilitation and inhibition, and movement-related cortical potentials. Work on response inhibition shows that right inferior frontal gyrus stimulation improves stopping only when the pulse is time-locked to the stop signal, indicating an effect on inhibitory timing rather than a general benefit. Pain studies targeting the insula and dorsal anterior cingulate cortex further show how online TUS can dissociate immediate regional roles in sensory intensity, salience evaluation, and autonomic regulation.
Offline TUS separates stimulation from the main behavioral or physiological measurement. It focuses on aftereffects that persist for minutes or longer after sonication has ended, and is therefore suited to questions about sustained changes in excitability, excitation-inhibition balance, neurochemical state, functional connectivity, and plasticity-like processes. In emotional research, offline stimulation of prefrontal and limbic regions has been associated with changes in helplessness, approach-avoidance behavior, mood, amygdala reactivity, and symptoms of anxiety or depression. These findings suggest that TUS can test not only cortical regulatory mechanisms, but also deep affective circuits that are difficult to access with transcranial magnetic or electrical stimulation. In response inhibition, offline TUS of basal ganglia and insular-prefrontal pathways provides causal evidence that stopping depends on a distributed hierarchy rather than a single prefrontal node. In memory and reward research, stimulation of the ventromedial anterior temporal lobe or nucleus accumbens has been linked to changes in semantic memory, reward sensitivity, feedback learning, and reward-network connectivity. Offline pain and motor studies further show that TUS can alter pain thresholds, temporal summation, chronic pain ratings, corticospinal excitability, interhemispheric balance, and motor symptoms, although these effects depend on target, baseline state, medication status, and acoustic dose.
Rigorous interpretation of TUS findings also requires stronger control of confounds and more transparent safety reporting. Online studies are especially vulnerable to auditory and somatosensory confounds because the acoustic pulse is synchronized with the task event. Sham stimulation, masking noise, active control sites, acoustic blocking, and vibration-matched controls should be selected according to the inference being tested. For safety, future studies should report nominal device settings, the basis of intensity estimation, in situ pressure and intensity when available, mechanical index or transcranial mechanical index, thermal estimates, stimulation duration, and skull-transmission assumptions. Recent adverse-event reports further indicate that conservative safety margins and individualized acoustic simulation are essential, especially for low-frequency, long-pulse, or higher-pressure protocols.
In sum, online and offline TUS answer different causal questions. Online TUS is most useful for identifying the timing of causal involvement in ongoing processing; offline TUS is most useful for testing how stimulation-induced brain-state changes alter later cognition, emotion, behavior, or symptoms. By distinguishing these modes, we aim to move TUS research in psychology from proof-of-principle demonstrations toward reproducible, mechanism-focused causal studies.

References | Related Articles | Metrics

Detecting without discerning: Characteristics and mechanisms of interoception in individuals with autism spectrum disorder

ZHU Xingchen, HU Jinsheng

2026, 34 (10): 1813-1831. doi: 10.3724/SP.J.1042.2026.1813

Abstract ( 2 )

PDF (624KB) ( 4 )

Interoception, the perception and integration of signals from within the body, has become a candidate mechanism for understanding the emotional and social difficulties observed in autism spectrum disorder (ASD). Despite growing empirical attention, the literature remains fragmented across measurement paradigms, developmental stages, and comorbidity profiles, leaving the conditions under which interoceptive atypicalities arise insufficiently characterized. This review synthesizes evidence across three dimensions of interoception, namely accuracy, sensibility, and awareness, and advances an integrative framework that organizes three established theoretical accounts into a hierarchical explanatory structure.
Behavioral, neural, and qualitative findings together suggest that individuals with ASD exhibit a dissociated profile rather than a uniform deficit. Objective accuracy on heartbeat detection tasks tends to be reduced, although estimates vary substantially across developmental stages and instruments. Self-reported sensibility appears elevated in adult samples on frequency-based measures such as the Body Perception Questionnaire, yet diminished on multidimensional instruments such as the MAIA, suggesting that conclusions are sensitive to what each instrument captures. Awareness, indexed by the discrepancy between objective performance and subjective evaluation, shows a systematic mismatch whereby individuals tend to overestimate their interoceptive capacity. This three-dimensional dissociation captures what generic deficit accounts cannot easily reconcile, namely the simultaneous presence of heightened bodily sensing and impaired bodily discerning.
The review further argues that three influential mechanistic accounts, predictive coding, weak central coherence, and the alexithymia hypothesis, are more parsimoniously understood as nested levels within a developmental cascade than as competing explanations. At the computational level, atypical precision weighting in Bayesian inference renders individuals with ASD overly reliant on bodily input while limiting context-sensitive adjustment. At the cognitive level, weak central coherence further impedes the integration of locally detected signals into coherent multimodal representations of bodily state. At the affective level, the resulting failure to anchor integrated interoceptive representations onto emotional categories manifests as alexithymia, which co-occurs with ASD in approximately half of cases. The insula and anterior cingulate circuitry, atypical across age groups in ASD, provides a plausible neural substrate for this organization. Marr's tri-level analytic framework, developmental timing considerations, and the non-universality of alexithymia within ASD jointly motivate the proposed nesting.
Several factors appear to moderate the expression of interoceptive atypicality. Age effects are dimension-specific. Accuracy differences are shaped by intellectual functioning, comorbidity, and task demands, and current cross-sectional evidence is insufficient to establish a unitary developmental trajectory. Signal type also matters, with cardiac and respiratory channels that depend on cross-modal integration appearing more vulnerable than channels with stable behavior-outcome mappings such as hunger. Comorbid anxiety, alexithymia, attention-deficit/hyperactivity disorder, and eating disorders each contribute to the heterogeneous expression observed in ASD. Methodological confounds in heartbeat counting tasks, including resting heart rate and body mass index, may inflate apparent group differences if left uncontrolled.
Three priorities are identified for future research. First, simultaneously measuring all three interoceptive dimensions within the same sample is needed to clarify their interrelations and to determine which dimension exerts the greatest functional impact on emotion regulation and social functioning. Second, partitioning the contributions of alexithymia and other comorbidities through stratified or mediation designs is necessary to test whether the proposed cascade applies uniformly across symptom domains. Third, longitudinal designs combined with multimodal indices that go beyond heartbeat detection are needed to capture the developmental dynamics of interoceptive atypicality. Implications for educational practice include staged interoceptive training tailored to a learner’s dimensional profile, metacognitive calibration exercises that address overconfidence in bodily judgment, and emotion regulation strategies adjusted to the individual's interoceptive capacity.
The three-dimensional dissociation identified here also has implications for educational and clinical practice. Rather than applying uniform interoceptive training, instruction may benefit from being staged according to each learner’s dimensional profile, beginning with body-scan exercises that link bodily cues to emotion vocabulary for those who detect signals but struggle to interpret them, and progressing to context-sensitive discrimination for those whose subjective and objective indices diverge. Metacognitive calibration through feedback comparison helps address the overconfidence that often accompanies impaired awareness, while emotion regulation instruction may need to shift from body-based strategies toward cognitive reappraisal when interoceptive accuracy is markedly reduced.
By foregrounding the dissociation between bodily sensing and discerning, and by situating the three dominant mechanistic accounts within a single developmental hierarchy, this review offers a more coherent framework for interpreting the heterogeneous interoceptive profile of ASD and for guiding targeted intervention.

References | Related Articles | Metrics

Neurotransmitter regulatory mechanisms of abnormal sensory attenuation in schizophrenia

CHEN Yiyue, ZENG Yaxin, HUANG Chaozheng, XIE Pei

2026, 34 (10): 1832-1842. doi: 10.3724/SP.J.1042.2026.1832

Abstract ( 2 )

PDF (772KB) ( 7 )

Schizophrenia is a severe and complex psychiatric disorder with high heterogeneity, characterized by positive symptoms (such as hallucinations and delusions), negative symptoms (including emotional blunting and social withdrawal), and extensive cognitive impairments, which severely affect patients’ cognitive, emotional, and behavioral functions. Among its core neurophysiological features, abnormal sensory processing—especially deficits in sensory attenuation—has attracted increasing attention in academic circles. Sensory attenuation plays a crucial role in regulating individuals’ adaptive responses to environmental stimuli: it enables the brain to dynamically adjust the responses of neuronal populations to repetitive sensory inputs, filter out irrelevant information, and prioritize the processing of important signals, thereby ensuring the efficient allocation of cognitive resources. However, schizophrenia patients often exhibit significant abnormalities in sensory attenuation, manifested as hypersensitivity to environmental stimuli and reduced ability to screen information. This neurophysiological disturbance not only directly leads to cognitive control deficits and abnormal subjective experiences but also is closely associated with the emergence of core symptoms like hallucinations, potentially serving as the neurobiological basis for the disease’s cognitive and behavioral abnormalities. Neurotransmitters, as key substances for information transmission between neurons, are critical regulators of sensory attenuation mechanisms. A large body of research has confirmed that functional abnormalities in multiple neurotransmitter systems—including glutamate, dopamine, and γ-aminobutyric acid (GABA)—are closely related to the pathogenesis of schizophrenia and are important factors contributing to the disruption of sensory attenuation mechanisms, though their specific regulatory mechanisms have not yet been fully and systematically clarified. Glutamate, the main excitatory neurotransmitter in the brain, affects sensory attenuation through abnormal activity of glutamatergic neurons (e.g., reduced activity in the prefrontal cortex and hippocampus) and dysfunction of N-methyl-D-aspartate (NMDA) receptors, which impairs synaptic plasticity and the brain’s adaptive responses to repetitive stimuli. Dopamine, a major modulatory neurotransmitter, exerts complex effects on sensory attenuation: abnormal activity of dopaminergic neurons in brain regions such as the striatum and prefrontal cortex, along with dysfunction of dopamine receptors (particularly D1 receptors), leads to biased processing and integration of sensory information, resulting in hypersensitivity to sensory stimuli. GABA, the primary inhibitory neurotransmitter in the brain, contributes to sensory attenuation deficits through reduced numbers or impaired function of GABAergic neurons in key brain regions (e.g., prefrontal cortex, hippocampus) and dysfunction of GABA-A receptors, disrupting the inhibitory regulation of neuronal excitability and thus the normal sensory attenuation process. Genetic factors further modulate the function of these neurotransmitter systems. For instance, the Val158Met polymorphism of the catechol-O-methyltransferase (COMT) gene affects dopamine metabolism, while polymorphisms in GRIN genes (which encode NMDA receptor subunits) regulate glutamate signaling; both ultimately influence sensory attenuation by altering neurotransmitter function. From a cognitive neuroscience perspective, neurotransmitter abnormalities disrupt the normal activity of visual, auditory, and tactile sensory pathways and impair the functional integrity of the default mode network (DMN), further exacerbating sensory attenuation deficits in schizophrenia patients. This review integrates multimodal evidence from neurobiology, molecular genetics, and cognitive neuroscience to systematically discuss the regulatory roles of neurotransmitters in abnormal sensory attenuation in schizophrenia. It emphasizes the need for future longitudinal studies combining dynamic neuroimaging techniques (e.g., fMRI, PET) to clarify the temporal relationship between neurotransmitter abnormalities and sensory attenuation deficits. Additionally, it proposes that future research should focus on developing targeted interventions (such as D1 receptor modulators and NMDA receptor partial agonists) and constructing individualized prediction models using multi-omics data and machine learning algorithms, with the goal of providing a scientific basis for precise intervention in schizophrenia and promoting improvements in patients’ prognosis and social function recovery.

References | Related Articles | Metrics

Fidelity measurement in psychotherapy: Current status and improvements

YANG Haoran, SUN Huanxiang, LIU Xiaoming

2026, 34 (10): 1843-1861. doi: 10.3724/SP.J.1042.2026.1843

Abstract ( 2 )

PDF (618KB) ( 2 )

Psychotherapy fidelity, defined as the extent to which a psychological intervention is delivered as intended, is considered to encompass both adherence (the extent to which pre-specified interventions are used) and competence (the skill with which they are implemented). While maintaining high fidelity is crucial for validating therapeutic efficacy and ensuring clinical quality, its measurement remains a significant challenge. Traditional methods, particularly the gold-standard observational coding, are hindered by high labor costs and inconsistent scoring among human raters. Alternative approaches, such as written surveys and behavioral rehearsals, fail to accurately capture real-world clinical execution. To address these issues, researchers have introduced frameworks (such as TREND, CONSORT, the BCC framework, and Carroll et al.’s framework), behavioral taxonomies (such as the Behavior Change Technique Taxonomy v1 [BCTTv1]), and machine learning methods for coding. However, these approaches remain insufficient; existing frameworks often offer broad recommendations without providing operational guidelines. Taxonomies can describe the presence of specific interventions (adherence) but tend to overlook the quality of execution (competence). Meanwhile, machine learning methods struggle with the high costs of manual data annotation and exhibit poor generalizability across different therapeutic modalities.
To address the dual challenges of the high cost and evaluation inconsistency, this paper proposes three future directions for fidelity measurement: refining measurement frameworks, establishing a unified knowledge base, and utilizing Large Language Models (LLMs) for automated assessment.
First, existing fidelity measurement frameworks can be continuously updated and refined under the guidance of evidence-based practice (EBP). A comprehensive framework should provide standardized description methods for all core dimensions of fidelity and clear rules for determining fidelity levels. To address current deficiencies in measuring therapist competence, future frameworks can integrate well-defined behavioral anchors with the Dreyfus model of skill acquisition to delineate different proficiency stages. Furthermore, grounding the framework in EBP ensures that the therapeutic components being evaluated are genuinely effective, preventing fidelity assessment from degrading into a superficial recording of clinical processes. As fidelity measurement is the prerequisite for successfully implementing evidence-based treatments, aligning these frameworks ensures that therapists effectively combine the best research evidence with clinical expertise and client characteristics.
Second, it is necessary to establish a unified fidelity measurement knowledge base to translate abstract frameworks into actionable, consistent criteria. This involves utilizing diverse strategies (such as shared component analysis and dismantling trials) to systematically extract valuable therapeutic components from various psychological treatment modalities. The paper advocates for adapting the Evidence to Decision (EtD) framework, originally used in medical guidelines, for use in psychotherapy. The EtD approach provides a systematic method for experts to review objective evidence and form consensus-based recommendations, effectively translating subjective expert experience and objective clinical evidence into unified, standardized scoring rules. This structured consensus is vital for mitigating the subjective discrepancies among expert raters.
Third, LLMs offer a technical solution to the high-cost and scalability issues of traditional human coding. Unlike traditional machine learning methods that require massive, expensive annotated datasets, LLMs possess strong few-shot learning capabilities. Given that psychotherapy is fundamentally language-based, the core task of fidelity measurement—identifying, categorizing, and quantifying verbal content and therapeutic functions—aligns with the natural language processing strengths of LLMs. Concerns regarding the inherent non-deterministic nature of LLMs can be effectively mitigated. LLMs inherently capture stable statistical regularities from massive corpora, ensuring overall consistency. Furthermore, fidelity measurement is a strictly bounded scoring task. Researchers can utilize the aforementioned structured frameworks and unified knowledge base to craft highly constrained prompts. This ensures that the evaluation boundaries, scoring dimensions, output formats, and decision rules are strictly defined. Combined with techniques like structured outputs and aggregation strategies, LLMs can provide scoring with substantially improved consistency. The efficiency and low cost of LLMs make comprehensive, large-scale quality assessments of psychotherapy processes practically feasible.
In conclusion, the future of psychotherapy fidelity measurement relies on the synergy of structured normative knowledge and advanced automated technologies. By employing EBP-guided frameworks, an EtD-based unified knowledge base, and LLM-assisted evaluation, the field can drive psychotherapy process evaluation from theoretical advocacy to a standardized, cost-effective, and highly reliable clinical practice.

References | Related Articles | Metrics

Computational cognitive mechanism of stereotype: Social learning and generalization

WANG Ying-Jie, ZHANG Ru-Yuan

2026, 34 (10): 1862-1875. doi: 10.3724/SP.J.1042.2026.1862

Abstract ( 1 )

PDF (1083KB) ( 6 )

Stereotypes—generalized beliefs that members of a social group tend to possess certain traits—are central to human social cognition. Traditional accounts have attributed stereotyping to motivational needs or cognitive resource constraints. However, these perspectives remain largely descriptive, and offer limited insight into the computational mechanisms governing how stereotypes are acquired, updated, and applied. Adopting a computational cognitive neuroscience perspective, this review integrates two dominant formal frameworks, reinforcement learning (RL) and Bayesian theory, to provide a unified, process-oriented account of stereotype formation and generalization.
With respect to social learning, we first examine how Bayesian structure learning enables the brain to infer latent group categories from observable social data, estimating the posterior probability of group assignments. Building on this, we delineate two functionally distinct pathways through which group-trait associations are established. The experiential pathway encompasses Pavlovian conditioning, instrumental learning, and observational learning. These operate primarily through RL mechanisms—particularly value updating, formalized as V(group) ← V(group) + αδ—to form associative representations. These automatic links between group labels and affective valence that influence behavior without requiring propositional evaluation. The linguistic pathway, by contrast, supports the acquisition and transmission of propositional representations (truth-apt beliefs, e.g., “group G possesses trait T”) through symbolic communication. The Bayesian framework provides a natural computational account for this type of representation, formalizing stereotypic beliefs as conditional probabilities, P(trait | group), that are updated in accordance with Bayes’ rule. Recent empirical evidence confirms that human stereotype judgments closely approximate Bayesian posteriors, and methodological advances further enable the direct quantification of propositional stereotype content in natural language corpora.
We further analyze how these associations are maintained or revised. Prediction error (PE) constitutes the core learning signal driving stereotype updating: counter-stereotypical information generates heightened PE, accelerating the revision of existing associations. However, pre-existing stereotypes resist updating through biased priors and asymmetric learning rates; individuals tend to learn more rapidly from evidence that confirms their stereotypic expectations than from disconfirming evidence. We identify the explore-exploit dilemma as a critical convergence point of the two frameworks: Bayesian priors shape RL-based exploration strategies, producing biased information sampling that in turn reinforces existing beliefs, creating a self-perpetuating cycle. This computational mechanism offers a principled explanation for why stereotypes resistant to change, even in the absence of motivational bias or cognitive limitations.
At the level of social generalization, we focus on the computational challenge of group identification: how the brain categorizes a novel individual into a known group to deploy learned associations. We distinguish two routes of similarity-based matching. Perceptual generalization operates on directly observable physical features in early visual processing, whereas functional generalization relies on abstract role and category information mediated by latent variable inference. The dynamic interaction between bottom-up perceptual signals and top-down conceptual knowledge, coordinated by the dorsomedial prefrontal cortex (dmPFC), means that group identification is not a serial process but a continuous integration of multiple cue types. Once group membership is established, both associative and propositional pathways are engaged during retrieval, with the former enabling rapid, resource-efficient behavioral responses and the latter supporting more deliberate trait inference.
Finally, we identify three promising directions for future research: (1) investigating how relational cues, such an individual’s position within a social network, contribute to stereotype generalization beyond feature-based similarity; (2) leveraging the social cognitive map framework, supported by recent evidence of distance and grid-like coding in the hippocampal-entorhinal system, to model the integrated representation of multi-dimensional social information; and (3) employing large language models (LLMs) to simulate linguistic transmission, in order to examine how minor initial biases become amplified and consolidated into shared cultural stereotypes through iterative communication.

References | Related Articles | Metrics

From self-serving to harm-seeking: Evolutionary pathways from negative to malevolent creativity and intervention

ZHOU Shujin, ZHU Jingyi, TU Hongyue, Li Yan

2026, 34 (10): 1876-1886. doi: 10.3724/SP.J.1042.2026.1876

Abstract ( 1 )

PDF (649KB) ( 6 )

Creativity is commonly understood as a neutral cognitive capacity that enables individuals to generate novel and appropriate ideas, solutions, or products. Depending on its motivational orientation and social consequences, creativity may lead to constructive outcomes, but it may also produce negative social consequences. The latter constitutes the negative side of creativity and includes two closely related forms: negative creativity and malevolent creativity. Both forms share the cognitive foundation of high novelty and high appropriateness, and both may result in negative social outcomes. However, they differ fundamentally in motivational orientation. Negative creativity is primarily driven by self-serving goals, with harm emerging mainly as an incidental byproduct. By contrast, malevolent creativity is motivated by an explicit intention to harm others, organizations, or society, with self-interest functioning as a possible but secondary motive.
In early childhood, behaviors that may contain elements of negative creativity, such as rule-bending, self-serving deception, and risky exploration, begin to emerge. However, the boundary between negative creativity and malevolent creativity is particularly difficult to determine during this developmental period. Children’s negative ideas are often situationally induced, unstable, and closely tied to immature cognitive control, moral reasoning, and social understanding. As a result, it is difficult to distinguish temporary negative ideation from stable malicious intent. Existing measurement tools, which are largely developed for adults and often rely on self-report scales or scenario-based tasks, are limited in their ability to identify children’s negative creativity and to differentiate it from malevolent creativity. This measurement dilemma constrains early identification and intervention.
Adopting a developmental perspective, this review systematically examines the multi-pathway mechanisms through which negative creativity may evolve into malevolent creativity. The proposed dynamic transformation model identifies four internal driving mechanisms: cognitive control failure, situational triggers, motivational reinforcement, and moral disengagement. It also highlights one external contextual factor: insufficient social regulation. These factors may operate independently, but they may also interact and accumulate, thereby increasing the likelihood that self-serving negative creativity becomes transformed into intentional harm. Specifically, cognitive control failure may weaken children’s ability to evaluate consequences and inhibit harmful responses; situational triggers such as anger, perceived injustice, or frustration may activate revenge-oriented ideation; motivational reinforcement may transform accidental gains into intentional value-seeking through harm; moral disengagement may reduce guilt and justify harmful behavior; and insufficient social regulation may weaken the feedback link between behavior and consequence. When these factors converge, the transformation from negative creativity to malevolent creativity may be accelerated. This framework moves beyond static trait-based conceptualizations and advances a process-oriented understanding of the negative side of creativity.
Building on this evolutionary framework, this review proposes the Cognitive-Motivational-Development (CMD) intervention model. The CMD model is grounded in Yeh’s ecological systems theory of creativity development and integrates three theoretical resources: social information processing theory, task motivation theory in the creativity component model, and moral disengagement theory. It consists of three interconnected tiers. The cognitive tier targets cognitive control failure and situational information misinterpretation by strengthening children’s social information processing and creative evaluation abilities. Specific strategies include guiding children to encode multifaceted situational cues, regulate emotion-induced arousal, and evaluate the possible consequences of creative ideas. The motivational tier aims to interrupt motivational reinforcement and prevent the escalation from self-interest to harm. Strategies include validating children’s creative motivation, making visible the social consequences of their actions, identifying morally disengaged justifications, and reconstructing the value link between self-benefit and prosocial outcomes. The developmental tier addresses insufficient social regulation and weak moral constraints by fostering a supportive moral ecology. Strategies include establishing clear behavioral norms and consistent feedback mechanisms, embedding moral dialogue into daily interactions, integrating cultural values into situated activities, and creating supportive physical and social environments.
The CMD model has three major implications. First, it shifts the focus of intervention from post-harm remediation to pre-harm upstream guidance, targeting negative creativity before malevolent intent becomes stabilized. Second, it translates the proposed evolutionary mechanisms into concrete, tiered intervention targets, thereby enhancing the applicability of the model in educational settings. Third, it integrates individual-level cognitive evaluation, motivational value reconstruction, and developmental ecological support into a coherent framework, offering a holistic approach to guiding children’s creative expressions.
The theoretical contribution of this review lies in clarifying the conceptual hierarchy among creativity, negative creativity, and malevolent creativity, and in reframing the negative side of creativity as a developmental process rather than a static trait. It explains how self-serving creativity may, under specific cognitive, motivational, emotional, moral, and environmental conditions, escalate toward intentional harm. The practical significance lies in providing an evidence-informed intervention framework that bypasses the measurement dilemma of malevolent creativity and enables early, developmentally appropriate guidance of children’s negative creativity. Future research should develop validated assessment tools for negative creativity in early childhood, conduct longitudinal and experimental studies to test the proposed transformation pathways and the effectiveness of the CMD model, and translate this framework into teacher-friendly and family-friendly intervention practices.

References | Related Articles | Metrics

Etiquette on the tip of the tongue: The influence of social norms on women’s eating behavior

WANG Qingyu, CHEN Yiyue, TANG Luyao, HUANG Chaozheng, XIE Pei

2026, 34 (10): 1887-1898. doi: 10.3724/SP.J.1042.2026.1887

Abstract ( 2 )

PDF (578KB) ( 5 )

This review advances a tripartite model—norm internalization → emotion activation → behavioral regulation—to explain how social norms shape women’s eating behaviors. Building on the focus theory of normative conduct, we distinguish descriptive norms (what others do) from injunctive norms (what ought to be done) and argue that their influence depends critically on attentional focus and the depth of internalization. Unlike previous fragmented accounts, our model specifies a continuous psychological pathway. First, norms transmitted through family, media, peers, and education are selectively attended to, interpreted, and either integrated into self-concept or rejected. When descriptive and injunctive norms conflict (e.g., family’s “eat more” vs. media’s “restrict calories”), cognitive dissonance, self-doubt, and anxiety arise, whereas consistent norms facilitate smooth internalization from “I must” to “I want”. Second, internalized norms become self-standards that trigger upward social comparisons, particularly under ubiquitous thin-ideal imagery on social media. This process generates body dissatisfaction, shame, and guilt—emotions that differentially motivate behavior: shame and guilt predominantly activate avoidance motivation (restrictive eating, food avoidance), while desire for social acceptance activates approach motivation (active conformity). The resulting “pleasure-guilt conflict” (reward from high-calorie foods followed by post-consumption shame) perpetuates a cycle of restriction and compensatory eating. Third, these motivational states translate into observable eating patterns through self-control and neuroendocrine mechanisms. Chronic normative stress dysregulates the hypothalamic‑pituitary‑adrenal axis, impairs prefrontal executive function, and sensitizes reward circuitry, predisposing women to oscillate between restrictive eating (which depletes self‑control resources) and loss‑of‑control eating. The moral licensing effect further explains why successful restraint sometimes leads to subsequent indulgence.
A key theoretical innovation is integrating attentional focus as the switch between norm types. When attention lands on descriptive norms, observational learning dominates; when it lands on injunctive norms, conformity to avoid social punishment prevails. However, when a norm is highly internalized and tied to social identity (e.g., “as a woman, I should eat lightly”), behavior becomes intrinsically motivated and resistant to external pressure changes. This model also incorporates neurophysiological evidence showing that social support modulates effective connectivity from insula to dorsolateral prefrontal cortex, strengthening inhibitory control over food cues—a mechanism that can buffer against maladaptive eating.
Based on this integrated framework, we derive targeted intervention strategies. For descriptive norms, correcting normative misperceptions (e.g., overestimating peers’ unhealthy snacking) promotes healthier choices, but interventions must avoid the boomerang effect among already‑healthy individuals by using upward comparisons only with caution and adding positive identity affirmation. For injunctive norms, policy‑level tools (e.g., sugar taxes, school meal programs) effectively shift short‑term behavior, yet without internalization behaviors rebound when enforcement ends; effective applications combine choice architecture, taste optimization, and value education rather than pure coercion. Critically, building multi‑level social support systems—emotional, instrumental, and informational—not only buffers negative affect but also directly enhances prefrontal inhibitory control, helping women shift from passive compliance to autonomous healthy eating.
Finally, we situate the model in cultural context. In collectivist settings such as China, traditional values (filial piety, communal eating) intersect with modern thinness ideals and wellness culture, intensifying norm internalization but also exacerbating conflicts and emotional burdens. Emerging descriptive norms on social media showing men engaging in meal preparation may decouple dietary responsibility from female identity, offering new avenues for intervention. Future research should employ longitudinal and multi‑method designs to test the dynamic interplay among norms, emotions, and eating behaviors across cultures, and to explore how positive, autonomy‑supportive norms (e.g., body functionality, intuitive eating) can be amplified through prosocial modeling and identity affirmation. This review provides an integrative theoretical foundation for understanding and intervening in normative influences on women’s dietary practices.

References | Related Articles | Metrics

Next Issue