http://psy-journal.hse.ru en-us Copyright 2026 Wed, 24 Jun 2026 22:20:17 +0300 https://psy-journal.hse.ru/en/2026-23-2/1171064577.html https://psy-journal.hse.ru/en/2026-23-2/1171064715.html The perception of voice plays an important role in auditory learning, as it is associated with cognitive load, emotional involvement, as well as memory-related processes. The study examines the types of narrating voice (natural human or AI-generated) as psychological mediators of learning activity. Within the cultural-historical and activity-based frameworks, listening is considered an internally structured cognitive–affective process influenced by mediational means. In our experimental design, undergraduate MIPT and MSLU students listened to an educational audio text presented in one of three voice conditions: professional female (natural), native male (natural), or synthetic. Immediate recall was assessed using a standardized comprehension test. The results demonstrated significantly higher retention under both natural voice conditions compared to the synthetic voice, regardless of the students’ disciplinary backgrounds. These findings suggest that natural prosody and emotional expressiveness may reduce extraneous cognitive load and facilitate deeper encoding of auditory information. From a psychological standpoint, voice quality can be seen as a mediating tool that promotes psychological comfort and sustained attention, resulting in reduced learning anxiety and improved subjective satisfaction with learning. These results can be used in recommendations for selecting and adapting auditory and hybrid learning formats. Specifically, when using AI voiceover in educational courses, its potential impact on learners’ emotional states and risks of cognitive and emotional strain should be considered. The study contributes to the substantiation of psychological approaches to the implementation of AI technologies in higher education pursuing psychological well-being and preventing emotional burnout. https://psy-journal.hse.ru/en/2026-23-2/1171064802.html This study investigates how explicit authorship labels — particularly those involving artificial intelligence (AI) — affect readers’ comprehension, trust, and intention to use scientific information. In a between-subjects online experiment (N = 183), university students read the same AI-generated science text about superconductors, with one of three authorship cues: (1) no label (control), (2) “AI-generated,” or (3) “AI-generated, human-edited.” All participants completed a five-item multiple-choice comprehension quiz (MCQ), a semantic differential scale assessing trust, and a self-report measure of behavioral intention to use or share the text. Results showed no statistically significant differences in comprehension scores across conditions, confirming that the inclusion of authorship labels did not impact factual understanding. However, the “AI-generated” label sharply reduced both trust and intended use compared to the control group (Hedges’ g > 1.1), indicating an aversion effect. Importantly, appending the phrase “human-edited” to the AI label eliminated this trust penalty: responses in the “AI+Human” condition were statistically indistinguishable from the control group. These findings align with prior cognitive and communication research, which suggests that while source cues rarely alter comprehension, they substantially influence user trust and perceived credibility. The study highlights the importance of carefully framing AI involvement in educational or scientific content. It suggests that emphasizing human oversight – through transparent labeling – can mitigate skepticism and foster appropriate trust, even when the content is AI-generated. Implications for ethical AI disclosure policies in higher education are discussed. https://psy-journal.hse.ru/en/2026-23-2/1171064996.html The demand for developmental diagnostics in personalized education reveals a fundamental crisis in classical psychometrics. This paper presents a conceptual blueprint for a new generation of LLM-based diagnostic systems engineered to ensure methodological validity in high-stakes educational contexts. The project rests on five interconnected principles: methodological grounding in evidence-centered design (ECD), a hybrid 'centaur' architecture, echeloned quality control, role-oriented utility, and data sovereignty. Its core innovation is an 'emulated expertise architecture' that emulates not the expert's consciousness but their idealized professional workflow, decomposing the act of judgment into a network of verifiable procedures – systematic falsification, independent peer review (supervision), and dialectical synthesis – executed by a multi-agent system of functionally specialized “prompt archetypes” (“researcher”, “auditor”, “communicator”). The entire pipeline is governed by the 'human-over-the-loop' (HOTL) ethical imperative, positioning the system as a tool for expert augmentation rather than replacement. This approach facilitates a paradigm shift from a “psychometrics of measurement” to a “psychometrics of understanding”, generating not merely a score but a dynamic, contextualized model of the learner's cognitive processes. Supplementary materials detail the five-phase methodology for translating psychometric instruments into behavioral analysis protocols and the complete microarchitectures of all agent archetypes. By integrating computational psychometrics with educational psychology, the proposed architecture opens a pathway toward valid, reliable, and interpretable tools, marks a shift from static measurement to a dynamic understanding of students' educational preferences and outcomes, and offers a practical instrument for advancing student-centered university education. https://psy-journal.hse.ru/en/2026-23-2/1171065069.html The relevance of this study is determined by the growing use of artificial intelligence in education and research, including tasks related to the interpretation of values and the formation of value orientations among young people. The aim of the article is to compare the semantic content of the value “dignity” represented in Russian linguoculture with the interpretation of this concept generated by artificial intelligence. The research material consisted of lexicographic and encyclopedic definitions of the Russian word dostoinstvo (“dignity”) and texts generated by ChatGPT 3.5 and 4.0. The methodological framework combined psycholinguistic and value-semantic approaches. The study employed definitional analysis, comparative analysis, and a specialized semantic differential technique; 30 respondents aged 18–24 participated in the experiment. The data matrix was processed using principal component analysis followed by Equamax rotation with Kaiser normalization. Six factors in the understanding of dignity were identified: inner self-worth, a humanistic understanding of dignity, personal autonomy and freedom, honesty, rejection of superiority over others, and interpersonal respect. The findings show that artificial intelligence represents dignity mainly as a rational moral and legal category, whereas dictionary and encyclopedic sources demonstrate the emotional-evaluative, cultural, and behavioral saturation of this concept. The results may be applied in educational psychology when developing practices for the critical use of AI in educational and value-oriented work. The study concludes that hybrid models are needed: AI may function as a cognitive and analytical tool, but the interpretation and transmission of value-laden meanings require human participation. https://psy-journal.hse.ru/en/2026-23-2/1171065137.html This review of research synthesizes how large language models (LLMs) are being used to teach Russian as a foreign language (RFL) across three research traditions – in Russia, English-language contexts, and China – covering January 2023 to October 2025. The article draws on 55 peer-reviewed and practitioner sources to identify recurring themes, benefits, risks, and implementation conditions. Across settings, the literature reports five recurring benefits: (1) expanded out-of-class practice via chatbots and conversational agents; (2) personalization that supports learner autonomy and differentiation; (3) development of all four skills and cultural competence through multimodal, task-based activities; (4) faster formative feedback on writing and grammar through exemplars and targeted drills; and (5) access to authentic, level-adaptable materials. Reported risks are comprised of factual and grammatical errors, weak pragmatic or cultural fit, over-reliance that may erode independent strategies, unequal access and data-privacy constraints, and unsettled assessment practices in AI-rich classrooms. Regional emphases differ: Russian studies foreground linguodidactic integration and teacher oversight; English-language scholarship brings into focus ethics, critical AI literacy, and customized tutoring; Chinese research advances human–AI co-teaching and “virtual + real” classroom models. Practical guidance follows: design AI-mediated tasks with explicit objectives and boundaries, train verification habits (cross-checking, source attribution, corpus-based checks), and pair AI feedback with human review. Research priorities include Russian-specific models and prompting resources, longitudinal evidence on learning outcomes, validated rubrics for AI-assisted work, and teacher professional development. Overall, LLMs can enhance RFL when their use is deliberately orchestrated and educators remain the cultural and pedagogical anchor. https://psy-journal.hse.ru/en/2026-23-2/1171065224.html Metacognitive monitoring plays a key role in the problem-solving process, enabling individuals to evaluate their progress toward a solution and, when necessary, adjust the strategies they employ. Confidence is one of the universal indicators of metacognitive experiences, used for making decisions about continuing, modifying, or terminating work on a problem. The main theoretical framework of this study is the Diminishing Criterion Model (DCM; Ackerman, 2014), which posits that the confidence threshold sufficient for accepting an answer decreases over time, allowing for an optimal balance between cognitive effort and solution quality. We assumed that when help is available, the criterion line parameters would change: its slope would decrease, reflecting higher requirements for answer quality. Forty university students (aged 18–44) participated in the experiment. They solved Compound Remote Associates (CRA) problems. In the first phase, help was unavailable; in the second, participants could request an additional cue word. For each trial, response time and subjective confidence were measured both at the moment of seeking help and after completing the task. Modeling revealed differences in the slopes of the criterion lines: when help was available, the decline in confidence requirements occurred more slowly. Intercepts corresponding to confidence in spontaneous answers did not differ. Cluster analysis identified several strategies for seeking help. According to the obtained data, the possibility of seeking help increases the minimal confidence level for accepting an answer, even when help is not actually used. This confirms the role of help availability as an additional external resource taken into account in metacognitive evaluations. https://psy-journal.hse.ru/en/2026-23-2/1171065358.html Social networks have intensified conspicuous consumption, turning visual travel‑related content into a key tool for status self‑presentation and the affirmation of social position. The present study is aimed at examining the socio‑demographic markers of conspicuous consumption on social networks and its relationship with loneliness. The study involved 100 respondents (M = 39.34; SD = 10.31), including 43% men and 57% women, who actively travel and post photographs from their trips on social networks. The following instruments were used: the Conspicuous Consumption of Posting Travel Photos Scale (CCPS) and the Social and Emotional Loneliness Scale for Adults (SELSA). Data were analyzed using confirmatory factor analysis, as well as correlational and comparative analyses. Four factors of tourism snobbery were identified: “pure conspicuous consumption” — the publication of status‑related photographs emphasizing wealth and belonging to the upper class; “group inclusion” — the desire to conform to trends through images from popular destinations; “uniqueness of the place” — the demonstration of rare locations to stand out and emphasize the exclusivity of the experience; and “uniqueness of the self” — the desire to showcase originality and aesthetic taste through visual content. Gender differences manifest in the fact that women more often use travel‑related content to express their uniqueness. With age, the need for conspicuous consumption decreases; income level also affects the nature of self‑presentation, as respondents with higher income are less likely to emphasize status. In addition, the results showed that posting photographs with the aim of joining a group is positively correlated with romantic and social loneliness, but negatively correlated with family loneliness. A relationship was found between social loneliness and the motivation to demonstrate the uniqueness of the visited place and one’s own personality, which indicates an attempt to compensate for a lack of social ties through self‑presentation in the digital environment. Overall, conspicuous consumption on social networks can be viewed as a dynamic phenomenon situated at the intersection of social comparison, self‑presentation, and loneliness. https://psy-journal.hse.ru/en/2026-23-2/1171065431.html The article is devoted to the analysis of contemporary charitable practices among Russians as a form of social solidarity. Charity is conceptualized as an activity characterized by specific motivations (both positive and negative), goals, and a set of stereotypical actions. The study examines the influence of interpersonal factors (empathy, sympathy, social phobias), anxious expectations, and media preferences on charitable attitudes, understood as internalized behavioral dispositions. The theoretical framework draws on modern models and our previous research on solidarity during the pandemic, which demonstrated that empathy is not a principal leading predictor of prosocial behavior, whereas motivation and existential fears play a significant role. The hypotheses were tested on a sample of 181 respondents (January 2024 – May 2025). The questionnaire was developed in field research (participant observation, expert interviews, document analysis) and included scales for cognitive and affective empathy (QCAE, adapted by Belousova and Geyvandova, 2021), Osgood’s semantic differential for measuring sympathy, scales of positive and negative motivation, as well as indicators of anxious expectations and media preferences. Data analysis was performed using structural equation modeling. The findings indicate that charity is more strongly driven by positive motivation than by interpersonal relationships. Sympathy was found to be unrelated to prior experience with charitable giving; however, empathy may enhance certain forms of assistance directed toward specific groups in need. Anxious expectations are primarily associated with negative motivation. Among all media types, only the press demonstrates a consistent positive association with charity and positive motives, which can be explained by the cultural authority of the printed word in Russia. The results of the study make it possible to develop recommendations for the media, charitable organizations, and educational institutions of various levels. https://psy-journal.hse.ru/en/2026-23-2/1171065483.html Within the domain of creativity, a significant challenge arises concerning inadvertent plagiarism, where individuals believe they have created a novel and original product, yet in fact unwittingly incorporate elements from earlier sources. Specifically, it remains unclear how this phenomenon relates to insight, particularly whether the aha-experience might accompany the generation of ideas which are actually recollections of previously encountered information, thus reflecting source-monitoring errors. The current experimental study aimed to test the hypothesis that during independent generation of verbal puzzles, participants would borrow ideas from puzzles they had previously solved, thus demonstrating the phenomenon of inadvertent plagiarism. Additionally, it was investigated whether the generation of ideas with borrowings would occur more frequently with insight and be accompanied by an aha-experience, due to their higher retrieval fluency. The study involved 48 volunteers. The procedure consisted of two stages. At first, participants were presented with remote associate tasks and attempted to solve them. One week later, in the second stage participants were asked to generate new verbal puzzle problems using provided solution words, half of which they had already encountered during the first phase, whereas the other half were new. The results demonstrated that the proportion of borrowing in the experimental condition, i.e., generating verbal puzzles corresponding to familiar solution words, significantly exceeded chance level, thus being evidence of inadvertent plagiarism. However, neither the reported instances of insight nor the ratings of aha-experiences differed significantly between the generation of verbal puzzles involving borrowings and the generation of completely new ones. These findings contribute to existing theories of inadvertent plagiarism and suggest directions for integrating research on insight, plagiarism, and creativity. https://psy-journal.hse.ru/en/2026-23-2/1171065541.html Professional identity (PI) is a complex substantive and procedural phenomenon, a context determined by readiness for perception and perceptual attitudes. Modern requirements make it necessary to study the functional role of a teacher's PI as a context. Creativity is considered as a meaningful component of PI and a factor in teaching effectiveness. The purpose of this work is to assess the impact of contextual priming of professional success on the dynamics of teachers' professional identity and creativity experimentally. The sample consisted of 220 people aged 22–59 years (M=40.60), 88% women, 12% men. The participants were divided into two professional groups: teachers (n=98) and other professionals (n=122), who were then randomly assigned to the experimental conditions (priming successful/beginner). Research methods: Test of Twenty Statements (TTS), the method of studying professional identity (MIPI) by L.B. Shneider, an adapted version of the method of diagnostics of verbal creativity by R. Bekkhat. As an experimental impact, priming with textual settings, reinforced by a visual context, was used. The results of the study showed that priming for professional success significantly increases the overall level of PI, but also leads to an increase in the number of participants with the pseudo-identity status. The mature PI status (achieved identity), on the other hand, showed resistance to short-term experimental exposure. The PI of teachers is a variable context for the development of their creativity. A positive effect of integrating the context of the attitude towards success and the degree of agreement with it has been identified, which contributes to increased fluency and flexibility in non-pedagogical fields. As for teachers, the integration of the contexts of the attitude towards success and PI restrains the spontaneous flexibility of verbal creativity and is qualified as a specific conflict of contexts (functional rigidity). The practical significance of the results lies in the understanding that the manifestation of teachers' creativity can be influenced by setting appropriate contexts for professional success and integrating them with the context of professional identity to prevent rigidity and burnout. https://psy-journal.hse.ru/en/2026-23-2/1171065645.html The present study focused on the basic level advantage effect in hybrid search. Basic level advantage effect refers to preferable categorization at the basic level as compared to the subordinate or superordinate levels. The goal of the study was to reveal the guidance process if multiple target templates are held in memory. An eye‑tracking method was used to separate the entire search process into a guidance and a verification. Classical hybrid search procedure was used: participants had to search for 1, 2, 3 or 4 targets among 8, 12 or 16 stimuli on the screen. Targets were defined verbally at the beginning of the block by the names of category – basic or superordinate. Images of photorealistic objects were used as stimuli; pictures were generated by Stable Diffusion neural network. Category levels affected the guidance duration (basic level targets were found faster as compared to superordinate) in conditions with the visual set size 16 and the memory set size 1, but not in other conditions. For verification, however, there was no effect of category level, revealing different ways in which representations are stored. The results are consistent with Guided Search 6.0 theory and reveal the existence of the basic level advantage for the classical visual search, but not for the hybrid search.