The study, published in Frontiers in Psychology, tested a moderated dual-path model using an electronic questionnaire distributed from January to February 2026 to undergraduate and postgraduate design majors at comprehensive universities and specialist art and design institutions in multiple regions of China. Recruitment combined convenience and snowball sampling. Of 632 submitted questionnaires, 568 were retained as valid, a valid response rate of 89.87%. The sample included 286 men and 282 women, with 78.86% aged 18–26. Students represented associate-degree, bachelor’s, master’s and doctoral programs, and 51.58% reported having received formal AI training. A pilot survey of 102 design students in the Nanjing area was used to test and refine the questionnaire before the formal data collection.

The final instrument contained 28 seven-point Likert-scale items measuring five constructs. AI technological disruption awareness, abbreviated AITDA, captured perceptions that generative AI and intelligent design tools could replace existing design skills, reduce the value of learning traditional techniques, or intensify competition and collaboration with AI. Promotion focus measured the tendency to use AI in pursuit of creative goals, positive outcomes and improved skills. Prevention focus covered caution, ethical compliance, reliability, error checking and meeting course requirements. AI critical thinking, or AICT, reflected the monitoring, evaluation and adjustment of students’ design reasoning when using AI. Design innovation competence, or DIC, encompassed novel concepts, responsiveness to users and contexts, multiple solution paths, actionable prototyping, and functional, experiential or social value.

The model was grounded in regulatory focus theory, which distinguishes motivation directed toward growth and gains from motivation directed toward safety and loss avoidance. AITDA was treated as a situational signal that could activate either orientation. An opportunity appraisal could encourage experimentation with new tools, exploration of human–machine collaboration and pursuit of ambitious creative goals. A threat appraisal could increase vigilance about professional displacement, academic performance and responsible use, potentially narrowing conceptual exploration or producing cautious strategy choices. Confirmatory factor analysis supported a five-factor measurement model, with CFI and TLI both 0.986, RMSEA of 0.028 and a chi-square-to-degrees-of-freedom ratio of 1.46. Cronbach’s alpha values for the five constructs ranged from 0.897 to 0.932, and a split-half analysis produced a similar five-factor structure in separate halves of the sample.

Hierarchical regression supported the predicted sequence of statistical associations. Awareness of AI disruption was positively associated with promotion focus, while promotion focus was positively associated with design innovation competence. In the model containing both mediators, promotion focus had an unstandardized coefficient of 0.209, while prevention focus had a coefficient of −0.393; both were significant at p < 0.001. Awareness of disruption also had a positive direct association with competence, with an unstandardized coefficient of 0.342 in that model. Bootstrap mediation analysis with 5,000 resamples estimated a positive indirect effect through promotion focus of 0.0929, with a bias-corrected 95% confidence interval from 0.0532 to 0.1416. The corresponding effect through prevention focus was negative at −0.0884, with a confidence interval from −0.1229 to −0.0532.

Because the two indirect effects were similar in absolute magnitude but opposite in direction, their combined total indirect effect was 0.0045, with a 95% confidence interval from −0.0433 to 0.0596 that included zero. The promotion pathway accounted for 26.8% of the total association, while the prevention pathway accounted for −25.5%, making the balancing pattern nearly symmetrical under the researchers’ predefined criterion. The near-null aggregate indirect effect does not mean that the motivational mechanisms were unimportant. Instead, the results indicate that averaging across students concealed two meaningful but offsetting associations with AI disruption. Because all variables were measured at one time, however, the model identifies statistical relationships rather than causal effects, and it cannot show whether awareness produces innovation.

AICT significantly changed how disruption awareness related to both regulatory orientations. The interaction between AITDA and AICT was positive for promotion focus, with an unstandardized coefficient of 0.2088, and also positive for prevention focus, with a coefficient of 0.1867; both interactions had p < 0.001. The second result contradicted the researchers’ hypothesis that critical thinking would weaken the association between disruption awareness and prevention focus. At low AICT, the simple slope from AITDA to prevention focus was negative, at −0.1688, but at high AICT it became positive, at 0.3270. By contrast, the association between AITDA and promotion focus increased from 0.2033 at low AICT to 0.7578 at high AICT. Additional tests found no significant moderation in the stage linking either regulatory focus to design innovation competence, indicating that AICT was associated principally with the interpretation of disruption and activation of a motivational orientation.

Conditional mediation analysis showed that the indirect associations depended on the level of critical thinking. Through promotion focus, the indirect association was positive and significant at all three tested levels: 0.042 at low AICT, 0.100 at mean AICT and 0.158 at high AICT. The prevention pathway changed direction across the same range. Its conditional indirect effect was significantly positive at low AICT, at 0.066, was not significant at the mean, at −0.031, and became significantly negative at high AICT, at −0.129. The indices of moderated mediation were significant for both pathways: 0.044 for promotion focus and −0.073 for prevention focus. The authors interpret the pattern as a possible change in the character of prevention motivation, arguing that students with higher AICT may express reasoned vigilance based on recognized limitations, bias, ethics and social impact rather than diffuse anxiety. That interpretation remains a theoretical explanation of the measured association, not a demonstrated mechanism.

Subgroup analyses broadly preserved the principal pattern. The positive association between disruption awareness and design innovation competence, the positive promotion-mediated pathway and the negative prevention-mediated pathway remained significant across groups divided by education, age, AI experience and formal AI training. The AITDA-by-AICT interactions also remained positive and significant in education-level and AI-experience subgroups. Formal training was associated with a weaker AITDA–prevention-focus coefficient among trained students, 0.186, than among untrained students, 0.300. The authors suggest that structured preparation may attenuate threat-based prevention reactions, but the study did not randomly assign training and did not test that explanation experimentally. All respondents studied design, preventing comparisons across specialties, and the non-probability sample restricts how broadly the results can be generalized beyond the surveyed Chinese university population.

For design education, the authors recommend cultivating AICT through activities such as checking AI-generated proposals, examining bias and intellectual-property concerns, reflecting on ethical responsibility and evaluating social consequences. They also propose differentiated support: students with lower AICT may benefit from structured experiences that frame AI as an enabling partner, while students with higher AICT may benefit from opportunities to turn vigilance into output auditing, bias checking and responsible-innovation practices. Future-oriented human–machine design projects and classroom environments that permit exploration and failure could further help students approach disruption as a challenge. These are educational implications drawn from observational associations; the research did not test a curriculum, workshop or training intervention, so the effectiveness of those proposals remains unverified. The results also do not establish that critical thinking is more useful than tool training alone.

Several limitations constrain the conclusions. The cross-sectional, single-wave design precludes causal inference and cannot show whether AI critical thinking changes regulatory focus over time, or whether changes in regulation precede changes in innovation. All core variables were self-reported, leaving possible common-method bias; although Harman’s single-factor test and a latent method-factor analysis suggested that common method bias was not severe, the first factor accounted for 31.54% of the variance. Design innovation competence also lacked objective measures such as faculty ratings or assessed portfolios. Non-probability sampling, possible non-response bias and the restriction to Chinese design students limit external validity, while the absence of design-specialty comparisons leaves disciplinary boundary conditions unresolved. The authors call for longitudinal or experimental studies, objective innovation assessments, cross-national and cross-specialty samples, and research examining additional factors such as self-efficacy, peer climate, institutional support and interactions with teachers, schools and industry. Until such evidence is available, the results describe a nuanced statistical pattern rather than a proven causal relationship between AI awareness, critical thinking and design innovation.

Subject of Research: How AI disruption awareness and AI critical thinking relate to design innovation through promotion and prevention motivation among Chinese design students

Article Title: The double-edged sword effect of AI technological disruption awareness on design innovation competence: the mediating role of regulatory focus and the moderating role of AI critical thinking

Article References: Wang, X., Hou, J., Luo, T., Du, J., & Yuan, Z. (2026). The double-edged sword effect of AI technological disruption awareness on design innovation competence: the mediating role of regulatory focus and the moderating role of AI critical thinking. Frontiers in Psychology, 17, Article 1883577. https://doi.org/10.3389/fpsyg.2026.1883577

Image Credits: AI Generated

DOI: 10.3389/fpsyg.2026.1883577

Keywords: artificial intelligence, design education, design innovation, AI disruption awareness, regulatory focus, AI critical thinking, promotion motivation, prevention motivation, cross-sectional survey, Chinese design students, double-edged, sword

The study focused on self-care, and specifically on mindfulness practices, a category of mental health behavior that has surged in popularity through apps, podcasts, and guided audio exercises. Unlike many physical health behaviors whose outcomes are distant and probabilistic, self-care is something people can do immediately and privately, which makes it an ideal testbed for framing effects. The researchers distinguished between two types of framed messages. Gain-framed messages emphasized the benefits of self-care, describing what people could obtain by practicing it, such as improved wellbeing and reduced distress. Nonloss-framed messages highlighted the distress that self-care could prevent, framing the behavior in terms of avoiding negative outcomes. A control condition received information unrelated to self-care, giving the team a baseline against which to compare the two framing strategies.

Participants were recruited through social media and university mass email and were randomly assigned to view one of the three message types. The design went beyond a simple comparison, however, because the authors were interested in whether framing effects depend on who is receiving the message. They measured two candidate moderators. The first was regulatory focus, a trait-level orientation from psychological theory that describes whether people chronically approach rewards, a promotion focus, or prioritize avoiding losses, a prevention focus. The second was perceived risk of worsening mental health, essentially how much a person believes their mental state could deteriorate. According to the matching hypothesis that has guided much of this literature, gain frames should work best for promotion-focused individuals, while nonloss frames should resonate with prevention-focused individuals or those who feel at risk.

The outcome measures captured three stages of behavioral engagement: intention to practice self-care, actual uptake, and depth of engagement, with assessments taken immediately after the message and again one week later. Uptake and engagement were operationalized around mindfulness audio materials, allowing the researchers to observe not just what participants said they would do but what they actually did, including how long they played the audio. The analysis relied on analyses of covariance and regression models, with attention to correcting for multiple testing, a methodological caution the authors credit in part to consultations during the review process.

The headline finding was a gain-framing advantage for self-care intention. Participants who read the gain-framed message reported higher intention to practice self-care than those in the control condition, and the data pointed suggestively in the same direction when compared with the nonloss-framed condition. In other words, telling people what they stand to gain from caring for their minds appears to be a more effective motivational strategy than telling them what they might avoid losing, at least when it comes to forming intentions. This aligns with a broader pattern in the framing literature, where gain frames tend to outperform loss frames for behaviors perceived as preventive and low-risk.

Interestingly, the behavioral data told a slightly different story. There was tentative evidence that participants in the nonloss-framing condition played the mindfulness audio for a longer duration than those in the control condition. This divergence between intention and behavior is a recurring puzzle in health psychology: the message that best motivates stated plans is not always the message that best drives actual engagement. The authors describe this behavioral effect as tentative, reflecting the statistical caution they applied, but it hints that nonloss framing may have a subtle role in sustaining engagement even when it does not win the intention contest.

The moderation results are where the study delivers its most consequential message. For people with lower perceived risks of worsening mental health, gain framing produced higher self-care intention than both the control condition and, tentatively, the nonloss condition. This suggests that when individuals do not feel their mental health is in jeopardy, emphasizing positive benefits is the more persuasive route. However, no evidence of moderation by perceived risk emerged for the other dependent variables, and, perhaps most strikingly, the results showed no support for moderation by regulatory focus at all. The long-standing theoretical expectation that promotion-focused people respond better to gain frames and prevention-focused people to nonloss frames did not hold in this mental health context.

This null result matters because it contradicts a substantial body of work on regulatory focus and framing interactions in physical health. The authors are explicit about the implication: their findings do not support applying tailored framing based on regulatory focus when encouraging mental health self-care. For intervention designers, that is both a warning and a relief. Tailoring messages to individual personality profiles is technically demanding, requiring assessment of traits and dynamic message generation. If that complexity does not buy measurable improvements in this domain, resources may be better spent elsewhere. At the same time, the finding that perceived risk moderates the framing effect offers a simpler, more actionable tailoring variable: people who feel vulnerable may not need benefit-focused persuasion, while those who feel secure may respond well to it.

The study was conducted with ethics approval from the Survey and Behavioural Research Ethics Committee of The Chinese University of Hong Kong, and it forms part of a larger research program funded by the Research Grants Council of Hong Kong on personalized nudging to promote uptake of and engagement in self-care through digital mental health interventions. The work began as part of the first author’s doctoral dissertation and evolved substantially through peer review, with the authors acknowledging input from colleagues on study design, moderation methodology, and statistical corrections. The team is transparent about the limitations inherent in their design, noting that the effects they were hunting for may be very small.

Indeed, the authors themselves frame the path forward clearly. They call for field studies, longitudinal designs, and well-powered studies capable of detecting very small effects to determine whether gain framing holds an advantage for actual self-care behavior over longer periods. A single laboratory-style exposure followed by one week of follow-up may simply be too brief and too controlled a setting to capture how framing shapes habits that unfold over months. Still, for a field increasingly dominated by digital interventions that must decide, algorithmically and at scale, what message to show whom, this experiment offers a valuable calibration. The safest evidence-based default for encouraging mental health self-care appears to be gain framing, with perceived risk, not personality, as the most promising axis for personalization, and with a lingering, tantalizing hint that loss-avoidance language may quietly extend the time people spend actually practicing mindfulness once they begin.

Subject of Research: Message framing effects on mental health self-care intention and mindfulness engagement

Article Title: Impact of gain and nonloss framing in mental health self-care: investigating regulatory focus and perceived risk of worsening mental health as moderators

Article References: Yeung, S. K., Mak, W. W. S., Zhao, H., & Lee, J. C. M. (2026). Impact of gain and nonloss framing in mental health self-care: investigating regulatory focus and perceived risk of worsening mental health as moderators. BMC Psychology. https://doi.org/10.1186/s40359-026-05715-8

Image Credits: AI Generated

DOI: 10.1186/s40359-026-05715-8

Keywords: message framing, mental health self-care, mindfulness, regulatory focus, perceived risk, gain framing, nonloss framing, health communication, digital mental health interventions, health psychology, behavior change, randomized experiment