Christopher M. Dobmeier

This study examined self-relevance cues in narrative health messages by combining deepfake-based facial similarity (visual cue) and perspective-taking (linguistic cue). A lab experiment (N = 204) found that both high facial similarity and first-person perspective increased persuasion. Identification, however, was influenced only by perspective-taking, not facial similarity. AI literacy moderated these effects distinctly: for identification, high-literacy participants showed suppressed identification when third-person framing and lower similarity co-occurred, indicating sensitivity to the absence of both cues. For persuasion, high-literacy participants were persuaded under first-person framing regardless of similarity, while low-literacy participants required both first-person framing and high similarity together to be persuaded—suggesting they compensate for limited metacognitive resources by relying on concurrent cues. These findings advance understanding of tailored health-promoting messages by revealing how technological and linguistic self-relevance cues jointly, yet contingently, shape persuasion processes in AI-mediated contexts.

This study examines the use of deepfakes in self-disclosure interventions within mental health contexts. Specifically, we investigate how videos featuring self-deepfakes, celebrity deepfakes, and virtual agents disclosing mental health challenges shape affective resistance and intention to seek support, considering the moderating influence of individual baseline mental health. The findings indicate that self-deepfakes elicited greater affective resistance than celebrity deepfakes, leading to reduced help-seeking intention, whereas no significant differences were observed between self-deepfakes and virtual agent disclosures. Also, the moderation analysis showed that participants with lower baseline mental health were especially prone to heightened affective resistance toward self-disclosure videos featuring deepfake representations of themselves. Our findings indicate that artificial intelligence (AI)-generated self-deepfakes, which personalize content without affording users agency, may reverse the conventional self-referencing effect, provoking affective resistance rooted in identity threat. Since these counterproductive effects are most salient among individuals with negative self-schemas who struggle with greater mental health challenges, AI-driven technologies should be applied in health communication with caution, accompanied by tailored strategies designed to curb impulsive, emotion-driven resistance.

The media effects discipline has bequeathed scholars with an endlessly pliable set of theories helping establish the field of communication. Recent decades, however, witnessed a growing chorus of criticism pointing to theories’ anachronistic nature to suggest that ideas once known for their originality and conceptual heft overstayed their welcome. If there is intellectual disagreement, it is in fact a propitious moment to infuse the marketplace of ideas with critical data. To this end, meta- and interrupted time series-analyses of seven key media effects theories (K = 588, N = 525,670) identify a decline in effect sizes over time, while moderation analyses point to methodological and technological culprits. As it turns out, effect sizes linked with media effects theories get smaller but more statistically significant. Optimistically, this may signal that, just like humans, theories that grow long in the tooth lose some of their youthful spark but gain stability.

Mental models play an adaptive role across communication processes, including information-seeking, processing, and transmission. However, the abstract nature of mental modeling, coupled with conceptual and operational inconsistencies, has led to fault lines within communication scholarship. The present article synthesizes the current state of mental models research to advance theoretical development in two ways. First, it establishes a common conceptualization of mental models by examining their core features—concentrated nature, personal construction, internal representation of external phenomena, and influence on cognitive-behavioral responses—across different communication frameworks and theories. Second, it identifies five key moderating variables—centrality, coherence, completeness, deliberation, and rigidity—that shape how mental models function in communication processes. The article concludes with testable propositions designed to probe these moderating variables and extend the application of mental models in communication research. Overall, this work offers a synthesized perspective of mental models to better understand the fundamental processes that define human communication.

Rooted in the emotions-as-frames model (EFM), this research examines how hope, fear, and annoyance are evoked through health news headline scanning, and how these emotions influence perceptions of news and medical science institutions as well as health behavioral intentions. A sample of U.S. adults (N = 327) were assigned to one of four headline framing conditions expected to associate with different emotions (positive future frame-hope; threat frame-fear/anxiety; reversal frame-annoyance; and control-neutral) and then asked about their emotional states, trust in science and news, and health-related behavioral intentions. Overall, health news headlines generated more hope than any other emotion across all conditions, and positive future-framed headlines evoked more hope than other framed headlines. Felt hope, in turn, generated greater trust in news and science, higher expectations of medical breakthroughs and cures, and greater intention to engage in preventative health behaviors. Felt anxiety had marginal positive benefits whereas felt annoyance negatively impacted the outcomes of interest. Notably, felt emotion mediated the headline frame-outcome relationships in the positive future/hope condition. These findings offer some support for the EFM and demonstrate that scanning headlines imbued with specific emotional frames can influence important health-related outcomes through the emotions they evoke. We discuss both the theoretical and practical implication of these findings.

A meta-analysis evaluated the persuasive effects of political satire and how such effects may differ between Horatian (light-hearted) and Juvenalian (acerbic) satire types. A synthesis of 33 studies ( N = 7,469) revealed that political satire is no more persuasive than serious news, but the data find several contingencies when distinguishing between satire types. Namely, satire type was found to interact with issue involvement and format, but not ideology or perceived humor. Particularly, persuasion was enhanced by Juvenalian sketch format (compared with news format) satire, as well as Juvenalian satire for high-involvement individuals, but was inhibited by Horatian satire for those same individuals.