Folk Theories of Language and the Brain: Public Beliefs About Speech, Bilingualism, and Neural Function

Folk Theories of Language and the Brain: Public Beliefs About Speech, Bilingualism, and Neural Function | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Folk Theories of Language and the Brain: Public Beliefs About Speech, Bilingualism, and Neural Function Yixuan Liang, Colin Wilson, Alan L. Yuile This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7941468/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Public understanding of how the brain processes language remains shaped more by folk theories than by modern neuroscience. Misconceptions and oversimplification such as “language lives in the left brain,” “bilingualism causes delays,” or “AI understands language like humans” persist across education, media, and healthcare discourse. Yet scientific models, including Damasio and Damasio’s tripartite neural framework, reveal a far more nuanced picture—where language emerges from interactions between bilateral conceptual networks, left-lateralized linguistic encoders, and intermediary systems that mediate between meaning and form. This study investigates the prevalence and origins of these folk theories using a mixed-methods approach, including public surveys, qualitative interviews, and media analysis. We identify the most common public beliefs, map them against neuroscientific models, and explore how misinformation travels through education, personal experience, and popular media. Our findings reveal not only widespread misalignment between public beliefs and neuroscientific consensus, but also thematic patterns shaped by metaphor, oversimplification, and cognitive bias. We conclude by offering communication strategies for scientists, educators, and policymakers seeking to close the gap between brain science and public perception of language. Cognitive Neuroscience Folk Theories Language Processing Damasio’s tripartite neural framework Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Significance Statement Public beliefs about how the brain processes language often rely on outdated models—such as the idea that language is confined to the left hemisphere or that AI systems “understand” language like humans. This study examines the prevalence and origins of these folk theories, using surveys, interviews, and media analysis to explore how they are shaped by education, media, and interactions with large language models (LLMs). By comparing public beliefs with current neuroscientific models, we identify key gaps in understanding and propose strategies for improving science communication. The findings have implications for education, clinical practice, and public discourse around language, brain function, and emerging AI technologies. Introduction How do people outside neuroscience think about the relationship between language and the brain? Research on folk theories —intuitive models that may oversimplify or distort scientific findings—suggests that public beliefs shape attitudes toward education, health, and technology (Niedzielski & Preston, 2000; Lilienfeld et al., 2015). Within language studies, prior work has documented both the myths commonly endorsed by the public (Wagner et al., 2023) and the knowledge experts believe the public should possess (Lahecka, 2021). Popular accounts, such as Kaplan’s Women Talk More Than Men , demonstrate how such ideas circulate beyond academic settings. Yet, compared with domains like intelligence and memory, systematic evidence on how people conceptualize the link between language and the brain remains limited. Several enduring misconceptions illustrate this gap. Misconceptions such as ‘language lives in the left hemisphere’ or ‘bilingualism confuses children’ persist despite advances in cognitive neuroscience. Many of these ideas trace back to early localization studies, such as Broca’s (1861) identification of cortical speech areas, which, while historically influential, have since been replaced by network-based accounts of brain function. Nevertheless, metaphors like “rewiring the brain” or “language centers” continue to reinforce outdated views in classrooms and media (O’Connor & Joffe, 2020). Beliefs about bilingualism provide a telling example. Concerns that dual-language exposure delays children’s development have been debated for decades, and some evidence suggests that endorsement of this myth may be declining (Byers-Heinlein & Lew-Williams, 2013). Still, such misconceptions can carry real-world consequences: families may avoid bilingual exposure, misjudge recovery potential after brain injury (Elman, 2016), or resist evidence-based educational policies (Bishop, 2014). Other myths are newer: large language models (LLMs) such as ChatGPT produce fluent text that leads many users to infer human-like understanding, echoing Weizenbaum’s (1966) “Eliza effect.” Scientific theories of language provide a far more complex picture. Rather than a single “language center,” contemporary accounts emphasize distributed and interactive systems. For instance, emergentist theories argue that language is not an encapsulated module but instead arises from the interaction of general cognitive, perceptual, and social processes (Bates & MacWhinney, 1989; Tomasello, 2003). Statistical learning approaches highlight how infants and adults extract structure from linguistic input through domain-general pattern detection, showing that exposure alone can drive segmentation, categorization, and rule formation (Saffran, Aslin, & Newport, 1996; Elman, 2004). In contrast, debates over modularity emphasize whether certain aspects of language, such as syntax, are handled by specialized, encapsulated systems or instead emerge from more general neural architectures (Fodor, 1983). Within this broader landscape, we highlight Damasio and Damasio’s tripartite model (1992) as one influential example. This model, grounded in lesion studies and neuroimaging, emphasizes interactions between bilateral conceptual networks, left-hemisphere linguistic encoders, and mediation systems linking meaning to form. We do not present this framework as definitive; rather, we use it heuristically to illustrate how contemporary scientific accounts diverge from folk metaphors such as “left brain,” “language centers,” or “AI that understands.” Taken together, these tensions point to an urgent need to assess the current landscape of public belief. This study addresses three central questions: ( 1 ) What are the most prevalent folk theories about how language is processed in the brain? ( 2 ) How are these beliefs shaped or reinforced by media, education, and interactions with AI? and ( 3 ) Where do public beliefs most sharply diverge from contemporary scientific frameworks? To answer these questions, we combine large-scale surveys, qualitative interviews, and media content analysis. By mapping public beliefs against current science and examining how those beliefs are sustained, we aim to inform more effective communication strategies. These findings have implications for education, clinical practice, and policy—especially in an era when human and machine language are increasingly entangled in everyday discourse. Methods This study used a mixed-methods design combining quantitative surveys, qualitative interviews, and media content analysis to investigate public beliefs about brain–language relationships. All procedures involving human participants were approved by the Institutional Review Board at Johns Hopkins University. Informed consent was obtained from all participants. 1. Survey Study 1.1 Participants Two groups participated (N = 687): General public sample (n = 363): recruited through Qualtrics and Prolific in the United States, China, India, Spain, France, and Germany. Participants were stratified by age, educational background, and native language. Expert sample (n = 324): recruited through neuroscience and cognitive science departments at Johns Hopkins University, Harvard University, the University of Southern California, Cold Spring Harbor Laboratory, Peking University, Salamanca University, and Ludwig Maximilian University of Munich. Demographic information collected included: age, gender, country of residence, native language, and educational background (STEM vs. non-STEM). STEM status was emphasized in analyses because prior literature suggests science education moderates belief in “folk” explanations (Lilienfeld et al., 2015). Other demographic variables were also tested; nonsignificant effects (age, gender) are reported in Supplementary Table S1. Participants received $ 15 for completing the survey. 1.2 Survey Instrument The instrument included 35 multiple-choice and Likert-style items across three domains: Language localization (e.g., hemispheric specialization, Broca’s and Wernicke’s areas), Bilingualism and development, Artificial intelligence and language processing. Items were written in plain, accessible language. The complete list of items is provided in Appendix A. 1.3 Coding and Analysis Responses were coded as follows: Myth endorsement = 1: selection of a scientifically unsupported statement (e.g., “language resides entirely in the left hemisphere”). Accurate response = 0: selection of a statement consistent with current consensus (e.g., “distributed across both hemispheres”). Don’t know: coded separately, excluded from accuracy scores but retained for descriptive statistics. Where scholarly debate exists, we justified our coding decisions. For example, the item “Bilingualism enhances cognitive flexibility” was coded as accurate, in line with developmental research showing advantages of bilingual exposure (Byers-Heinlein & Lew-Williams, 2013). We acknowledge, however, that some debates remain in the literature. This approach was chosen to avoid collapsing “debated but widely supported” views into myths. Data analysis proceeded in several steps. Descriptive statistics were first calculated to estimate endorsement rates for each belief. Group differences between the general public and experts, as well as by STEM background and national subgroup, were assessed using chi-square (χ²) tests. Associations between continuous variables, such as belief confidence and media exposure, were examined with Pearson correlations. To quantify the strength of group differences in myth rejection, odds ratios were computed. Exploratory factor analysis (principal axis factoring, Varimax rotation) on all 35 survey items to test whether misconceptions clustered into coherent domains. The factor analysis served both as a validity check (i.e., did our survey tap the three intended domains?) and as a way to test whether folk theories naturally clustered in public cognition. 2. Qualitative Interviews 2.1 Participants and Procedure A stratified random subsample of 60 participants (30 general public, 30 experts) completed semi-structured interviews via Zoom (30–45 minutes). Interviews were audio-recorded, transcribed verbatim, and anonymized. 2.2 Interview Guide The guide covered four domains: Brain–language relationships (e.g., “What happens in the brain when someone speaks?”), Bilingualism (e.g., “How do you think being bilingual affects the brain?”), Artificial intelligence (e.g., “Do you think AI systems understand language like humans?”), Belief sources and confidence (e.g., “Where did you learn this?”). Follow-up questions probed contradictions (e.g., “Earlier you said language is in the left hemisphere, but now you mentioned both sides—can you explain?”). This ensured that mixed or inconsistent beliefs were captured rather than smoothed over. Full question wording is provided in Appendix B. 2.3 Coding and Reliability Two independent coders analyzed transcripts using NVivo. Codes were developed deductively (e.g., “localization myth,” “AI-human equivalence”) and inductively (e.g., emergent metaphors such as “rewiring”). Contradictory statements were coded under multiple categories and flagged as “mixed responses.” Inter-rater reliability was strong (Cohen’s κ = 0.81). Discrepancies were resolved through discussion. Unexpected themes (e.g., AI having “emotions”) were retained for qualitative description. 3. Media Content Analysis 3.1 Sampling Strategy We analyzed 150 English-language media items published between 2015–2023, identified via YouTube, TED, Google News, and science blogs. Search terms included “language and brain,” “bilingualism,” “aphasia,” and “AI language.” Inclusion criteria: >500 words or > 5 minutes in length, targeted at a lay audience. To contextualize influence, we recorded popularity metrics (views, shares, subscriber counts). However, we deliberately treated each source equally in coding to avoid biasing analyses toward high-traffic items. Instead, popularity metrics are reported descriptively. 3.2 Coding Scheme Each media source was systematically coded along several dimensions. We assessed the type of model used to describe brain–language relationships (modular vs. distributed), the presence of metaphors (e.g., “rewiring the brain,” “left-brain logic”), the degree of scientific accuracy (alignment with peer-reviewed research), and the extent of anthropomorphic framing of AI (e.g., use of verbs such as “understands” or “thinks”). In addition, we coded credibility markers, including author credentials (neuroscientist, journalist, layperson), venue type (e.g., TED vs. personal blog), and whether peer-reviewed research was cited. To quantify lexical patterns across the corpus, we conducted lexicometric analyses—such as frequency counts and collocation analysis—using Python (v3.11). Results We analyzed survey data from 687 participants, comprising 363 individuals from the general public and 324 with formal training in neuroscience or cognitive science. Our investigation focused on three domains of public belief: language lateralization, bilingualism, and artificial intelligence (AI). The results revealed significant discrepancies between public beliefs and current neuroscientific models. Language Lateralization In the domain of language lateralization, 62% of general public respondents (225 out of 363) endorsed the belief that language resides exclusively in the left hemisphere. In contrast, only 8% of expert respondents (26 out of 324) endorsed this view. The difference was statistically significant, χ²(1, N = 687) = 210.4, p < .001, with a large effect size (Cramer’s V = .55). Only 13% of all participants referenced bilateral conceptual networks—an essential element in Damasio and Damasio’s (1992) tripartite model. Additionally, we found that education level was positively associated with belief accuracy, with a correlation coefficient of r = 0.36, 95% CI [0.27, 0.44], p = 0.004ᵇ. Figure 1. Endorsement of Key Folk Theories by Group Bar graph comparing endorsement rates of three major folk theories—language lateralization, bilingualism delay, and AI-human language equivalence—between general public respondents (n = 363) and neuroscience experts (n = 324). The general public showed significantly higher belief in all three misconceptions. Bars represent percentage endorsement. Bilingualism Regarding bilingualism, 41% of the general public believed that exposing children to two languages causes developmental delays, while only 29% selected the accurate statement that bilingualism enhances cognitive flexibility. By contrast, only 6% of experts (19/324) endorsed the delay myth, while 78% endorsed the flexibility advantage. This group difference was statistically significant, χ²(1, N = 687) = 104.7, p < .001, Cramer’s V = .39. Participants from Germany and France were significantly more likely to choose the scientifically supported view, in contrast to those from the United States and India, χ²(3, N = 363) = 18.9, p = 0.0021ᶜ. These regional differences likely reflect the influence of multilingual education policies in shaping belief accuracy. Public Perception of AI and Language Public perception of AI and language also revealed widespread misunderstanding. A total of 58% of general public respondents (211 out of 363) agreed with the statement that “AI understands language like humans.” In contrast, only 7% of experts (24 out of 324) endorsed this statement. The group difference was highly significant, χ²(1, N = 687) = 192.6, p < .001, Cramer’s V = .53. This belief was significantly more prevalent among individuals who reported frequent use of large language models such as ChatGPT, χ²(1, N = 363) = 45.2, p = 0.00091ᵈ. Factor Analysis of Misconceptions A factor analysis of survey responses was conducted across all 35 survey questions, rather than only those related to AI. Items clustered into three dominant misconception categories: “Localization Myths” (e.g., “Broca’s area handles grammar alone”), “Developmental Determinism” (e.g., “Children are blank slates”), and “AI-Human Equivalence” (e.g., “ChatGPT thinks like a person”). This analysis was used to validate that the most common misconceptions identified empirically map onto the three conceptual domains that guided survey design. Figure 2. Factor Loadings of Folk Theories on Three Key Misconception Clusters Heatmap showing the results of exploratory factor analysis on survey responses. Items loaded onto three latent misconception clusters: Localization Myths, Developmental Determinism, and AI-Human Equivalence. Strong loadings (≥ 0.70) are indicated in red, weaker in blue. Factor analysis was conducted using principal axis factoring with Varimax rotation (n = 687). Belief Confidence and Media Exposure To complement the quantitative findings, we conducted qualitative interviews with a randomized subset of 60 participants—30 from the general public and 30 from the expert cohort. Among the general public, 63% reported media sources, including TED Talks and YouTube, as their primary source of neuroscience knowledge. Another 22% attributed their beliefs to school-based education, frequently citing textbook diagrams of “language centers,” while the remaining 15% drew on personal anecdotes, such as a child’s delayed speech. Interview data also highlighted the influence of metaphor in shaping public understanding. Phrases like “rewiring the brain” were cited in 47% of interviews, while 32% referenced the notion of “left-brain logic.” Notably, belief confidence was positively associated with frequent exposure to media narratives, with a correlation of r = 0.41, 95% CI [0.26, 0.54], p = 0.009ᵉ, even among participants who held demonstrably inaccurate views. Media exposure was measured using a survey item asking participants, “How often do you engage with media content (e.g., TED Talks, YouTube videos, blogs, or articles) about the brain and language?” (1 = never, 5 = very often). Figure 3. Reported Sources of Belief About Language and the Brain Pie chart showing self-reported sources of belief among general public interview participants (n = 30). Most cited media and TED Talks (63%), followed by formal education (22%) and personal experience (15%). Responses were coded from open-ended qualitative interviews. Media Content Analysis A media content analysis of 150 public-facing texts—including articles, TED Talks, and educational videos—offered further insights into the origin of these beliefs. Only 18% of sources accurately reflected the tripartite model of language processing. In contrast, 67% relied on outdated or oversimplified terminology, such as "language module" or "speech center." Technical terms like “bilateral conceptual networks” appeared in just 12% of the media corpus, while metaphorical analogies—such as likening the brain to a computer—were found in 88% of cases. Lexicometric analysis supported these patterns: “left-brain” appeared nearly four times more frequently than “bilateral.” In AI-focused content, 72% of sources used anthropomorphic verbs such as “understands” and “learns,” contributing to the widespread misconception that AI possesses human-like cognitive capacities. Figure 4. Frequency of Dominant Metaphors in Media Coverage Bar graph depicting term frequency (n = 150 media samples) for technical and metaphorical language describing brain-language processes and artificial intelligence. Metaphors such as “left-brain,” “rewiring,” and “understands” appeared far more frequently than scientific terms like “mediation” or “bilateral.” Individual Differences Across Demographics Cross-demographic and cross-cultural analyses revealed further disparities. Participants with STEM backgrounds were 2.3 times more likely to reject localization myths than those without, with an odds ratio of 2.3, 95% CI [1.7, 3.1], p = 0.0017ᶠ. Belief in AI-human equivalence was highest among participants from the United States and India, where 65% endorsed this view, compared to only 51% of European participants. This cross-national difference was statistically significant, χ²(2, N = 363) = 14.3, p = 0.0064ᵍ. Meanwhile, participants from Germany and France were more likely to select scientifically accurate statements about bilingualism, suggesting a potential link between public policy and belief accuracy. Age and gender were also collected but did not show significant associations with belief accuracy. Full descriptive and inferential statistics for these variables are provided in Supplementary Table S1. Figure 5. Cross-Cultural Variation in Misconception Endorsement Grouped bar graph displaying endorsement rates of three misconception categories across six countries: United States, France, Germany, China, India, and Spain. Cross-cultural variation was most pronounced in beliefs about bilingualism and AI equivalence. Total survey sample: n = 363. Qualitative Interview Themes Finally, we assessed the degree to which public beliefs align with contemporary neuroscientific frameworks. Overall, 89% of respondents referenced the left hemisphere as the exclusive site of language, while only 6% mentioned intermediary mediation structures, and just 2% acknowledged any role for subcortical regions. In contrast, expert participants frequently described language as a dynamic, distributed process involving the interaction of syntax, semantics, and broader cognitive functions. The general public, however, continued to rely on static, modular metaphors that do not reflect the current state of neuroscience. Figure 6. Public vs. Expert Emphasis on Brain-Language Concepts Horizontal bar graph comparing the relative emphasis on five types of language-brain concepts in open-ended responses from the general public (n = 30) and expert participants (n = 30). Experts were more likely to reference distributed systems and mediation structures, while public responses favored modular or static explanations. Discussion Our study indicates a persistent and multi-layered disconnect between contemporary neuroscience and public beliefs about how the brain processes language. Across survey, interview, and media data, we found that folk theories consistently cluster around three dominant misconceptions: localization myths, developmental determinism, and AI anthropomorphism. These clusters represent not only epistemic simplifications but also deeply entrenched conceptual frameworks shaped by language, metaphor, and cultural narrative. They align with prior work showing that folk theories in other cognitive domains—such as memory (as “storage”), attention (as “spotlight”), or intelligence (as fixed “capacity”)—rely on similarly reductive heuristics (e.g., Furnham, 1996; Roediger, 1980). Our findings therefore extend the broader folk theory literature by illustrating how language, like other cognitive functions, is filtered through familiar but misleading conceptual shortcuts. The most prevalent folk theory—localization of language in the left hemisphere—stands in stark contrast to modern models of distributed and bilateral neural processing (Damasio & Damasio, 1992). While localization theories have historical significance (e.g., Broca, 1861), our findings suggest they have been overgeneralized in the public mind. Only a small minority of participants mentioned mediation structures or bilateral conceptual networks—key components in contemporary models. These omissions reflect a widespread reliance on spatial and modular metaphors (“language center,” “speech area”) that obscure the dynamic and interactive nature of language processing. While it is true that this myth may not carry the same societal weight as misconceptions about bilingualism or AI, it still matters: if the public believes language resides exclusively in one spot, they may misunderstand the consequences of brain injury, underestimate recovery potential, or support oversimplified clinical interventions. In this sense, even a “low-stakes” myth about lateralization shapes how people think about health, disability, and rehabilitation. Importantly, acknowledging alternative models (e.g., Friederici, Poeppel, Pylkkänen) reinforces that “not Damasio” does not mean “folk theory”; rather, the folk version is the outdated idea of a single left-hemisphere “language organ.” A second cluster of misconceptions centers on developmental determinism—the belief that language acquisition is a passive process governed by immutable developmental stages or “blank slate” brains. This belief continues to inform public opinion on bilingualism, with 41% of participants believing it causes delays. This is in direct opposition to empirical research showing cognitive and social advantages associated with dual-language exposure (Byers-Heinlein & Lew-Williams, 2013). The persistence of this myth suggests that both educational materials and clinical messaging have not sufficiently integrated findings from neuroscience on plasticity, sensitive periods, and variability in language development. Unlike the lateralization myth, this misconception has immediate and tangible social consequences, influencing parenting choices, educational practice, and immigration debates. As with folk models of intelligence that emphasize innate ability over learning, these deterministic views simplify complexity but carry consequences for how children are raised, taught, and assessed. The third and most contemporary cluster involves AI anthropomorphism. Over half of participants agreed that AI systems “understand language like humans,” with higher endorsement among those frequently interacting with LLMs. This belief is both intuitive and misleading. The fluency of models like ChatGPT triggers anthropomorphic biases, leading users to ascribe cognitive and emotional capacities where none exist (Bender & Koller, 2020; Weizenbaum, 1966). Our media analysis showed that 72% of AI-focused content used human-like verbs such as “understands” or “thinks,” reinforcing the illusion of consciousness. This linguistic framing matters—not just because it distorts public understanding, but because it may shape attitudes toward technology, education policy, and even mental health interventions. Here, as in folk theories of memory and intelligence, metaphorical reasoning produces powerful but inaccurate intuitions that shape real-world decision-making. A key theme emerging from both interviews and media content is the power of metaphor in shaping cognitive models. Phrases like “rewiring the brain” and “language centers” function as communicative shortcuts, but they also encode simplified and sometimes inaccurate representations. Only 18% of media sources accurately described distributed models of brain-language interaction, while nearly 90% leaned on familiar but outdated metaphors. The “so what” of these results is clear: metaphors are not neutral—they shape folk theories, and folk theories in turn shape behavior, from whether parents raise children bilingually to how societies interpret the promises and risks of AI. What connects our three focal myths is precisely their metaphorical foundation: folk phrenology in the case of lateralization, biological clocks and blank slates in the case of development, and anthropomorphic personification in the case of AI. Each myth survives because it is grounded in metaphors that are intuitively compelling, easy to communicate, and widely reinforced in culture. To move forward, we need better metaphors: for example, describing Broca’s area as a “hub in a distributed network” rather than a “center,” or AI as a “mirror of statistical patterns” rather than a “learner.” Such metaphors retain accessibility while reducing distortion. The observed demographic differences—particularly between STEM-trained participants and those without formal science education—point to promising avenues for intervention. Participants with STEM backgrounds were significantly more likely to reject localization myths and AI anthropomorphism, suggesting that science education serves as a protective factor against misinformation. Cross-cultural differences, particularly higher belief accuracy in multilingual countries with policy-driven education (e.g., Germany, France), further reinforce the importance of systemic educational strategies. Additionally, media literacy training could help the public better evaluate claims presented in news articles, TED Talks, and AI-generated content. Overall, our findings underscore the importance of targeted science communication strategies. Addressing public misconceptions about language and the brain will require a combination of accurate metaphor use, integration of current neuroscience into curricula, and critical engagement with how AI technologies are described in public discourse. As generative language technologies become more visible and persuasive, the risk of reinforcing intuitive but incorrect folk theories grows. Bridging the divide between expert models and public belief is therefore not just an educational challenge—it is a necessary foundation for evidence-based policy, ethical technology design, and a more cognitively informed society. Declarations Acknowledgments We thank the participants from the general public and neuroscience communities for generously sharing their time and insights. We are also grateful to the faculty and staff at Johns Hopkins University for facilitating recruitment. This research was supported by the Neuroscience and Cognitive Science Departments at Johns Hopkins University. The authors affirm that all research involving human subjects was conducted in accordance with institutional ethics guidelines. References Bates E, MacWhinney B (1989) Functionalist approaches to grammar. In: MacWhinney B, Bates E (eds) The crosslinguistic study of sentence processing. Cambridge University Press, pp 3–73 Bender EM, Koller A (2020) Climbing towards NLU: On meaning, form, and understanding in the age of data. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (pp. 5185–5198). Association for Computational Linguistics. https://doi.org/10.18653/v1/2020.acl-main.463 Bishop DVM (2014) Ten questions about terminology for children with unexplained language problems. Int J Lang Communication Disorders 49(4):381–415. https://doi.org/10.1111/1460-6984.12101 Broca P (1861) Remarks on the seat of the faculty of articulated language, following an observation of aphemia (loss of speech). Bulletins de la Société Anatomique de Paris 6:330–357 Byers-Heinlein K, Lew-Williams C (2013) Bilingualism in the early years: What the science says. Learn Landscapes 7(1):95–112. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6168212/ Damasio AR, Damasio H (1992) Brain and language. Sci Am 267(3):88–95. https://doi.org/10.1038/scientificamerican0992-88 Elman JL (2004) An alternative view of the mental lexicon. Trends Cogn Sci 8(7):301–306. https://doi.org/10.1016/j.tics.2004.05.003 Elman JL (2016) Language, plasticity, and learning: Challenges of the new millennium. In Proceedings of the 40th Annual Boston University Conference on Language Development Fodor JA (1983) The modularity of mind. MIT Press Friederici AD (2011) The brain basis of language processing: From structure to function. Physiol Rev 91(4):1357–1392. https://doi.org/10.1152/physrev.00006.2011 Furnham A (1996) Lay theories: Everyday understanding of problems in the social sciences. Hum Relat 49(2):157–173 Kuhl PK (2010) Brain mechanisms in early language acquisition. Neuron 67(5):713–727. https://doi.org/10.1016/j.neuron.2010.08.038 Kaplan J (2016) Women talk more than men … and other myths about language explained. Cambridge University Press Lahecka T (2021) Folk linguistics in applied contexts: Teachers’ beliefs about language learning. J Lang Educ 7(1):20–34 Lilienfeld SO, Lynn SJ, Ruscio J, Beyerstein BL (2015) 50 great myths of popular psychology: Shattering widespread misconceptions about human behavior. Wiley-Blackwell Mayberry RI (2023) Sensitive periods in language development: Implications for education and clinical practice. Oxford University Press Niedzielski N, Preston D (2000) Folk linguistics. Mouton de Gruyter O’Connor C, Joffe H (2020) How has neuroscience affected lay understandings of personhood? A review of the evidence. Public Underst Sci 29(1):88–102. https://journals.sagepub.com/doi/full/10.1177/0963662513476812 Poeppel D (2014) The neuroanatomic and neurophysiological infrastructure for speech and language. Curr Opin Neurobiol 28:142–149. https://doi.org/10.1016/j.conb.2014.07.005 Pylkkänen L (2019) The neural basis of combinatory syntax and semantics. Science 366(6461):62–66. https://doi.org/10.1126/science.aax0050 Roediger HL (1980) Memory metaphors in cognitive psychology. Mem Cognit 8(3):231–246. https://doi.org/10.3758/BF03213419 Saffran JR, Aslin RN, Newport EL (1996) Statistical learning by 8-month-old infants. Science 274(5294):1926–1928. https://doi.org/10.1126/science.274.5294.1926 Tomasello M (2003) Constructing a language: A usage-based theory of language acquisition. Harvard University Press Wagner K, Clopper CG, Pate JK (2023) Lay beliefs about language learning and bilingualism. Appl Psycholinguist 44(2):241–262 Weizenbaum J (1966) ELIZA—A computer program for the study of natural language communication between man and machine. Commun ACM 9(1):36–45. https://doi.org/10.1145/365153.365168 Tables Table 1. Descriptive Summary Statistics for Key Beliefs about Language and the Brain Variable N Mean Min Max SD Left hemisphere = language 363 0.62 0 1 0.49 Bilingualism delays speech 363 0.41 0 1 0.48 AI understands like humans 363 0.58 0 1 0.49 Belief accuracy (STEM group) 324 0.83 0.3 1 0.27 Belief accuracy (non-STEM group) 363 0.51 0.15 0.95 0.34 The table presents descriptive statistics for belief endorsement across five key statements related to language processing and cognition. Mean values represent the proportion of participants endorsing each belief (e.g., selecting “agree” or equivalent response options). “Left hemisphere = language” reflects belief in strict lateralization; “Bilingualism delays speech” reflects the common misconception that early dual-language exposure hinders development; “AI understands like humans” gauges anthropomorphic beliefs about language models. Accuracy scores for STEM vs. non-STEM groups reflect composite scores based on alignment with current neuroscientific models. SD = standard deviation. Table 2. Summary of Statistical Analyses Conducted Across Study Designs Purpose Statistical Method Sample Size (N) Key Outputs Compare belief accuracy by education level Pearson correlation (r) 687 r = 0.36, 95% CI [0.27, 0.44], p = 0.004 Compare public vs. expert belief rates Chi-square test 687 χ²(1) = 210.4, p = 0.00012 Compare AI beliefs by LLM exposure frequency Chi-square test 363 χ²(1) = 45.2, p = 0.00091 Examine belief confidence vs. media exposure Pearson correlation (r) 60 r = 0.41, 95% CI [0.26, 0.54], p = 0.009 Compare bilingualism beliefs by country Chi-square test 363 χ²(3) = 18.9, p = 0.0021 Compare AI beliefs by region (U.S./Japan vs. Europe) Chi-square test 363 χ²(2) = 14.3, p = 0.0064 Compare myth rejection rates by STEM background Odds ratio, chi-square test 687 OR = 2.3, 95% CI [1.7, 3.1], p = 0.0017 The table summarizes all major statistical analyses performed in the study, categorized by design type. It includes the structure of the data, statistical test employed, sample sizes, and corresponding results. Chi-square and correlation analyses compared belief endorsement across groups, while exploratory factor analysis revealed latent misconception clusters. Cross-cultural and between-group comparisons further evaluated demographic effects. p-values and confidence intervals are reported where applicable to characterize the significance and uncertainty of observed effects. Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7941468","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":534555363,"identity":"5b7a35c7-6c0d-4726-a155-d703a0ddd20e","order_by":0,"name":"Yixuan Liang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAyElEQVRIiWNgGAWjYDACdsYGIPlPjp/hDJBmI0YLM1jLAWPJBuK1gMkDiQYHeIjUYnCYufFzwa87CcYHzx5g+FB2mBgtjM3SM/ue5ZkdOJfAOOMccVoapHl7mIvNDpwxYOZtI9KW30AtiZsbgFr+EqmlTZrnx+HEDQxALYzEaJEEarHmbUgzlgA67GDPuXTCWviOtz++zfPHRo5/xhnDBz/KrAlrUTgAJBjbgITEAYYDhNUDgXwDiPwDxPwNRGkYBaNgFIyCEQgAsnJDFR3SU/gAAAAASUVORK5CYII=","orcid":"https://orcid.org/0009-0005-2857-1619","institution":"Johns Hopkins University","correspondingAuthor":true,"prefix":"","firstName":"Yixuan","middleName":"","lastName":"Liang","suffix":""},{"id":534555364,"identity":"574b1d2d-5bfb-4ceb-ab16-a6bd7ea47084","order_by":1,"name":"Colin Wilson","email":"","orcid":"https://orcid.org/0009-0005-3059-1133","institution":"Charité – Universitätsmedizin Berlin, Berlin, Germany","correspondingAuthor":false,"prefix":"","firstName":"Colin","middleName":"","lastName":"Wilson","suffix":""},{"id":534555365,"identity":"679a6e3c-c9de-4d7d-b6f0-78228da30341","order_by":2,"name":"Alan L. Yuile","email":"","orcid":"","institution":"Johns Hopkins University","correspondingAuthor":false,"prefix":"","firstName":"Alan","middleName":"L.","lastName":"Yuile","suffix":""}],"badges":[],"createdAt":"2025-10-25 13:18:44","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-7941468/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7941468/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":94583517,"identity":"983723e0-89af-4164-94a2-ecdb38fd3094","added_by":"auto","created_at":"2025-10-28 18:14:06","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":46602,"visible":true,"origin":"","legend":"","description":"","filename":"FolkTheoriesofLanguageandtheBrainREVISED.docx","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/c5fb3a3195e44c8c3b8c96b9.docx"},{"id":94584174,"identity":"a376cc86-005c-41f1-ba07-18b691d66127","added_by":"auto","created_at":"2025-10-28 18:14:58","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":342,"visible":true,"origin":"","legend":"","description":"","filename":"rs7941468.json","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/77efdd7d0a2b206df217ec8c.json"},{"id":94584565,"identity":"754faf91-fe1c-41e9-bac4-b69de60dca4c","added_by":"auto","created_at":"2025-10-28 18:15:23","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":80301,"visible":true,"origin":"","legend":"","description":"","filename":"rs79414680enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/e7bd8b68270c25e11890c2fe.xml"},{"id":94584207,"identity":"752587ea-c7d5-4433-b3d3-23c4a5e08e24","added_by":"auto","created_at":"2025-10-28 18:15:00","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":74224,"visible":true,"origin":"","legend":"","description":"","filename":"rs79414680structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/f8040b4b40962e457714df24.xml"},{"id":94584150,"identity":"f50bd0d8-6ca1-4534-90fb-ac28ea607b41","added_by":"auto","created_at":"2025-10-28 18:14:54","extension":"html","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":89310,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/8fb2645afbf0722dca9a761e.html"},{"id":94583677,"identity":"8a6f8b16-b88e-430c-a1c6-73a7ccd197a1","added_by":"auto","created_at":"2025-10-28 18:14:22","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":157325,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eEndorsement of Key Folk Theories by Group\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eBar graph comparing endorsement rates of three major folk theories—language lateralization, bilingualism delay, and AI-human language equivalence—between general public respondents (n = 363) and neuroscience experts (n = 324). The general public showed significantly higher belief in all three misconceptions. Bars represent percentage endorsement.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/ca830792cf7453a42e0f3295.png"},{"id":94583577,"identity":"3635b1be-7165-46aa-95f9-94a621d79ce8","added_by":"auto","created_at":"2025-10-28 18:14:11","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":271029,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eFactor Loadings of Folk Theories on Three Key Misconception Clusters\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eHeatmap showing the results of exploratory factor analysis on survey responses. Items loaded onto three latent misconception clusters: Localization Myths, Developmental Determinism, and AI-Human Equivalence. Strong loadings (≥0.70) are indicated in red, weaker in blue. Factor analysis was conducted using principal axis factoring with Varimax rotation (n = 687).\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/bb9b86aa72b006821dd185ba.png"},{"id":94584599,"identity":"0760afb6-017e-4bbf-a5e2-9b3d7ae8a5ce","added_by":"auto","created_at":"2025-10-28 18:15:25","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":207960,"visible":true,"origin":"","legend":"\u003cp\u003eFrequency of Dominant Metaphors in Media Coverage\u003c/p\u003e\n\u003cp\u003eBar graph depicting term frequency (n = 150 media samples) for technical and metaphorical language describing brain-language processes and artificial intelligence. Metaphors such as “left-brain,” “rewiring,” and “understands” appeared far more frequently than scientific terms like “mediation” or “bilateral.”\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/51ef77b37ade4bbab7213132.png"},{"id":94584173,"identity":"6350ffaa-51a7-4913-a81a-bca80510bd99","added_by":"auto","created_at":"2025-10-28 18:14:58","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":155430,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eReported Sources of Belief About Language and the Brain\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003ePie chart showing self-reported sources of belief among general public interview participants (n = 30). Most cited media and TED Talks (63%), followed by formal education (22%) and personal experience (15%). Responses were coded from open-ended qualitative interviews.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/773ba7e41f83e14281b81cb2.png"},{"id":94584294,"identity":"84f14ae7-2eaa-4ba5-b3f1-b8456340e1b8","added_by":"auto","created_at":"2025-10-28 18:15:07","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":91123,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eCross-Cultural Variation in Misconception Endorsement\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eGrouped bar graph displaying endorsement rates of three misconception categories across six countries: United States, France, Germany, China, India, and Spain. Cross-cultural variation was most pronounced in beliefs about bilingualism and AI equivalence. Total survey sample: n = 363.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/b30aa8d7cc7cf143f057893e.png"},{"id":94583625,"identity":"daafb09d-c7b0-4d1a-9a4d-cf90e151c2f6","added_by":"auto","created_at":"2025-10-28 18:14:17","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":174957,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003ePublic vs. Expert Emphasis on Brain-Language Concepts\u003c/em\u003e\u003cbr\u003e\nHorizontal bar graph comparing the relative emphasis on five types of language-brain concepts in open-ended responses from the general public (n = 30) and expert participants (n = 30). Experts were more likely to reference distributed systems and mediation structures, while public responses favored modular or static explanations.\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/3c95bf23685d2c32d68f77c8.png"},{"id":95654397,"identity":"4be03f9e-54ea-4a1a-aa26-aa43ab5e5b3a","added_by":"auto","created_at":"2025-11-11 16:11:36","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1280362,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7941468/v1/4101bb16-9d49-4e05-b5ca-1528d46704e1.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eFolk Theories of Language and the Brain: Public Beliefs About Speech, Bilingualism, and Neural Function\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"Significance Statement","content":"\u003cp\u003ePublic beliefs about how the brain processes language often rely on outdated models\u0026mdash;such as the idea that language is confined to the left hemisphere or that AI systems \u0026ldquo;understand\u0026rdquo; language like humans. This study examines the prevalence and origins of these folk theories, using surveys, interviews, and media analysis to explore how they are shaped by education, media, and interactions with large language models (LLMs). By comparing public beliefs with current neuroscientific models, we identify key gaps in understanding and propose strategies for improving science communication. The findings have implications for education, clinical practice, and public discourse around language, brain function, and emerging AI technologies.\u003c/p\u003e"},{"header":"Introduction","content":"\u003cp\u003eHow do people outside neuroscience think about the relationship between language and the brain? Research on \u003cem\u003efolk theories\u003c/em\u003e\u0026mdash;intuitive models that may oversimplify or distort scientific findings\u0026mdash;suggests that public beliefs shape attitudes toward education, health, and technology (Niedzielski \u0026amp; Preston, 2000; Lilienfeld et al., 2015). Within language studies, prior work has documented both the myths commonly endorsed by the public (Wagner et al., 2023) and the knowledge experts believe the public should possess (Lahecka, 2021). Popular accounts, such as Kaplan\u0026rsquo;s \u003cem\u003eWomen Talk More Than Men\u003c/em\u003e, demonstrate how such ideas circulate beyond academic settings. Yet, compared with domains like intelligence and memory, systematic evidence on how people conceptualize the link between language and the brain remains limited.\u003c/p\u003e\u003cp\u003eSeveral enduring misconceptions illustrate this gap. Misconceptions such as \u0026lsquo;language lives in the left hemisphere\u0026rsquo; or \u0026lsquo;bilingualism confuses children\u0026rsquo; persist despite advances in cognitive neuroscience. Many of these ideas trace back to early localization studies, such as Broca\u0026rsquo;s (1861) identification of cortical speech areas, which, while historically influential, have since been replaced by network-based accounts of brain function. Nevertheless, metaphors like \u0026ldquo;rewiring the brain\u0026rdquo; or \u0026ldquo;language centers\u0026rdquo; continue to reinforce outdated views in classrooms and media (O\u0026rsquo;Connor \u0026amp; Joffe, 2020).\u003c/p\u003e\u003cp\u003eBeliefs about bilingualism provide a telling example. Concerns that dual-language exposure delays children\u0026rsquo;s development have been debated for decades, and some evidence suggests that endorsement of this myth may be declining (Byers-Heinlein \u0026amp; Lew-Williams, 2013). Still, such misconceptions can carry real-world consequences: families may avoid bilingual exposure, misjudge recovery potential after brain injury (Elman, 2016), or resist evidence-based educational policies (Bishop, 2014). Other myths are newer: large language models (LLMs) such as ChatGPT produce fluent text that leads many users to infer human-like understanding, echoing Weizenbaum\u0026rsquo;s (1966) \u0026ldquo;Eliza effect.\u0026rdquo;\u003c/p\u003e\u003cp\u003eScientific theories of language provide a far more complex picture. Rather than a single \u0026ldquo;language center,\u0026rdquo; contemporary accounts emphasize distributed and interactive systems. For instance, \u003cem\u003eemergentist theories\u003c/em\u003e argue that language is not an encapsulated module but instead arises from the interaction of general cognitive, perceptual, and social processes (Bates \u0026amp; MacWhinney, 1989; Tomasello, 2003). \u003cem\u003eStatistical learning approaches\u003c/em\u003e highlight how infants and adults extract structure from linguistic input through domain-general pattern detection, showing that exposure alone can drive segmentation, categorization, and rule formation (Saffran, Aslin, \u0026amp; Newport, 1996; Elman, 2004). In contrast, debates over \u003cem\u003emodularity\u003c/em\u003e emphasize whether certain aspects of language, such as syntax, are handled by specialized, encapsulated systems or instead emerge from more general neural architectures (Fodor, 1983). Within this broader landscape, we highlight Damasio and Damasio\u0026rsquo;s tripartite model (1992) as one influential example. This model, grounded in lesion studies and neuroimaging, emphasizes interactions between bilateral conceptual networks, left-hemisphere linguistic encoders, and mediation systems linking meaning to form. We do not present this framework as definitive; rather, we use it heuristically to illustrate how contemporary scientific accounts diverge from folk metaphors such as \u0026ldquo;left brain,\u0026rdquo; \u0026ldquo;language centers,\u0026rdquo; or \u0026ldquo;AI that understands.\u0026rdquo;\u003c/p\u003e\u003cp\u003eTaken together, these tensions point to an urgent need to assess the current landscape of public belief. This study addresses three central questions: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) What are the most prevalent folk theories about how language is processed in the brain? (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) How are these beliefs shaped or reinforced by media, education, and interactions with AI? and (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) Where do public beliefs most sharply diverge from contemporary scientific frameworks? To answer these questions, we combine large-scale surveys, qualitative interviews, and media content analysis. By mapping public beliefs against current science and examining how those beliefs are sustained, we aim to inform more effective communication strategies. These findings have implications for education, clinical practice, and policy\u0026mdash;especially in an era when human and machine language are increasingly entangled in everyday discourse.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eThis study used a mixed-methods design combining quantitative surveys, qualitative interviews, and media content analysis to investigate public beliefs about brain\u0026ndash;language relationships. All procedures involving human participants were approved by the Institutional Review Board at Johns Hopkins University. Informed consent was obtained from all participants.\u003c/p\u003e\u003cp\u003e1. Survey Study\u003c/p\u003e\u003cp\u003e1.1 Participants\u003c/p\u003e\u003cp\u003eTwo groups participated (N\u0026thinsp;=\u0026thinsp;687):\u003c/p\u003e\u003cp\u003eGeneral public sample (n\u0026thinsp;=\u0026thinsp;363): recruited through Qualtrics and Prolific in the United States, China, India, Spain, France, and Germany. Participants were stratified by age, educational background, and native language.\u003c/p\u003e\u003cp\u003eExpert sample (n\u0026thinsp;=\u0026thinsp;324): recruited through neuroscience and cognitive science departments at Johns Hopkins University, Harvard University, the University of Southern California, Cold Spring Harbor Laboratory, Peking University, Salamanca University, and Ludwig Maximilian University of Munich.\u003c/p\u003e\u003cp\u003eDemographic information collected included: age, gender, country of residence, native language, and educational background (STEM vs. non-STEM). STEM status was emphasized in analyses because prior literature suggests science education moderates belief in \u0026ldquo;folk\u0026rdquo; explanations (Lilienfeld et al., 2015). Other demographic variables were also tested; nonsignificant effects (age, gender) are reported in Supplementary Table S1.\u003c/p\u003e\u003cp\u003eParticipants received \u003cspan\u003e$\u003c/span\u003e15 for completing the survey.\u003c/p\u003e\u003cp\u003e1.2 Survey Instrument\u003c/p\u003e\u003cp\u003eThe instrument included 35 multiple-choice and Likert-style items across three domains:\u003c/p\u003e\u003cp\u003eLanguage localization (e.g., hemispheric specialization, Broca\u0026rsquo;s and Wernicke\u0026rsquo;s areas),\u003c/p\u003e\u003cp\u003eBilingualism and development,\u003c/p\u003e\u003cp\u003eArtificial intelligence and language processing.\u003c/p\u003e\u003cp\u003eItems were written in plain, accessible language. The complete list of items is provided in Appendix A.\u003c/p\u003e\u003cp\u003e1.3 Coding and Analysis\u003c/p\u003e\u003cp\u003eResponses were coded as follows:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eMyth endorsement\u0026thinsp;=\u0026thinsp;1: selection of a scientifically unsupported statement (e.g., \u0026ldquo;language resides entirely in the left hemisphere\u0026rdquo;).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eAccurate response\u0026thinsp;=\u0026thinsp;0: selection of a statement consistent with current consensus (e.g., \u0026ldquo;distributed across both hemispheres\u0026rdquo;).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eDon\u0026rsquo;t know: coded separately, excluded from accuracy scores but retained for descriptive statistics.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eWhere scholarly debate exists, we justified our coding decisions. For example, the item \u0026ldquo;Bilingualism enhances cognitive flexibility\u0026rdquo; was coded as accurate, in line with developmental research showing advantages of bilingual exposure (Byers-Heinlein \u0026amp; Lew-Williams, 2013). We acknowledge, however, that some debates remain in the literature. This approach was chosen to avoid collapsing \u0026ldquo;debated but widely supported\u0026rdquo; views into myths.\u003c/p\u003e\u003cp\u003eData analysis proceeded in several steps. Descriptive statistics were first calculated to estimate endorsement rates for each belief. Group differences between the general public and experts, as well as by STEM background and national subgroup, were assessed using chi-square (χ\u0026sup2;) tests. Associations between continuous variables, such as belief confidence and media exposure, were examined with Pearson correlations. To quantify the strength of group differences in myth rejection, odds ratios were computed.\u003c/p\u003e\u003cp\u003eExploratory factor analysis (principal axis factoring, Varimax rotation) on all 35 survey items to test whether misconceptions clustered into coherent domains. The factor analysis served both as a validity check (i.e., did our survey tap the three intended domains?) and as a way to test whether folk theories naturally clustered in public cognition.\u003c/p\u003e\u003cp\u003e2. Qualitative Interviews\u003c/p\u003e\u003cp\u003e2.1 Participants and Procedure\u003c/p\u003e\u003cp\u003eA stratified random subsample of 60 participants (30 general public, 30 experts) completed semi-structured interviews via Zoom (30\u0026ndash;45 minutes). Interviews were audio-recorded, transcribed verbatim, and anonymized.\u003c/p\u003e\u003cp\u003e2.2 Interview Guide\u003c/p\u003e\u003cp\u003eThe guide covered four domains:\u003c/p\u003e\u003cp\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eBrain\u0026ndash;language relationships (e.g., \u0026ldquo;What happens in the brain when someone speaks?\u0026rdquo;),\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eBilingualism (e.g., \u0026ldquo;How do you think being bilingual affects the brain?\u0026rdquo;),\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eArtificial intelligence (e.g., \u0026ldquo;Do you think AI systems understand language like humans?\u0026rdquo;),\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eBelief sources and confidence (e.g., \u0026ldquo;Where did you learn this?\u0026rdquo;).\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003cp\u003eFollow-up questions probed contradictions (e.g., \u0026ldquo;Earlier you said language is in the left hemisphere, but now you mentioned both sides\u0026mdash;can you explain?\u0026rdquo;). This ensured that mixed or inconsistent beliefs were captured rather than smoothed over. Full question wording is provided in Appendix B.\u003c/p\u003e\u003cp\u003e2.3 Coding and Reliability\u003c/p\u003e\u003cp\u003eTwo independent coders analyzed transcripts using NVivo. Codes were developed deductively (e.g., \u0026ldquo;localization myth,\u0026rdquo; \u0026ldquo;AI-human equivalence\u0026rdquo;) and inductively (e.g., emergent metaphors such as \u0026ldquo;rewiring\u0026rdquo;). Contradictory statements were coded under multiple categories and flagged as \u0026ldquo;mixed responses.\u0026rdquo; Inter-rater reliability was strong (Cohen\u0026rsquo;s κ\u0026thinsp;=\u0026thinsp;0.81). Discrepancies were resolved through discussion. Unexpected themes (e.g., AI having \u0026ldquo;emotions\u0026rdquo;) were retained for qualitative description.\u003c/p\u003e\u003cp\u003e3. Media Content Analysis\u003c/p\u003e\u003cp\u003e3.1 Sampling Strategy\u003c/p\u003e\u003cp\u003eWe analyzed 150 English-language media items published between 2015\u0026ndash;2023, identified via YouTube, TED, Google News, and science blogs. Search terms included \u0026ldquo;language and brain,\u0026rdquo; \u0026ldquo;bilingualism,\u0026rdquo; \u0026ldquo;aphasia,\u0026rdquo; and \u0026ldquo;AI language.\u0026rdquo; Inclusion criteria: \u0026gt;500 words or \u0026gt;\u0026thinsp;5 minutes in length, targeted at a lay audience.\u003c/p\u003e\u003cp\u003eTo contextualize influence, we recorded popularity metrics (views, shares, subscriber counts). However, we deliberately treated each source equally in coding to avoid biasing analyses toward high-traffic items. Instead, popularity metrics are reported descriptively.\u003c/p\u003e\u003cp\u003e3.2 Coding Scheme\u003c/p\u003e\u003cp\u003eEach media source was systematically coded along several dimensions. We assessed the type of model used to describe brain\u0026ndash;language relationships (modular vs. distributed), the presence of metaphors (e.g., \u0026ldquo;rewiring the brain,\u0026rdquo; \u0026ldquo;left-brain logic\u0026rdquo;), the degree of scientific accuracy (alignment with peer-reviewed research), and the extent of anthropomorphic framing of AI (e.g., use of verbs such as \u0026ldquo;understands\u0026rdquo; or \u0026ldquo;thinks\u0026rdquo;). In addition, we coded credibility markers, including author credentials (neuroscientist, journalist, layperson), venue type (e.g., TED vs. personal blog), and whether peer-reviewed research was cited. To quantify lexical patterns across the corpus, we conducted lexicometric analyses\u0026mdash;such as frequency counts and collocation analysis\u0026mdash;using Python (v3.11).\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eWe analyzed survey data from 687 participants, comprising 363 individuals from the general public and 324 with formal training in neuroscience or cognitive science. Our investigation focused on three domains of public belief: language lateralization, bilingualism, and artificial intelligence (AI). The results revealed significant discrepancies between public beliefs and current neuroscientific models.\u003c/p\u003e\n\u003ch3\u003eLanguage Lateralization\u003c/h3\u003e\n\u003cp\u003eIn the domain of language lateralization, 62% of general public respondents (225 out of 363) endorsed the belief that language resides exclusively in the left hemisphere. In contrast, only 8% of expert respondents (26 out of 324) endorsed this view. The difference was statistically significant, χ\u0026sup2;(1, N\u0026thinsp;=\u0026thinsp;687)\u0026thinsp;=\u0026thinsp;210.4, p\u0026thinsp;\u0026lt;\u0026thinsp;.001, with a large effect size (Cramer\u0026rsquo;s V\u0026thinsp;=\u0026thinsp;.55). Only 13% of all participants referenced bilateral conceptual networks\u0026mdash;an essential element in Damasio and Damasio\u0026rsquo;s (1992) tripartite model. Additionally, we found that education level was positively associated with belief accuracy, with a correlation coefficient of r\u0026thinsp;=\u0026thinsp;0.36, 95% CI [0.27, 0.44], p\u0026thinsp;=\u0026thinsp;0.004ᵇ.\u003c/p\u003e\u003cp\u003e\u003cem\u003eFigure\u0026nbsp;1. Endorsement of Key Folk Theories by Group\u003c/em\u003e\u003c/p\u003e\u003cp\u003eBar graph comparing endorsement rates of three major folk theories\u0026mdash;language lateralization, bilingualism delay, and AI-human language equivalence\u0026mdash;between general public respondents (n\u0026thinsp;=\u0026thinsp;363) and neuroscience experts (n\u0026thinsp;=\u0026thinsp;324). The general public showed significantly higher belief in all three misconceptions. Bars represent percentage endorsement.\u003c/p\u003e\n\u003ch3\u003eBilingualism\u003c/h3\u003e\n\u003cp\u003eRegarding bilingualism, 41% of the general public believed that exposing children to two languages causes developmental delays, while only 29% selected the accurate statement that bilingualism enhances cognitive flexibility. By contrast, only 6% of experts (19/324) endorsed the delay myth, while 78% endorsed the flexibility advantage. This group difference was statistically significant, χ\u0026sup2;(1, N\u0026thinsp;=\u0026thinsp;687)\u0026thinsp;=\u0026thinsp;104.7, p\u0026thinsp;\u0026lt;\u0026thinsp;.001, Cramer\u0026rsquo;s V\u0026thinsp;=\u0026thinsp;.39. Participants from Germany and France were significantly more likely to choose the scientifically supported view, in contrast to those from the United States and India, χ\u0026sup2;(3, N\u0026thinsp;=\u0026thinsp;363)\u0026thinsp;=\u0026thinsp;18.9, p\u0026thinsp;=\u0026thinsp;0.0021ᶜ. These regional differences likely reflect the influence of multilingual education policies in shaping belief accuracy.\u003c/p\u003e\n\u003ch3\u003ePublic Perception of AI and Language\u003c/h3\u003e\n\u003cp\u003ePublic perception of AI and language also revealed widespread misunderstanding. A total of 58% of general public respondents (211 out of 363) agreed with the statement that \u0026ldquo;AI understands language like humans.\u0026rdquo; In contrast, only 7% of experts (24 out of 324) endorsed this statement. The group difference was highly significant, χ\u0026sup2;(1, N\u0026thinsp;=\u0026thinsp;687)\u0026thinsp;=\u0026thinsp;192.6, p\u0026thinsp;\u0026lt;\u0026thinsp;.001, Cramer\u0026rsquo;s V\u0026thinsp;=\u0026thinsp;.53. This belief was significantly more prevalent among individuals who reported frequent use of large language models such as ChatGPT, χ\u0026sup2;(1, N\u0026thinsp;=\u0026thinsp;363)\u0026thinsp;=\u0026thinsp;45.2, p\u0026thinsp;=\u0026thinsp;0.00091ᵈ.\u003c/p\u003e\n\u003ch3\u003eFactor Analysis of Misconceptions\u003c/h3\u003e\n\u003cp\u003eA factor analysis of survey responses was conducted across all 35 survey questions, rather than only those related to AI. Items clustered into three dominant misconception categories: \u0026ldquo;Localization Myths\u0026rdquo; (e.g., \u0026ldquo;Broca\u0026rsquo;s area handles grammar alone\u0026rdquo;), \u0026ldquo;Developmental Determinism\u0026rdquo; (e.g., \u0026ldquo;Children are blank slates\u0026rdquo;), and \u0026ldquo;AI-Human Equivalence\u0026rdquo; (e.g., \u0026ldquo;ChatGPT thinks like a person\u0026rdquo;). This analysis was used to validate that the most common misconceptions identified empirically map onto the three conceptual domains that guided survey design.\u003c/p\u003e\u003cp\u003e\u003cem\u003eFigure\u0026nbsp;2. Factor Loadings of Folk Theories on Three Key Misconception Clusters\u003c/em\u003e\u003c/p\u003e\u003cp\u003eHeatmap showing the results of exploratory factor analysis on survey responses. Items loaded onto three latent misconception clusters: Localization Myths, Developmental Determinism, and AI-Human Equivalence. Strong loadings (\u0026ge;\u0026thinsp;0.70) are indicated in red, weaker in blue. Factor analysis was conducted using principal axis factoring with Varimax rotation (n\u0026thinsp;=\u0026thinsp;687).\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eBelief Confidence and Media Exposure\u003c/h2\u003e\u003cp\u003eTo complement the quantitative findings, we conducted qualitative interviews with a randomized subset of 60 participants\u0026mdash;30 from the general public and 30 from the expert cohort. Among the general public, 63% reported media sources, including TED Talks and YouTube, as their primary source of neuroscience knowledge. Another 22% attributed their beliefs to school-based education, frequently citing textbook diagrams of \u0026ldquo;language centers,\u0026rdquo; while the remaining 15% drew on personal anecdotes, such as a child\u0026rsquo;s delayed speech.\u003c/p\u003e\u003cp\u003eInterview data also highlighted the influence of metaphor in shaping public understanding. Phrases like \u0026ldquo;rewiring the brain\u0026rdquo; were cited in 47% of interviews, while 32% referenced the notion of \u0026ldquo;left-brain logic.\u0026rdquo; Notably, belief confidence was positively associated with frequent exposure to media narratives, with a correlation of r\u0026thinsp;=\u0026thinsp;0.41, 95% CI [0.26, 0.54], p\u0026thinsp;=\u0026thinsp;0.009ᵉ, even among participants who held demonstrably inaccurate views. Media exposure was measured using a survey item asking participants, \u0026ldquo;How often do you engage with media content (e.g., TED Talks, YouTube videos, blogs, or articles) about the brain and language?\u0026rdquo; (1\u0026thinsp;=\u0026thinsp;never, 5\u0026thinsp;=\u0026thinsp;very often).\u003c/p\u003e\u003cp\u003e\u003cem\u003eFigure\u0026nbsp;3. Reported Sources of Belief About Language and the Brain\u003c/em\u003e\u003c/p\u003e\u003cp\u003ePie chart showing self-reported sources of belief among general public interview participants (n\u0026thinsp;=\u0026thinsp;30). Most cited media and TED Talks (63%), followed by formal education (22%) and personal experience (15%). Responses were coded from open-ended qualitative interviews.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eMedia Content Analysis\u003c/h3\u003e\n\u003cp\u003eA media content analysis of 150 public-facing texts\u0026mdash;including articles, TED Talks, and educational videos\u0026mdash;offered further insights into the origin of these beliefs. Only 18% of sources accurately reflected the tripartite model of language processing. In contrast, 67% relied on outdated or oversimplified terminology, such as \"language module\" or \"speech center.\" Technical terms like \u0026ldquo;bilateral conceptual networks\u0026rdquo; appeared in just 12% of the media corpus, while metaphorical analogies\u0026mdash;such as likening the brain to a computer\u0026mdash;were found in 88% of cases. Lexicometric analysis supported these patterns: \u0026ldquo;left-brain\u0026rdquo; appeared nearly four times more frequently than \u0026ldquo;bilateral.\u0026rdquo; In AI-focused content, 72% of sources used anthropomorphic verbs such as \u0026ldquo;understands\u0026rdquo; and \u0026ldquo;learns,\u0026rdquo; contributing to the widespread misconception that AI possesses human-like cognitive capacities.\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;4. Frequency of Dominant Metaphors in Media Coverage\u003c/p\u003e\u003cp\u003eBar graph depicting term frequency (n\u0026thinsp;=\u0026thinsp;150 media samples) for technical and metaphorical language describing brain-language processes and artificial intelligence. Metaphors such as \u0026ldquo;left-brain,\u0026rdquo; \u0026ldquo;rewiring,\u0026rdquo; and \u0026ldquo;understands\u0026rdquo; appeared far more frequently than scientific terms like \u0026ldquo;mediation\u0026rdquo; or \u0026ldquo;bilateral.\u0026rdquo;\u003c/p\u003e\n\u003ch3\u003eIndividual Differences Across Demographics\u003c/h3\u003e\n\u003cp\u003eCross-demographic and cross-cultural analyses revealed further disparities. Participants with STEM backgrounds were 2.3 times more likely to reject localization myths than those without, with an odds ratio of 2.3, 95% CI [1.7, 3.1], p\u0026thinsp;=\u0026thinsp;0.0017ᶠ. Belief in AI-human equivalence was highest among participants from the United States and India, where 65% endorsed this view, compared to only 51% of European participants. This cross-national difference was statistically significant, χ\u0026sup2;(2, N\u0026thinsp;=\u0026thinsp;363)\u0026thinsp;=\u0026thinsp;14.3, p\u0026thinsp;=\u0026thinsp;0.0064ᵍ. Meanwhile, participants from Germany and France were more likely to select scientifically accurate statements about bilingualism, suggesting a potential link between public policy and belief accuracy.\u003c/p\u003e\u003cp\u003eAge and gender were also collected but did not show significant associations with belief accuracy. Full descriptive and inferential statistics for these variables are provided in Supplementary Table S1.\u003c/p\u003e\u003cp\u003e\u003cem\u003eFigure\u0026nbsp;5. Cross-Cultural Variation in Misconception Endorsement\u003c/em\u003e\u003c/p\u003e\u003cp\u003eGrouped bar graph displaying endorsement rates of three misconception categories across six countries: United States, France, Germany, China, India, and Spain. Cross-cultural variation was most pronounced in beliefs about bilingualism and AI equivalence. Total survey sample: n\u0026thinsp;=\u0026thinsp;363.\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eQualitative Interview Themes\u003c/h2\u003e\u003cp\u003eFinally, we assessed the degree to which public beliefs align with contemporary neuroscientific frameworks. Overall, 89% of respondents referenced the left hemisphere as the exclusive site of language, while only 6% mentioned intermediary mediation structures, and just 2% acknowledged any role for subcortical regions. In contrast, expert participants frequently described language as a dynamic, distributed process involving the interaction of syntax, semantics, and broader cognitive functions. The general public, however, continued to rely on static, modular metaphors that do not reflect the current state of neuroscience.\u003c/p\u003e\u003cp\u003e\u003cem\u003eFigure\u0026nbsp;6. Public vs. Expert Emphasis on Brain-Language Concepts\u003c/em\u003e\u003c/p\u003e\u003cp\u003eHorizontal bar graph comparing the relative emphasis on five types of language-brain concepts in open-ended responses from the general public (n\u0026thinsp;=\u0026thinsp;30) and expert participants (n\u0026thinsp;=\u0026thinsp;30). Experts were more likely to reference distributed systems and mediation structures, while public responses favored modular or static explanations.\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eOur study indicates a persistent and multi-layered disconnect between contemporary neuroscience and public beliefs about how the brain processes language. Across survey, interview, and media data, we found that folk theories consistently cluster around three dominant misconceptions: localization myths, developmental determinism, and AI anthropomorphism. These clusters represent not only epistemic simplifications but also deeply entrenched conceptual frameworks shaped by language, metaphor, and cultural narrative. They align with prior work showing that folk theories in other cognitive domains\u0026mdash;such as memory (as \u0026ldquo;storage\u0026rdquo;), attention (as \u0026ldquo;spotlight\u0026rdquo;), or intelligence (as fixed \u0026ldquo;capacity\u0026rdquo;)\u0026mdash;rely on similarly reductive heuristics (e.g., Furnham, 1996; Roediger, 1980). Our findings therefore extend the broader folk theory literature by illustrating how language, like other cognitive functions, is filtered through familiar but misleading conceptual shortcuts.\u003c/p\u003e\u003cp\u003eThe most prevalent folk theory\u0026mdash;localization of language in the left hemisphere\u0026mdash;stands in stark contrast to modern models of distributed and bilateral neural processing (Damasio \u0026amp; Damasio, 1992). While localization theories have historical significance (e.g., Broca, 1861), our findings suggest they have been overgeneralized in the public mind. Only a small minority of participants mentioned mediation structures or bilateral conceptual networks\u0026mdash;key components in contemporary models. These omissions reflect a widespread reliance on spatial and modular metaphors (\u0026ldquo;language center,\u0026rdquo; \u0026ldquo;speech area\u0026rdquo;) that obscure the dynamic and interactive nature of language processing. While it is true that this myth may not carry the same societal weight as misconceptions about bilingualism or AI, it still matters: if the public believes language resides exclusively in one spot, they may misunderstand the consequences of brain injury, underestimate recovery potential, or support oversimplified clinical interventions. In this sense, even a \u0026ldquo;low-stakes\u0026rdquo; myth about lateralization shapes how people think about health, disability, and rehabilitation. Importantly, acknowledging alternative models (e.g., Friederici, Poeppel, Pylkk\u0026auml;nen) reinforces that \u0026ldquo;not Damasio\u0026rdquo; does not mean \u0026ldquo;folk theory\u0026rdquo;; rather, the folk version is the outdated idea of a single left-hemisphere \u0026ldquo;language organ.\u0026rdquo;\u003c/p\u003e\u003cp\u003eA second cluster of misconceptions centers on developmental determinism\u0026mdash;the belief that language acquisition is a passive process governed by immutable developmental stages or \u0026ldquo;blank slate\u0026rdquo; brains. This belief continues to inform public opinion on bilingualism, with 41% of participants believing it causes delays. This is in direct opposition to empirical research showing cognitive and social advantages associated with dual-language exposure (Byers-Heinlein \u0026amp; Lew-Williams, 2013). The persistence of this myth suggests that both educational materials and clinical messaging have not sufficiently integrated findings from neuroscience on plasticity, sensitive periods, and variability in language development. Unlike the lateralization myth, this misconception has immediate and tangible social consequences, influencing parenting choices, educational practice, and immigration debates. As with folk models of intelligence that emphasize innate ability over learning, these deterministic views simplify complexity but carry consequences for how children are raised, taught, and assessed.\u003c/p\u003e\u003cp\u003eThe third and most contemporary cluster involves AI anthropomorphism. Over half of participants agreed that AI systems \u0026ldquo;understand language like humans,\u0026rdquo; with higher endorsement among those frequently interacting with LLMs. This belief is both intuitive and misleading. The fluency of models like ChatGPT triggers anthropomorphic biases, leading users to ascribe cognitive and emotional capacities where none exist (Bender \u0026amp; Koller, 2020; Weizenbaum, 1966). Our media analysis showed that 72% of AI-focused content used human-like verbs such as \u0026ldquo;understands\u0026rdquo; or \u0026ldquo;thinks,\u0026rdquo; reinforcing the illusion of consciousness. This linguistic framing matters\u0026mdash;not just because it distorts public understanding, but because it may shape attitudes toward technology, education policy, and even mental health interventions. Here, as in folk theories of memory and intelligence, metaphorical reasoning produces powerful but inaccurate intuitions that shape real-world decision-making.\u003c/p\u003e\u003cp\u003eA key theme emerging from both interviews and media content is the power of metaphor in shaping cognitive models. Phrases like \u0026ldquo;rewiring the brain\u0026rdquo; and \u0026ldquo;language centers\u0026rdquo; function as communicative shortcuts, but they also encode simplified and sometimes inaccurate representations. Only 18% of media sources accurately described distributed models of brain-language interaction, while nearly 90% leaned on familiar but outdated metaphors. The \u0026ldquo;so what\u0026rdquo; of these results is clear: metaphors are not neutral\u0026mdash;they shape folk theories, and folk theories in turn shape behavior, from whether parents raise children bilingually to how societies interpret the promises and risks of AI. What connects our three focal myths is precisely their metaphorical foundation: \u003cem\u003efolk phrenology\u003c/em\u003e in the case of lateralization, \u003cem\u003ebiological clocks and blank slates\u003c/em\u003e in the case of development, and \u003cem\u003eanthropomorphic personification\u003c/em\u003e in the case of AI. Each myth survives because it is grounded in metaphors that are intuitively compelling, easy to communicate, and widely reinforced in culture. To move forward, we need better metaphors: for example, describing Broca\u0026rsquo;s area as a \u0026ldquo;hub in a distributed network\u0026rdquo; rather than a \u0026ldquo;center,\u0026rdquo; or AI as a \u0026ldquo;mirror of statistical patterns\u0026rdquo; rather than a \u0026ldquo;learner.\u0026rdquo; Such metaphors retain accessibility while reducing distortion.\u003c/p\u003e\u003cp\u003eThe observed demographic differences\u0026mdash;particularly between STEM-trained participants and those without formal science education\u0026mdash;point to promising avenues for intervention. Participants with STEM backgrounds were significantly more likely to reject localization myths and AI anthropomorphism, suggesting that science education serves as a protective factor against misinformation. Cross-cultural differences, particularly higher belief accuracy in multilingual countries with policy-driven education (e.g., Germany, France), further reinforce the importance of systemic educational strategies. Additionally, media literacy training could help the public better evaluate claims presented in news articles, TED Talks, and AI-generated content.\u003c/p\u003e\u003cp\u003eOverall, our findings underscore the importance of targeted science communication strategies. Addressing public misconceptions about language and the brain will require a combination of accurate metaphor use, integration of current neuroscience into curricula, and critical engagement with how AI technologies are described in public discourse. As generative language technologies become more visible and persuasive, the risk of reinforcing intuitive but incorrect folk theories grows. Bridging the divide between expert models and public belief is therefore not just an educational challenge\u0026mdash;it is a necessary foundation for evidence-based policy, ethical technology design, and a more cognitively informed society.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAcknowledgments\u003c/h2\u003e\u003cp\u003eWe thank the participants from the general public and neuroscience communities for generously sharing their time and insights. We are also grateful to the faculty and staff at Johns Hopkins University for facilitating recruitment. This research was supported by the Neuroscience and Cognitive Science Departments at Johns Hopkins University.\u003c/p\u003e\u003cp\u003e The authors affirm that all research involving human subjects was conducted in accordance with institutional ethics guidelines.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBates E, MacWhinney B (1989) Functionalist approaches to grammar. In: MacWhinney B, Bates E (eds) The crosslinguistic study of sentence processing. Cambridge University Press, pp 3\u0026ndash;73\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBender EM, Koller A (2020) Climbing towards NLU: On meaning, form, and understanding in the age of data. In \u003cem\u003eProceedings of the 58th Annual Meeting of the Association for Computational Linguistics\u003c/em\u003e (pp. 5185\u0026ndash;5198). Association for Computational Linguistics. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18653/v1/2020.acl-main.463\u003c/span\u003e\u003cspan address=\"10.18653/v1/2020.acl-main.463\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBishop DVM (2014) Ten questions about terminology for children with unexplained language problems. Int J Lang Communication Disorders 49(4):381\u0026ndash;415. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/1460-6984.12101\u003c/span\u003e\u003cspan address=\"10.1111/1460-6984.12101\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBroca P (1861) Remarks on the seat of the faculty of articulated language, following an observation of aphemia (loss of speech). Bulletins de la Soci\u0026eacute;t\u0026eacute; Anatomique de Paris 6:330\u0026ndash;357\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eByers-Heinlein K, Lew-Williams C (2013) Bilingualism in the early years: What the science says. Learn Landscapes 7(1):95\u0026ndash;112. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC6168212/\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6168212/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDamasio AR, Damasio H (1992) Brain and language. Sci Am 267(3):88\u0026ndash;95. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/scientificamerican0992-88\u003c/span\u003e\u003cspan address=\"10.1038/scientificamerican0992-88\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eElman JL (2004) An alternative view of the mental lexicon. Trends Cogn Sci 8(7):301\u0026ndash;306. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.tics.2004.05.003\u003c/span\u003e\u003cspan address=\"10.1016/j.tics.2004.05.003\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eElman JL (2016) Language, plasticity, and learning: Challenges of the new millennium. In \u003cem\u003eProceedings of the 40th Annual Boston University Conference on Language Development\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFodor JA (1983) The modularity of mind. MIT Press\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFriederici AD (2011) The brain basis of language processing: From structure to function. Physiol Rev 91(4):1357\u0026ndash;1392. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1152/physrev.00006.2011\u003c/span\u003e\u003cspan address=\"10.1152/physrev.00006.2011\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFurnham A (1996) Lay theories: Everyday understanding of problems in the social sciences. Hum Relat 49(2):157\u0026ndash;173\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKuhl PK (2010) Brain mechanisms in early language acquisition. Neuron 67(5):713\u0026ndash;727. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.neuron.2010.08.038\u003c/span\u003e\u003cspan address=\"10.1016/j.neuron.2010.08.038\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKaplan J (2016) Women talk more than men \u0026hellip; and other myths about language explained. Cambridge University Press\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLahecka T (2021) Folk linguistics in applied contexts: Teachers\u0026rsquo; beliefs about language learning. J Lang Educ 7(1):20\u0026ndash;34\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLilienfeld SO, Lynn SJ, Ruscio J, Beyerstein BL (2015) 50 great myths of popular psychology: Shattering widespread misconceptions about human behavior. Wiley-Blackwell\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMayberry RI (2023) Sensitive periods in language development: Implications for education and clinical practice. Oxford University Press\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNiedzielski N, Preston D (2000) Folk linguistics. Mouton de Gruyter\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eO\u0026rsquo;Connor C, Joffe H (2020) How has neuroscience affected lay understandings of personhood? A review of the evidence. Public Underst Sci 29(1):88\u0026ndash;102. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://journals.sagepub.com/doi/full/10.1177/0963662513476812\u003c/span\u003e\u003cspan address=\"https://journals.sagepub.com/doi/full/10.1177/0963662513476812\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePoeppel D (2014) The neuroanatomic and neurophysiological infrastructure for speech and language. Curr Opin Neurobiol 28:142\u0026ndash;149. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.conb.2014.07.005\u003c/span\u003e\u003cspan address=\"10.1016/j.conb.2014.07.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePylkk\u0026auml;nen L (2019) The neural basis of combinatory syntax and semantics. Science 366(6461):62\u0026ndash;66. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1126/science.aax0050\u003c/span\u003e\u003cspan address=\"10.1126/science.aax0050\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRoediger HL (1980) Memory metaphors in cognitive psychology. Mem Cognit 8(3):231\u0026ndash;246. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3758/BF03213419\u003c/span\u003e\u003cspan address=\"10.3758/BF03213419\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSaffran JR, Aslin RN, Newport EL (1996) Statistical learning by 8-month-old infants. Science 274(5294):1926\u0026ndash;1928. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1126/science.274.5294.1926\u003c/span\u003e\u003cspan address=\"10.1126/science.274.5294.1926\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTomasello M (2003) Constructing a language: A usage-based theory of language acquisition. Harvard University Press\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWagner K, Clopper CG, Pate JK (2023) Lay beliefs about language learning and bilingualism. Appl Psycholinguist 44(2):241\u0026ndash;262\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWeizenbaum J (1966) ELIZA\u0026mdash;A computer program for the study of natural language communication between man and machine. Commun ACM 9(1):36\u0026ndash;45. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/365153.365168\u003c/span\u003e\u003cspan address=\"10.1145/365153.365168\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1. Descriptive Summary Statistics for Key Beliefs about Language and the Brain\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"636\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 216px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariable\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMin\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMax\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 84px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 216px;\"\u003e\n \u003cp\u003eLeft hemisphere = language\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e363\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.49\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 216px;\"\u003e\n \u003cp\u003eBilingualism delays speech\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e363\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.48\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 216px;\"\u003e\n \u003cp\u003eAI understands like humans\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e363\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.49\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 216px;\"\u003e\n \u003cp\u003eBelief accuracy (STEM group)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e324\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.83\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.27\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 216px;\"\u003e\n \u003cp\u003eBelief accuracy\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;(non-STEM group)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e363\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 84px;\"\u003e\n \u003cp\u003e0.34\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eThe table presents descriptive statistics for belief endorsement across five key statements related to language processing and cognition. Mean values represent the proportion of participants endorsing each belief (e.g., selecting \u0026ldquo;agree\u0026rdquo; or equivalent response options). \u0026ldquo;Left hemisphere = language\u0026rdquo; reflects belief in strict lateralization; \u0026ldquo;Bilingualism delays speech\u0026rdquo; reflects the common misconception that early dual-language exposure hinders development; \u0026ldquo;AI understands like humans\u0026rdquo; gauges anthropomorphic beliefs about language models. Accuracy scores for STEM vs. non-STEM groups reflect composite scores based on alignment with current neuroscientific models. SD = standard deviation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 2. Summary of Statistical Analyses Conducted Across Study Designs\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"672\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 250px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePurpose\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 159px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eStatistical Method\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 70px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSample Size (N)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 193px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eKey Outputs\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 250px;\"\u003e\n \u003cp\u003eCompare belief accuracy by education level\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 159px;\"\u003e\n \u003cp\u003ePearson correlation (r)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 70px;\"\u003e\n \u003cp\u003e687\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 193px;\"\u003e\n \u003cp\u003er = 0.36, 95% CI [0.27, 0.44], p = 0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 250px;\"\u003e\n \u003cp\u003eCompare public vs. expert belief rates\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 159px;\"\u003e\n \u003cp\u003eChi-square test\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 70px;\"\u003e\n \u003cp\u003e687\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 193px;\"\u003e\n \u003cp\u003e\u0026chi;\u0026sup2;(1) = 210.4, p = 0.00012\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 250px;\"\u003e\n \u003cp\u003eCompare AI beliefs by LLM exposure frequency\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 159px;\"\u003e\n \u003cp\u003eChi-square test\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 70px;\"\u003e\n \u003cp\u003e363\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 193px;\"\u003e\n \u003cp\u003e\u0026chi;\u0026sup2;(1) = 45.2, p = 0.00091\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 250px;\"\u003e\n \u003cp\u003eExamine belief confidence vs. media exposure\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 159px;\"\u003e\n \u003cp\u003ePearson correlation (r)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 70px;\"\u003e\n \u003cp\u003e60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 193px;\"\u003e\n \u003cp\u003er = 0.41, 95% CI [0.26, 0.54], p = 0.009\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 250px;\"\u003e\n \u003cp\u003eCompare bilingualism beliefs by country\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 159px;\"\u003e\n \u003cp\u003eChi-square test\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 70px;\"\u003e\n \u003cp\u003e363\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 193px;\"\u003e\n \u003cp\u003e\u0026chi;\u0026sup2;(3) = 18.9, p = 0.0021\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 250px;\"\u003e\n \u003cp\u003eCompare AI beliefs by region (U.S./Japan vs. Europe)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 159px;\"\u003e\n \u003cp\u003eChi-square test\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 70px;\"\u003e\n \u003cp\u003e363\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 193px;\"\u003e\n \u003cp\u003e\u0026chi;\u0026sup2;(2) = 14.3, p = 0.0064\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 250px;\"\u003e\n \u003cp\u003eCompare myth rejection rates by STEM background\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 159px;\"\u003e\n \u003cp\u003eOdds ratio, chi-square test\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 70px;\"\u003e\n \u003cp\u003e687\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 193px;\"\u003e\n \u003cp\u003eOR = 2.3, 95% CI [1.7, 3.1], p = 0.0017\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eThe table summarizes all major statistical analyses performed in the study, categorized by design type. It includes the structure of the data, statistical test employed, sample sizes, and corresponding results. Chi-square and correlation analyses compared belief endorsement across groups, while exploratory factor analysis revealed latent misconception clusters. Cross-cultural and between-group comparisons further evaluated demographic effects. p-values and confidence intervals are reported where applicable to characterize the significance and uncertainty of observed effects.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Johns Hopkins University","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Folk Theories, Language Processing, Damasio’s tripartite neural framework","lastPublishedDoi":"10.21203/rs.3.rs-7941468/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7941468/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003ePublic understanding of how the brain processes language remains shaped more by folk theories than by modern neuroscience. Misconceptions and oversimplification such as \u0026ldquo;language lives in the left brain,\u0026rdquo; \u0026ldquo;bilingualism causes delays,\u0026rdquo; or \u0026ldquo;AI understands language like humans\u0026rdquo; persist across education, media, and healthcare discourse. Yet scientific models, including Damasio and Damasio\u0026rsquo;s tripartite neural framework, reveal a far more nuanced picture\u0026mdash;where language emerges from interactions between bilateral conceptual networks, left-lateralized linguistic encoders, and intermediary systems that mediate between meaning and form. This study investigates the prevalence and origins of these folk theories using a mixed-methods approach, including public surveys, qualitative interviews, and media analysis. We identify the most common public beliefs, map them against neuroscientific models, and explore how misinformation travels through education, personal experience, and popular media. Our findings reveal not only widespread misalignment between public beliefs and neuroscientific consensus, but also thematic patterns shaped by metaphor, oversimplification, and cognitive bias. We conclude by offering communication strategies for scientists, educators, and policymakers seeking to close the gap between brain science and public perception of language.\u003c/p\u003e","manuscriptTitle":"Folk Theories of Language and the Brain: Public Beliefs About Speech, Bilingualism, and Neural Function","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-28 16:30:17","doi":"10.21203/rs.3.rs-7941468/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"b53b9954-15c3-4c31-a9b1-675b528e5fe7","owner":[],"postedDate":"October 28th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":56839351,"name":"Cognitive Neuroscience"}],"tags":[],"updatedAt":"2025-10-28T16:30:17+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-28 16:30:17","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7941468","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7941468","identity":"rs-7941468","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}