Next Issue

, Volume 57 Issue 6 Previous Issue   
---
Academic Papers of the 27^th Annual Meeting of the China Association for Science and Technology
Reports of Empirical Studies
Commentary


For Selected:

Download Citations EndNote Ris BibTeX

Toggle Thumbnails

Academic Papers of the 27^th Annual Meeting of the China Association for Science and Technology

Select

When AI “possesses” personality: Roles of good and evil personalities influence moral judgment in large language models JIAO Liying, LI Chang-Jin, CHEN Zhen, XU Hengbin, XU Yan 2025, 57 (6):  929-946.  doi: 10.3724/SP.J.1041.2025.0929 Abstract ( 458 )   PDF (1144KB) ( 1208 )   Peer Review Comments The rapid advancement of artificial intelligence (AI) has raised significant ethical concerns, particularly regarding the moral decision-making capabilities of large language models (LLMs). One intriguing aspect is the potential for LLMs to exhibit characteristics akin to human personalities, which may influence the LLMs’ moral judgment. Understanding how personality traits, especially the moral traits, influence these decisions is crucial for developing AI systems that align with human ethical standards. Therefore, this study aims to explore how the roles of good and evil personalities shape the moral decision-making of LLMs, providing insights that are essential for the ethical development of AI. This study investigated the roles of good and evil personalities in shaping the moral decision-making of the ERNIE 4.0 and GPT-4. Good personality was characterized by traits such as conscientiousness and integrity, altruism and dedication, benevolence and amicability, and tolerance and magnanimity. Evil personality encompassed traits such as atrociousness and mercilessness, mendacity and hypocrisy, calumniation and circumvention, and faithlessness and treacherousness. Study 1 analyzed 4000 observations. Specific prompts corresponding to different personality dimensions were designed. After specifying the type of personality, ERNIE 4.0 completed a self-report scale for good and evil personalities, evaluated whether the descriptions matched the current personality traits and provided a numerical rating indicating the degree of agreement. Study 2 recruited 370 human participants and utilized 832 LLM observations, investigated the roles of good and evil personalities in shaping the moral decision-making of the LLMs and compared with human results. Significant score differences were observed across all eight personality dimensions, with high-level manipulations significantly higher than low-level manipulations. These results demonstrate LLMs’ ability to express levels of good and evil personality traits. A comparative analysis was conducted between human participants and LLMs to evaluate the impact of these traits on CAN model in Study 2. Results showed that the patterns of personality’s influence on moral judgment exhibited both similarities and differences between LLMs and humans. GPT-4's good personality manipulation aligns closely with human results, while ERNIE 4.0 scored higher than humans on sensitivity to consequences (C), sensitivity to moral norms (N), overall action/inaction preferences (A) parameters, and utilitarianism (U). GPT-4 demonstrated better moral alignment compared to ERNIE 4.0. Furthermore, a theoretical model of good and evil personality traits in LLMs was constructed within the domain of moral judgment. This study demonstrated that LLMs effectively simulated varying levels of good and evil personality traits through personality prompts, which significantly influenced their moral judgments. GPT-4’s moral judgments aligned more closely with humans under good personality prompts, while ERNIE 4.0 consistently scored higher than humans across moral judgment indicators. Under evil personality prompts, GPT-4 exhibited lower moral norm sensitivity and higher action tendency and utilitarianism. Additionally, the influence of personality on GPT-4’s moral judgment was stronger than on ERNIE 4.0. The impact of good and evil personalities on moral judgment showed hierarchical differences, with good personality traits, particularly conscientiousness, playing a more critical role in achieving human-AI alignment in moral judgments. This research provided valuable insights into enhancing AI ethical decision-making by integrating nuanced personality traits, guiding the development of more socially responsible AI systems. References | Related Articles | Metrics

Select

The linguistic strength and weakness of artificial intelligence: A comparison between Large Language Model(s) and real students in the Chinese context GAO Chenghai, DANG Baobao, WANG Bingjie, WU Michael Shengtao 2025, 57 (6):  947-966.  doi: 10.3724/SP.J.1041.2025.0947 Abstract ( 211 )   PDF (1033KB) ( 278 )   Peer Review Comments Previous research on generative artificial intelligence (AI) has been primarily conducted in the English context, but it remains unclear about linguistic strength and weakness of generative AI in the Chinese context. This study focuses on the accuracy and normativity, affectivity, and creativity of AI in generating language knowledge, and explores its cultural adaptability and ability to generate humanized and personalized content. Evaluating and analyzing these key indicators helps us gain a deeper understanding of the linguistic strengths and weaknesses of AI, as well as cultivating the unique advantages of humans in education. By combining quantitative and qualitative methods, we evaluated the differences in knowledge accuracy, normativity, affectivity, and creativity between large language models and real students. Specifically, using an explanatory sequential design in the mixed-methods framework, we first tested group differences in each indicator among GPT-4 and ERNIE-4 versus real students on knowledge accuracy, normativity, affectivity, and creativity to test the. Next, through content analyses, we explored the specific performance of large language models on each indicator and the mechanism of their linguistic strengths and weaknesses. Study 1 found that compared to real students, GPT-4 exhibited higher accuracy in modern text knowledge (especially conceptual knowledge), but lower accuracy in ancient poetry and language usage. The knowledge normativity of GPT-4 were comparable to those of real students, while its affectivity and creativity were lower than those of real students. Moreover, the highest individual scores of GPT-4 in normativity and emotionality were on comparable with the highest scores of real students. Study 2, based on ERNIE-4, confirmed the aforementioned results, and the accuracy in ancient poetry was still lower than that of real students. The results exhibited the advantages of artificial intelligence in the areas of modern knowledge and norms, its shortcomings in ancient poetry knowledge, and its potential in affective and creative expressions. Taken together, the current findings demonstrate the linguistic strength of generative AI in the knowledge accuracy of modern Chinese literary, and the weakness regarding ancient Chinese poetry and affective and creative writings, as well as generative AI’s potential in normative and affective expressions. This sheds light on the field of the cultural adaptability, affective and creative expressions of generative AI, and has valuable implications for the AI-assistant teaching practice in the Chinese context. References | Related Articles | Metrics

Select

Cognitive outsourcing based on generative artificial intelligence: An Analysis of interactive behavioral patterns and cognitive structural features WANG Fancong, TANG Xiaoyu, YU Shengquan 2025, 57 (6):  967-986.  doi: 10.3724/SP.J.1041.2025.0967 Abstract ( 228 )   PDF (3853KB) ( 333 )   Peer Review Comments The emergence of generative AI has profoundly impacted the field of education by enabling individuals to enhance both the efficiency and quality of cognitive tasks by delegating part of the tasks to generative AI. This process is referred to as cognitive outsourcing. However, individuals’ effectiveness in using AI varies. Empirical research on the educational applications of generative AI remains limited, primarily focusing on evaluating technical capabilities and the effects of learning support. At present, the cognitive and behavioral prerequisites for effective cognitive outsourcing remain unclear. Furthermore, the differences in prior knowledge, behavioral patterns, and cognitive structures among individuals with varying performances have yet to be thoroughly explored. In this study, we designed a cognitive outsourcing activity for graduate students involving a sample of 46 participants (10 males, 36 females; age: M = 26.39, SD = 6.91). The activity consisted of two sessions. In the first session, participants were allotted 30 minutes to independently construct a concept map on the topic "Artificial Intelligence and Teachers" using pen and paper, which served as a measure of their prior knowledge. In the second session, participants engaged with a generative AI system to compose an essay on the same topic within a 100-minute time frame using a computer. The entire process was video-recorded. Based on expert evaluations, participants were categorized into high-performance and low-performance groups according to their essay scores. Interactive behaviors and contents were coded, and behavioral sequence transitions between the two groups were mapped using Lag Sequence Analysis. Additionally, Epistemic Network Analysis was employed to construct cognitive structure mappings, followed by a comparative analysis of the differences between the two groups. The results indicate that the high-performance group exhibited significantly higher prior domain knowledge compared to the low-performance group. Significant differences were observed between the two groups, including the frequency of different interactive behaviors, the frequency of different cognitive elements, the behavioral sequences, and the cognitive network structures. From the behavioral perspective, the high-performance group demonstrated significantly more diversified behavioral transitions, forming a distinctive pattern characterized by "rapid and autonomous task comprehension and planning, efficient and precise human-computer interaction, selective information extraction and deep processing." From the cognitive perspective, the high-performance group exhibited a well-balanced and comprehensive cognitive structure characterized by diverse and tightly interconnected cognitive elements. In contrast, the low-performance group displayed an unbalanced and loosely connected cognitive structure, primarily engaging with lower cognitive-level interaction. Overall, the findings indicate that effective cognitive outsourcing is a multifaceted process that necessitates active participation and profound cognitive processing. It demands proficient integration between internal cognitive frameworks and external technological tools. These findings highlight the distinct behavioral patterns and cognitive structures of individuals with varying levels of success in cognitive outsourcing activities and elucidate the cognitive and behavioral requirements for effective cognitive outsourcing. By focusing on individuals’ prior knowledge and interactive processes, this study examines the influence of cognitive and behavioral characteristics on the efficacy of generative AI-assisted writing, thus contributing to empirical research on generative AI-supported education. Additionally, it extends the theoretical understanding of cognitive outsourcing and provides insight for future research and educational practices. Furthermore, the interactive behavior and content coding framework established in this study, along with the application of Lag Sequence Analysis and Epistemic Network Analysis, provide valuable methodological references. Future research should further investigate the long-term and deep-seated effects of cognitive outsourcing on individuals with different characteristics, as well as the intrinsic neural mechanisms underlying effective cognitive outsourcing. References | Related Articles | Metrics

Select

Suicidal ideation data augmentation and recognition technology based on large language models ZHANG Yanbo, HUANG Feng, MO Liuling, LIU Xiaoqian, ZHU Tingshao 2025, 57 (6):  987-1000.  doi: 10.3724/SP.J.1041.2025.0987 Abstract ( 190 )   PDF (618KB) ( 325 )   Peer Review Comments Suicide constitutes a significant global public health challenge, with the World Health Organization reporting substantial annual mortality rates. Traditional suicide detection methods primarily depend on self-assessment scales and clinical evaluations, which require considerable resources and rely on patients actively seeking assistance. The integrated motivational-volitional (IMV) model offers a theoretical framework for comprehending suicidal behavior progression, with suicidal ideation serving as a critical risk indicator. While text-based analysis presents a promising non-invasive approach for early identification, it encounters technical challenges due to limited annotated data and linguistic complexity. Large Language Models (LLMs) offer unprecedented capabilities in language understanding and generation, potentially addressing these challenges through their ability to comprehend diverse expressions of suicidal ideation and generate high-quality training data. This research employed a two-stage design leveraging LLMs to address the challenge of limited training data for suicidal ideation recognition. In Study I, we selected ChatGLM3-6B and Qwen-7B-Chat as foundation LLMs and implemented both zero-shot and few-shot learning approaches combined with supervised learning strategies. We extracted examples from an original dataset of Weibo comments to create high-quality training data for the LLMs. Comparative experiments evaluated model performance, with human coders assessing the quality of LLM-generated texts using established suicide risk evaluation criteria. In Study II, we evaluated the impact of LLM-based data augmentation on recognition models by comparing traditional machine learning approaches with LLM-based methods trained on both original and augmented datasets, measuring performance through accuracy and true negative rate metrics. In Study I, the two self-developed LLM-based models demonstrated excellent performance in suicidal ideation data augmentation, significantly outperforming baseline models according to comprehensive evaluation metrics. The success of these LLM-enhanced models highlighted the effectiveness of high-quality data construction through advanced language modeling capabilities. In Study II, all experimental models trained on LLM-augmented data significantly outperformed their corresponding baseline models in both accuracy and true negative rate. The highest-performing model utilized the ChatGLM3-6B architecture with few-shot learning, showing marked improvements compared to its baseline counterpart. These findings demonstrate the substantial impact of LLM-based data augmentation on model generalization ability, particularly in capturing diverse and subtle expressions of suicidal ideation that traditional approaches often miss. This study validates the effectiveness of LLM-based data augmentation methods in enhancing suicidal ideation recognition while addressing data scarcity challenges. The non-invasive approach developed through LLM technology has the potential to provide timely and effective early warning of suicide risk while protecting user privacy. This research contributes to both theoretical understanding of LLMs' capabilities in complex psychological text processing and practical applications in mental health monitoring. Future research should explore cross-platform applicability of LLMs, model interpretability, and ethical considerations to further advance this promising technology in suicide prevention and broader mental health applications. References | Related Articles | Metrics

Reports of Empirical Studies

Select

Object category differences regulate the sensory dominance of the response level in an audiovisual cross-modal conflict ZHOU Heng, WANG Ai-Jun, YUAN Xiang-Yong, JIANG Yi 2025, 57 (6):  1001-1012.  doi: 10.3724/SP.J.1041.2025.1001 Abstract ( 102 )   PDF (1533KB) ( 113 )   Peer Review Comments Sensory dominance is a phenomenon in which the brain selectively processes specific sensory information when presented with multisensory inputs, thereby enhancing human perception of external stimuli. Previous studies have discussed sensory dominance at perceptual and response levels. However, how the intermediate processing level between perceptual and response levels affects sensory dominance remains unknown. Therefore, the present study adopted the cross-modal 2-1 mapping paradigm, and object categories were manipulated through three studies to investigate the role of the intermediate processing level on sensory dominance in a cross-modal conflict. In this paradigm, based on key mapping, cognitive processing levels can be defined into the preresponse level (including perceptual and semantic levels) and the response level. The difference between the audiovisual incongruent condition and the audiovisual congruent condition is called the conflict effect, and sensory dominance can be obtained by comparing the conflict effect of attention to vision and auditory. Experiment 1 manipulated the degree of difference in object categories to explore its impact on sensory dominance. Experiments 1a-1c involved animal objects (small differences), tool objects (moderate differences), and animal and musical instrument objects (large differences). A total of 30 participants were recruited for each experiment. Visual pictures reached perceptual representation early, whereas auditory sounds reached semantic representation early. Therefore, Experiment 2 (34 participants) changed visual pictures into visual words on the basis of Experiment 1c to explore the effects of the visual presentation way of object categories on sensory dominance. In Experiment 3 (20 participants), transcranial direct current stimulation (tDCS) was used on the left anterior temporal lobe, an important brain region responsible for processing object categories to study casually the effects of object category on the sensory dominance of the response level further. The results of Experiment 1 showed that, regardless of the difference in object categories, the conflict effect of attention to auditory at the preresponse level was significantly greater than that of attention to vision, that is, visual dominance. However, visual dominance at the response level appeared when the object category difference was small (Experiment 1a). Moreover, no sensory dominance was observed when the object category difference was moderate (Experiment 1b), and auditory dominance appeared when the object category difference was large (Experiment 1c). The results of Experiment 2 and Experiment 1c were consistent, that is, auditory dominance, indicating that this behavior pattern was not affected by the bottom-up visual presentation way. The results of Experiment 3 showed that under the cathodal tDCS condition, the preresponse level still showed visual dominance. However, the response level no longer showed sensory dominance. This result showed the effects of object categories on the sensory dominance of the response level from the causal level. The mechanism of sensory dominance is still under investigation. The present study was the first to find that object categories affected the sensory dominance of the response level. From the perspective of cognitive processing level, the intermediate processing level played a regulating role in the sensory dominance of the response level. This finding can enrich the explanatory theory of sensory dominance and can provide a new perspective for the study of sensory dominance in a cross-modal conflict. References | Related Articles | Metrics

Select

The multi-cue influence mechanism of spatial communication across different perspectives JIAN Jinhan, ZHANG Junheng, YAN Bihua, JI Ming 2025, 57 (6):  1013-1040.  doi: 10.3724/SP.J.1041.2025.1013 Abstract ( 222 )   PDF (18237KB) ( 186 )   Peer Review Comments Spatial communication refers to the process of exchanging spatial information among collaborators in spatial cooperation tasks. Previous research has shown that social-spatial cues, environmental cues, and layout cues can influence spatial communication. However, the exploration of their multi-cue impact mechanisms has been insufficient. Additionally, factors like perspective and field cognitive style are worthy of investigation as they may affect spatial communication through their influence on spatial perspective-taking processes and spatial cue extraction. Therefore, this study investigated the effects of consistency in multiple spatial cues, spatial perspective-taking ability, and field cognitive style on spatial communication under different perspectives. Building upon the classical paradigm of spatial communication, this study investigated the multi-cue impact mechanisms by constructing more realistic small-scale indoor virtual environments. Study 1 examined the multi-cue effects on the expression process from the first-person perspective (Experiment 1) and the third-person perspective (Experiment 2). Study 2 explored the multi-cue effects on the reception process from the first-person perspective (Experiment 3) and the third-person perspective (Experiment 4). Study 1 employed a three-factor mixed design, with the core within-subject independent variable being the consistency of spatial cues, derived from social-spatial, environmental, and layout cues. The remaining between-subject variables were spatial perspective-taking ability and field cognitive style. Dependent variables included the ratio of choosing self or other-centric reference frames, duration of language organization and expression. Study 2 utilized a four-factor mixed design, introducing an additional within-subject independent variable, the reference frame of expression statement. Dependent variables comprised accuracy and time taken for understanding expression statements. All data were analyzed using a generalized linear mixed model. Findings revealed that, in the first-person perspective, greater support for spatial reference frames led to a higher likelihood of expresser using them for spatial information, shorter language organization duration, and faster comprehension and higher comprehension correctness of corresponding spatial statements by receiver. Moreover, layout cues had a greater support effect than environmental cues. The higher the spatial perspective-taking ability, the more inclined the expresser is to choose to express spatial language using other-centric reference frame, and the more correct and time-consuming it is for the receiver to understand the spatial utterance. Field-independent expresser showed a more pronounced effect of spatial cue support in choosing reference frames compared to field-dependent expresser. In spatial communication from a third-person perspective, layout cues continued to have a support effect, while environmental cues did not. In both perspectives, receiver comprehended spatial language expressed using receiver-centric frame more quickly. However, overall, no significant difference was found in interaction performance between the two perspectives. The results indicate that: First, in the first-person perspective, there is a presence of spatial cue support effects, with the support effect of layout cues significantly outweighing that of environmental cues; regarding reference frame selection, the consistency of spatial cues has a greater impact on field-independent expresser; the higher the spatial perspective-tasking ability, the more the expresser tends to take on a higher cognitive load and the higher the efficiency of the receiver’s comprehension. Second, the third-person perspective diminishes the supportive effects of spatial cues and the influence of spatial perspective-taking ability on communication, increases the likelihood of using a self-centric reference frame to describe spatial information, and complicates the process of representation to varying degrees for the two types of field cognitive style expressers. But the use of a receiver’s frame of reference for linguistic representation is the optimal method for improving comprehension efficiency regardless of perspective. References | Related Articles | Metrics

Select

The processing bias of verbal emotional information in depression prone individuals LI Yutong, LI Xuan, YUE Zeming, LI Yahong, SUI Xue 2025, 57 (6):  1041-1055.  doi: 10.3724/SP.J.1041.2025.1041 Abstract ( 234 )   PDF (689KB) ( 340 )   Peer Review Comments Individuals with depressive tendencies are characterized by certain depressive symptoms or mild depressive conditions that do not meet the clinical diagnostic criteria for depression. Such individuals are more susceptible to developing depression compared to the general population. Previous research has demonstrated that, similarly to those with depression, individuals with depressive tendencies are predisposed to pay more attention to negative information in words or images, often at the expense of positive information. Emotional speech conveys complex information; words can more precisely articulate emotional subtleties, expressing deeper thoughts and feelings, thereby necessitating unique language processing approaches. However, studies focusing on the processing of verbal emotional information in those prone to depression are relatively scarce. This study seeks to determine whether individuals with a depressive disposition exhibit similar cognitive biases towards negative and positive information when processing textual emotional content—specifically, the extent of their biases towards negative information and their neglect of positive information. This research employs behavioral experimentation and eye-tracking technologies, designing three experiments to investigate the bias in processing emotional information in speech among those with depressive tendencies. Experiment 1 employed a 2 (groups: healthy, depressive tendency) × 3 (vocabulary valence: negative, neutral, positive) mixed experimental design, where valence served as the within-subjects factor. A total of 40 positive words, 40 negative words, and 40 neutral words were selected, along with 120 pseudowords as fillers. Participants, both with depressive tendencies and healthy, were tasked with identifying true words versus pseudowords. The results indicated no significant differences in response time or accuracy between the two groups, suggesting that when emotional information was processed indirectly and the task was straightforward, individuals with depressive tendencies did not exhibit a cognitive bias. Experiment 2 utilized the same experimental design but focused solely on emotional words (excluding pseudowords). Participants were required to assess the valence of the words. Findings revealed that individuals with depressive tendencies responded more quickly to both negative and positive words compared to the healthy group; they also showed quicker responses to positive and negative words over neutral words, whereas the healthy group responded faster to positive words than to neutral words. Experiment 3 utilized a 2 (groups: healthy, depressive tendency) × 3 (vocabulary valence: negative, neutral, positive) mixed experimental design. Emotional words were embedded within similarly structured sentences, creating 10 positive, 10 negative, and 10 neutral sentences. These sentences were divided into three interest zones: before the target word, the target word, and after the target word. Using an eye tracker, metrics such as the duration of the first fixation, the duration of the first glance, and the total duration of fixations were recorded during the reading process. The results indicated no differences in eye movement indices between the two groups in the area before the target word. However, in the target word area, a significant interaction between group and word valence was observed in the duration of the first glance, with the depressive tendency group exhibiting a significantly shorter fixation time on positive words than the healthy group. In the area following the target word, significant interactions between group and word valence were noted in the duration of the first fixation, with the depressive tendency group showing a significantly longer duration of the first fixation on negative words compared to the healthy group. Based on the results of these experiments, the following conclusions can be drawn: (1) Compared to healthy individuals, those with depressive tendencies display a processing bias towards negative emotional verbal information, although this bias is relatively mild; (2) Individuals with depressive tendencies pay less attention to positive emotional verbal information; (3) The processing bias of depression-prone individuals to negative information in context is reflected through the spillover effect. References | Related Articles | Metrics

Select

The joint role of childhood emotional abuse and bullying victimization in the development of adolescent depressive symptoms: Sequential mediation or enhanced moderation? LI Xi, ZHANG Chang, YU Ruize, YIN Yijia, ZHOU Tong, LIU Wei, CHEN Ning 2025, 57 (6):  1056-1069.  doi: 10.3724/SP.J.1041.2025.1056 Abstract ( 233 )   PDF (644KB) ( 374 )   Peer Review Comments Adolescence is a distinctive developmental phase bridging childhood and adulthood, characterized by both opportunities and challenges across mental health and interpersonal domains. Depressive symptoms and bullying victimization are prevalent in adolescence. Meanwhile, the negative developmental consequences of childhood emotional abuse also tend to emerge during this period. The interpersonal risk model highlights the risk of developing depressive symptoms from interpersonal victimization, yet several details still require empirical validation: (1) Does childhood emotional abuse positively predict the developmental trajectories of bullying victimization and depressive symptoms in adolescence? (2) Is there a time-dependent correlation between bullying victimization and the developmental trajectory of depressive symptoms in adolescence? (3) Does childhood emotional abuse moderate the longitudinal relationship between bullying victimization and depressive symptoms in adolescence? (4) Does the developmental trajectory of bullying victimization mediate the longitudinal relationship between childhood emotional abuse and adolescent depressive symptoms? Given the negative impacts of childhood emotional abuse, bullying victimization, and depression on adolescent development, exploring their trajectories and interrelationships over time has important theoretical and practical implications. A total of 521 high school students from China (Mage = 13.83 years, SDage = 1.40, 59.31% boys) were recruited to participate in a three-wave longitudinal survey study over a two-year period. The latent growth curve model (LGCM) and its variants were used to examine the mediating role of bullying victimization in the association between childhood emotional abuse and adolescent depressive symptoms, as well as the moderating effect of childhood emotional abuse on the relationship between bullying victimization and depressive symptoms. Parameter estimation was conducted using the Bayesian method. Convergence of the Markov Chain was indicated by the Potential Scale Reduction (PSR) below 1.025. Model fit was considered acceptable if the 95% credible interval of the Posterior Predictive Checking (PPC) includes zero. The results revealed that: (1) childhood emotional abuse significantly and positively predicted both adolescent bullying victimization and depressive symptoms, as shown by both the initial intercept and the growth slope. (2) Bullying victimization significantly and positively predicted adolescent depressive symptoms in terms of both initial intercept and growth slope. (3) Bullying victimization served as a mediator in the relationship between childhood emotional abuse and adolescent depressive symptoms, influencing both the initial intercept and the growth slope. (4) Childhood emotional abuse weakened the effect of the initial intercept of bullying victimization on the initial intercept of depressive symptoms, but did not significant moderate the growth slope of depressive symptoms. Findings indicate that childhood emotional abuse and adolescent bullying victimization both independently and jointly predict the development of depressive symptoms in adolescence. The joint effect was within a longitudinal sequential mediation model rather than an enhanced moderation model. Such results integrate the interpersonal risk model and the cumulative risk model into a cumulative interpersonal risk model, demonstrating how interpersonal risk factors operate sequentially across developmental stages and relational systems to influence the development of adolescent depressive symptoms. Practically, these findings highlight the need for a timely and multilayered approach in the prevention of bullying victimization and depression. Given the negative effect of childhood emotional abuse on individual’s ability to cope with subsequent interpersonal risks, timely implemented of interventions is crucial for mitigating the cumulative effect of risks over time, which may prevent the onset of severe mental health issues. References | Related Articles | Metrics

Select

The influence of moral role-based perspectives on moral judgments of third-party bystanders SONG Ru, WU Jun, LIU Caixia, LIU Jie, CUI Fang 2025, 57 (6):  1070-1082.  doi: 10.3724/SP.J.1041.2025.1070 Abstract ( 181 )   PDF (1543KB) ( 255 )   Peer Review Comments Previous studies have revealed inconsistencies in moral judgments of the same behavior when evaluated by decision-makers, victims, and bystanders in specific scenarios. However, no prior research has examined how the experiences of being a decision-maker or a victim influence individuals’ moral judgments when they assume the role of a third-party bystander. The present study employed a modified harm paradigm combined with event-related potential (ERP) techniques to investigate how pre-existing experiences with different moral roles modulate third-party moral judgments. This study comprised two experiments, each structured as a two-stage task: a moral role priming experiment and a main moral judgment experiment. In the first, participants were assigned to a decision-maker perspective group, a victim perspective group, or a control group. Participants in the decision-maker group completed moral decision-making tasks, while those in the victim group evaluated the decisions made by the decision-makers; the control group did not engage in any priming task. All three groups then participated in the second experiment, in which they assessed the morality of decisions made by anonymous decision-makers from the perspective of a third-party observer. Experiment 1 was a behavioral study involving 90 healthy adults (44 males, mean age 20.58 ± 1.92 years). It employed a mixed design with three factors: 3 (Moral Role Perspective: Decision-maker, Victim, Control) × 3 (Decision-makers’ Benefit: High, Medium, Low) × 2 (Decision Outcome: Accept, Reject). Experiment 2 utilized ERP techniques and included 54 healthy adults (28 males, mean age 21.18 ± 2.21 years), also following a mixed design comprising 2 (Moral Role Perspective: Decision-maker, Victim) × 2 (Decision-makers’ Benefit: High, Low) × 2 (Decision Outcome: Accept, Reject). The results indicated that different primed moral role perspectives significantly influenced third-party moral judgments. In both experiments, participants in the victim perspective group rendered stricter moral judgments compared to those in the decision-maker perspective group. Additionally, the level of benefits gained by decision-makers from immoral actions moderated this effect; as these benefits increased, the differences in moral judgments between the groups regarding various decision outcomes became more pronounced. ERP findings suggested distinct neural patterns associated with role-based perspectives. Participants in the decision-maker perspective group exhibited larger N1 and P2 amplitudes when observing others’ moral decisions compared to the victim perspective group. Notably, N1 amplitudes were modulated by the level of benefits, with higher benefits eliciting significantly greater amplitudes than lower benefits. Conversely, participants in the victim perspective demonstrated a significantly larger feedback-related negativity (FRN) amplitude than those in the decision-maker group. FRN results aligned with the behavioral results, showing an interaction between role-based perspective and decision outcomes. Specifically, individuals in the victim perspective group exhibited higher FRN amplitudes for “accept” decisions than for “reject” decisions, while no such significant differences were observed for participants in the decision-maker perspective group. This study highlights the challenges third-party bystanders face in maintaining impartiality in moral judgments, as prior experiences involving morality lead to varying preferences that evoke either stricter or more lenient evaluations. Under different perspective priming conditions, individuals’ moral judgments are shaped by morality-related roles they have previously occupied (Bartels et al., 2015). The ERP results indicate that perspective priming primarily influences early attentional selection and emotional arousal processes, as reflected in the N1, P2, and FRN components. These findings provide neurophysiological evidence for the role of past experiences in modulating of moral judgments. They further support the dual-process theory of morality by underscoring the importance of early emotional responses in moral decision-making. This research may enhance our understanding of how past experiences shape and update individuals’ moral standards and associated judgments. Further, it highlights the flexible nature of moral decision-making and illustrates how experiences with morality influence and refine personal standards, ultimately contributing to a deeper comprehension of the mechanisms underlying moral judgment. References | Related Articles | Metrics

Select

The double-edged sword effect of proactive helping behavior on coworker relationships CHEN Liangyong, CAO Zhonghuai, WAN Wenhua, ZHANG Weiting 2025, 57 (6):  1083-1097.  doi: 10.3724/SP.J.1041.2025.1083 Abstract ( 290 )   PDF (554KB) ( 447 )   Peer Review Comments As a common form of interpersonal interaction in organizational settings, proactive helping behavior has a substantial influence on coworker relationships. Investigating this impact is essential with regard to efforts to establish a positive team environment and facilitate healthy organizational development. Researchers have not yet reached a consensus regarding the effects of proactive helping behavior on the relationship between the helper and the recipient, and most studies on this topic have focused exclusively on either positive or negative outcomes. We propose to adopt a dialectical perspective to comprehensively explore the multifaceted effects of proactive helping behavior on coworker relationships. On the basis of affective events theory and the affect theory of social exchange, we consider the interaction effect of proactive helping behaviors and recipients' humility levels on coworker relationships. By identifying recipients' emotional responses, specifically gratitude and ability-based Mianzi stress, as potential mechanisms in this context, we provide a new perspective on the impact of proactive helping behavior on recipients' coworker relationships. We conducted two studies to test our hypotheses. In Study 1, we employed an experience sampling method to conduct a questionnaire survey among employees of several companies in China. The questionnaire survey process included an initial, one-time entry survey as well as daily surveys over a period of two weeks. The final sample included 507 usable observations that were collected from 53 employees. In Study 2, to enhance our ability to draw causal inferences, we conducted a scenario-based experiment. In this study, we recruited 216 full-time workers from companies in southern China. These respondents were randomly divided into two groups: the experimental group (proactive helping behavior, n = 107) and the control group (no proactive helping behavior, n = 109). The above studies supported our hypotheses. The results showed that the relationship between proactive helping behavior and recipient gratitude was positive when recipients exhibit a high level of humility. Furthermore, for recipients with high levels of humility, proactive helping behavior promoted their coworker relationships by activating their gratitude. However, for recipients with low levels of humility, this indirect effect was not significant. We also found that the relationship between proactive helping behavior and ability-based Mianzi stress was positive when recipients exhibit low levels of humility. In addition, for recipients with low levels of humility, proactive helping behavior inhibited their coworker relationships by eliciting their ability-based Mianzi stress. However, for recipients with high levels of humility, this indirect effect was not significant. The theoretical contributions of this study are as follows. First, this study highlights the 'double-edged sword' effect of proactive helping behavior on coworker relationships, thereby helping researchers understand the effects of proactive helping behavior in a more comprehensive and balanced manner. Second, from the perspective of the affective events theory, this study clarifies the differential emotional reactions of recipients with different humility levels when facing proactive helping behaviors. This enhances our understanding of how proactive helping behaviors influence the process of coworker relationships. Third, we identified the humility trait of recipients as a new “key” to disentangle the differentiated impacts of proactive helping behaviors. References | Related Articles | Metrics

Commentary

Select

The so-called influence relationship requires caution: Commentary on Wen et al. (2024) GE Xiaoyu 2025, 57 (6):  1098-1107.  doi: 10.3724/SP.J.1041.2025.1098 Abstract ( 1303 )   PDF (798KB) ( 1074 )   Peer Review Comments This is a commentary on a paper entitled “The influence relationship among variables and types of multiple influence factors working together” by Wen et al., published in Acta Psychologica Sinica in October 2024. They proposed a new concept called the “influence relationship.” This new concept is problematic. First, Wen et al. provided no definition for the “influence relationship,” which is unacceptable for a new-conception paper. Second, according to their proposed inference requirement, if researchers fail to disprove alternative explanations that threaten causal inferences, then they can use the term, “influence relationship,” when reporting their studies. However, this argument is a manifestation of the misunderstanding of inference requirements of causal relationships. Third, “influence” is a term that poses causal meanings according to Chinese and English dictionaries, previous academic articles, and empirical evidence. Thus, the suggestion by Wen et al. to describe a noncausal relationship using “influence” can result in an overstatement of research significance and misunderstanding among fellow academics and public readers. This scenario is contradictory to the increasing expectations of researchers of more rigorous scientific language. Fourth, Wen et al. were confused with goals and the realization of such goals. Failure to disprove alternative explanations is a compromise or a limitation in methods instead of a unique goal. Wen et al. stressed that a “directional correlation” lacked an appropriate name in academia. Therefore, they called it the “influence relationship.” This stance is seemingly an unfair description of the academic status quo because researchers typically adopt the word, “predict,” to describe a directional correlation. Based on previous articles, this commentary proposes another framework for the categorization of variable relationships. At the goal level, causal goals—in which researchers hypothesize a difference in Y if X is deliberately changed—can be distinguished from noncausal goals. Furthermore, noncausal goals can be classified as predictive goals (e.g., using texts to predict mental disorder risks or test scores to predict future performance) and purely correlational goals (e.g., a shopping basket analysis or a correlation analysis between a newly proposed personality construct and the Big Five). Neither is concerned with alterations to X. At the realization level, if a researcher opts for a causal goal but fails to provide sufficient evidence to support causal relationships, then they are expected to avoid causal language (e.g., “influence”) when reporting results and key conclusions. Alternatively, they can use terms such as “be associated with” and “predict” if appropriate. Moreover, this commentary provides authors and reviewers with several practical suggestions. (A) Clearly define research goals because the different criteria to evaluate causal, predictive, and purely correlational studies should be followed. (B) Enable researchers to discuss causal meanings conveyed by their results even if they fail to offer sufficient causal evidence when targeting causal goals. This statement does not mean an encouragement of overstatement; conversely, only if researchers clearly define their causal goals can they admit the extent to which they are realizing such goals. (C) Use noncausal language to report noncausal results frankly rather than using euphemisms as a strategy for impression management. (D) Avoid an all-or-none attitude toward causal evidence; instead, value every effort that helps disprove alternative explanations and provides more confidence in causal propositions. (E) Do not rely on a single study (even a randomized experiment) to provide conclusive answers to causal questions; instead, value the accumulation of evidence and triangulation. References | Related Articles | Metrics

Select

Distinguishing between causality, influence, correlation, and prediction WEN Zhonglin, MA Peng, MENG Jin, WANG Yifan 2025, 57 (6):  1108-1118.  doi: 10.3724/SP.J.1041.2025.1108 Abstract ( 1211 )   PDF (518KB) ( 1201 )   Peer Review Comments Wen et al. (2024) discussed the following three issues: (1) explaining why it is inappropriate to understand influence relationship between variables as causal or correlation relationship, and then providing the definitions for two terms, influence relationship and influence factor; (2) summarizing several ways to find evidence for justifying the directionality when modeling the influence relationship. (3) categorizing multiple influence factors working together. Ge (2025) questioned Wen et al.’s article as follows: (1) the concept of "influence relationship" is not clearly defined; (2) influence relationship and causal relationship are indistinguishable; (3) one cannot create a new goal just because the means cannot provide causal evidence for the goal of causality; (4) the so-called influence relationship should be called “prediction”. In response to these concerns, the present article offers clarifications and justifications. First, the influence relationship has been rigorously defined by using a "Genus and Differentia" approach in Wen et al.’s paper. The influence relationship can be determined by using logical reasoning and statistical correlation testing. We also explain “correlation”, the genus concept of the influence relationship. Furthermore, any causal relationship is the influence relationship, and the two are equivalent in studies through randomized controlled experiments, whereas in other contexts influence relationship may not necessarily be causal relationship. We provided easily understandable cases where the influence relationship was established but the causal relationship was not. It also presented and explained the status and role of proxy effects in those cases. Beyond experimentation, establishing influence relationships is a suitable goal for investigation and research, which is better than the goal of establishing correlation. In statistics, any “correlation” can be used for “prediction”, and the direction of prediction can differ from the actual direction of the variable relationship. Prediction is essentially statistical inference based on the relationships between variables, but it is not the relationships themselves. Causality, influence, and correlation are all relationships between variables, whereas prediction concerns the application of variable relationships. Therefore, prediction is not an appropriate substitute for influence relationship. References | Related Articles | Metrics