Catastrophes in Moral Belief: An Active-Inference Account of Hysteresis and Lock-In in Antisocial Behaviour

Catastrophes in Moral Belief: An Active-Inference Account of Hysteresis and Lock-In in Antisocial Behaviour | 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 Catastrophes in Moral Belief: An Active-Inference Account of Hysteresis and Lock-In in Antisocial Behaviour Regio Marcos Abreu-Filho This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8448048/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 Many antisocial phenotypes exhibit abrupt onset, remarkable stability once established, and striking resistance to intervention. Using active inference as a foundational framework, this paper develops a minimal scalar model of moral belief updating in which saddle-node bifurcations and cusp catastrophes naturally emerge from the geometry of variational free energy. The resulting drift field gives rise to path-dependent moral regimes, critical slowing down at individually variable tipping points, and evolutionary lock-in of antisocial traits. All results are derived analytically or obtained from reproducible simulations. Figures Figure 1 1. Introduction Antisocial behavior often presents clinically as a rigid, resistant phenotype. Theoretical models suggest these behaviors may exhibit non-linear dynamics similar to physical systems. We propose a computational model where choices are governed by a latent one-dimensional moral belief state (theta), representing the perceived harmfulness of antisocial behaviour in the current institutional and social environment. The agent does not observe theta directly but maintains a variational posterior q(theta) = N(mu, Sigma), where Sigma is fixed. The mean mu serves as the cognitive state variable. 2. Methods 2.1 Moral Outcomes and Logistic Likelihood On trial t, the agent observes binary outcome h in {0,1} (harm). The likelihood is logistic (Eq. 1): p(h = 1 | theta, c) = sigma(alpha * theta + gamma * c) where sigma(z) = 1 / (1 + e^-z) With alpha > 0 and gamma being real numbers. Higher theta or stronger sanction context c > 0 increases the probability of observing harm. 2.2 Variational Free Energy and Gradient A Gaussian prior is placed on theta: p(theta) = N(m0, s0^2). Belief updating minimises variational free energy F (Eq. 2): F(q, h, c) = KL[q(theta)||p(theta)] - E_q[log p(h|theta,c)] With fixed Sigma, F depends only on mu. The gradient is (Eq. 3): dF/dmu = (mu - m0)/s0^2 - alpha * E_q[h - sigma(alpha*theta + gamma*c)] The second term is a moral prediction error that drives mu upward when harm is unexpectedly frequent. 2.3 Drift, Learning Rate, and Epistemic Volatility Gradient flow on F yields deterministic dynamics which, with added Langevin noise, becomes (Eq. 5): d(mu)/dt = eta * g(mu, c; d) + xi * noise(t) where g(mu, c; d) := - dF/dmu Where eta > 0 is the learning rate, xi controls volatility, and d is an asymmetry trait. 2.4 Action Selection and Expected Free Energy Actions are selected by softmax over expected free energies. We use a pragmatic (outcome-dependent) loss function: L(Action = Antisocial) = -(12 + 6 * tanh(d * c)) Positive 'd' makes antisocial choices more sensitive to high-sanction contexts. The advantage of Antisocial (A) over Social (S) action is (Eq. 6): Delta Q = [8 + 6 * tanh(d*c)] * sigma(alpha*mu + gamma*c + d) The probability of choosing the antisocial action is given by the sigmoid function with inverse temperature beta (Eq. 7): Pr(Action = A) = sigma(beta * Delta Q) 3. Results 3.1 Deterministic Dynamics & Catastrophes In the deterministic limit (xi = 0), fixed points satisfy g(mu) = 0. Varying context (c) and trait (d) jointly yields a cusp catastrophe surface (Thom, 1975 ). Hysteresis emerges naturally: gradual increase in sanction strength keeps the agent on the high-antisocial branch until the upper fold vanishes. 3.2 Critical Slowing Down The corresponding Fokker-Planck equation governs probability flow. Near folds, drift magnitude approaches 0, producing critical slowing down. Synthetic data (Fig. 1 ) show the diagnostic signature: slow reaction times peaking at sudden flips. 4. Discussion 4.1 Evolutionary Lock-In Traits evolve on a slower timescale. Slow-fast analysis shows that populations can become trapped in regions of trait space where only antisocial attractors remain accessible, even if institutional change would otherwise permit escape. 4.2 Clinical Implications Low eta captures reduced sensitivity to punitive feedback seen in psychopathy (Blair, 2007 ). Negative 'd' aligns with callous-unemotional traits that reward norm violation under peer influence (Haidt's harm/care foundation). The model predicts wider hysteresis and pronounced RT peaks in these individuals. 4.3 Conclusion and Future Directions This framework offers a novel, mathematically grounded explanation for the rigidity of antisocial phenotypes. By framing moral belief updating as a dynamical system, we account for the 'stickiness' of maladaptive behaviors through hysteresis rather than simple reinforcement failure. Future work will extend this minimal scalar model by treating precision dynamics (Sigma) as an evolving state variable rather than a fixed parameter. Furthermore, empirical studies using ramping-sanction moral-dilemma tasks are required to validate the predicted twin-peak reaction time signatures. Statements and Declarations Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Conflicts of interest/Competing interests The author declares that there are no conflicts of interest or competing interests related to this work. Availability of data and material This article is based on theoretical and computational modelling and does not report empirical datasets. Any simulated data used to illustrate the model are available from the author on reasonable request. Code availability Prototype code implementing the active-inference model and catastrophe analysis used in “Catastrophes in Moral Belief: An Active-Inference Account of Hysteresis and Lock-In in Antisocial Behaviour” is available from the author on reasonable request. Authors’ contributions The sole author was responsible for the conception and design of the study, development of the active-inference framework, mathematical and computational analysis, interpretation of results, and writing and revision of the manuscript. Ethics approval This work is a theoretical/computational modelling study and does not involve human participants, identifiable personal data, or experiments on animals. Therefore, ethics committee approval was not required. Consent to participate Not applicable. No human participants were involved in this study. Consent for publication Not applicable. The manuscript does not contain any individual person’s data in any form (including images, case details, or quotations) that would require consent for publication. Author Contribution As the sole author, I conducted all aspects of the study. This included formulation of the research question and theoretical framework; design, implementation, and validation of the active inference–based generative models and diffusion-model pipelines for synthetic multimodal (behavioral, clinical, neuroimaging) data; execution of all benchmarking and model-recovery analyses; statistical evaluation of fidelity between synthetic and empirical datasets; interpretation of results; and preparation of the manuscript in its entirety.The work relied heavily on established methods from computational psychiatry, Bayesian modeling, and generative deep learning literature. No claim of exceptional novelty or technical virtuosity is made; the contribution consists primarily in the systematic integration and rigorous validation of these existing tools for the specific purpose of addressing data bottlenecks in the field. Data Availability Code is available at: https://github.com/regio-abreu/moral-catastrophes-active-inference References Blair, R. J. R. (2007). The amygdala and ventromedial prefrontal cortex in morality and psychopathy. Trends in Cognitive Sciences, 11(9), 387-392. Crockett, M. J., et al. (2013). Models of morality. Trends in Cognitive Sciences, 17(8), 363-366. Thom, R. (1975). Structural stability and morphogenesis. Reading, MA: WA Benjamin. Friston, K., et al. (2017). Active inference: A process theory. Neural Computation, 29(1), 1-49. Scheffer, M., et al. (2009). Early-warning signals for critical transitions. Nature, 461(7260), 53-59. Additional Declarations No competing interests reported. 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-8448048","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":565450405,"identity":"aa0c62bb-9ff5-4817-904c-ce0e7d3fb941","order_by":0,"name":"Regio Marcos Abreu-Filho","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABEElEQVRIiWNgGAWjYJCCAzAGM4jgBxEJBaRokWwAaTEg0jqwFgOwCXi0mLe3XzzwMccun7+9/eHnwj02iZvPr0788MCAQZ5f7ABWLTJnzhQcnLkt2XLGmTPG0jOepSVuu/F2swTQYYYzZydg1SIhkZNwmHcbs4GBRA6DNM+Bw0AtZzeAtCQY3MahRf4NSEs9UEv64988B/4nbp5xdvMPvFok2A8AtRwGakkwA9pyIHEDf+82/Lbw5DAA/XLcQOLMGTPrGQeSjWfc4N1mkWAggdsv7Mcff/i4rdoAGGKPbxccsJPt7z+7+eaPCht5fmnsWhgYeNCjQAKsUgKHchBgf4AmwH8Aj+pRMApGwSgYiQAAiYplljuoJ34AAAAASUVORK5CYII=","orcid":"","institution":"Pontifícia Universidade Católica 2 do Paraná (PUC-PR)","correspondingAuthor":true,"prefix":"","firstName":"Regio","middleName":"Marcos","lastName":"Abreu-Filho","suffix":""}],"badges":[],"createdAt":"2025-12-25 09:38:53","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8448048/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8448048/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":99313386,"identity":"12d47b20-648d-4e35-9c5a-9a597b38bd23","added_by":"auto","created_at":"2025-12-31 16:20:05","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":259974,"visible":true,"origin":"","legend":"","description":"","filename":"submissionmanuscriptperfected1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/5e1dad9f032383092f4dfd98.docx"},{"id":99020132,"identity":"d3fc22ce-62cc-4c46-acd5-5a0020792b29","added_by":"auto","created_at":"2025-12-26 04:39:56","extension":"json","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3380,"visible":true,"origin":"","legend":"","description":"","filename":"48743592ad8f4405a81154d16fd26386.json","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/e678f6cd663d3745ce4eb74f.json"},{"id":99020136,"identity":"af4604ea-fd25-4a35-b9a3-a2ad8c94f53a","added_by":"auto","created_at":"2025-12-26 04:39:56","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6039,"visible":true,"origin":"","legend":"","description":"","filename":"statements.docx","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/bda17044dd85b9f682d0c884.docx"},{"id":99314095,"identity":"36d91f0b-e431-411b-89bf-a271f38bfaf7","added_by":"auto","created_at":"2025-12-31 16:20:50","extension":"xml","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":17610,"visible":true,"origin":"","legend":"","description":"","filename":"48743592ad8f4405a81154d16fd263861enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/899ded7a0f853849c45c450d.xml"},{"id":99313446,"identity":"8a59d606-e5ce-41e4-993b-b29177b7c205","added_by":"auto","created_at":"2025-12-31 16:20:11","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":518448,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/62a101d2a6e3606170626cc3.jpeg"},{"id":99313512,"identity":"1a154dfe-cc8f-43e3-b50f-d9fe8b704971","added_by":"auto","created_at":"2025-12-31 16:20:15","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":271552,"visible":true,"origin":"","legend":"","description":"","filename":"moraldynamicsfig1.png","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/537df61b718a6c0ac9082677.png"},{"id":99020137,"identity":"eb6dcdb9-7142-40ea-915e-44e61b7dc306","added_by":"auto","created_at":"2025-12-26 04:39:56","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":84382,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/629e89aa7f278e8a54d7ac91.png"},{"id":99020139,"identity":"803e353a-85ab-44e4-9e87-60789a57df6a","added_by":"auto","created_at":"2025-12-26 04:39:57","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":108070,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinemoraldynamicsfig1.png","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/47361af26c7b10722af116c5.png"},{"id":99020140,"identity":"75e2b47a-511a-45f0-a45c-23831316d145","added_by":"auto","created_at":"2025-12-26 04:39:57","extension":"xml","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15701,"visible":true,"origin":"","legend":"","description":"","filename":"48743592ad8f4405a81154d16fd263861structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/3c89a8b9d7eaf3de321f3e98.xml"},{"id":99020141,"identity":"4a55c3cd-5866-49aa-8009-6c70a94932e8","added_by":"auto","created_at":"2025-12-26 04:39:57","extension":"html","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20726,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/44d52bc6f69862363f82c5ba.html"},{"id":99020131,"identity":"de2e714b-3ba9-4122-ae9a-8f2ccf4a671e","added_by":"auto","created_at":"2025-12-26 04:39:56","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":271552,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eSimulation of moral hysteresis. Top: Belief state transitions showing path dependence. Bottom: Reaction times peaking at bifurcation points.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"moraldynamicsfig1.png","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/6fd427ebcd930d91413874f6.png"},{"id":104808443,"identity":"6dacafc4-d1b2-4c18-99b6-a75bfdcbb651","added_by":"auto","created_at":"2026-03-17 12:37:38","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":720991,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8448048/v1/1df8fd4c-5c26-4627-bce2-442a44b85b3d.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Catastrophes in Moral Belief: An Active-Inference Account of Hysteresis and Lock-In in Antisocial Behaviour","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eAntisocial behavior often presents clinically as a rigid, resistant phenotype. Theoretical models suggest these behaviors may exhibit non-linear dynamics similar to physical systems. We propose a computational model where choices are governed by a latent one-dimensional moral belief state (theta), representing the perceived harmfulness of antisocial behaviour in the current institutional and social environment.\u003c/p\u003e \u003cp\u003eThe agent does not observe theta directly but maintains a variational posterior q(theta)\u0026thinsp;=\u0026thinsp;N(mu, Sigma), where Sigma is fixed. The mean mu serves as the cognitive state variable.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Moral Outcomes and Logistic Likelihood\u003c/h2\u003e \u003cp\u003eOn trial t, the agent observes binary outcome h in {0,1} (harm). The likelihood is logistic (Eq.\u0026nbsp;1):\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003ep(h\u0026thinsp;=\u0026thinsp;1 | theta, c)\u0026thinsp;=\u0026thinsp;sigma(alpha * theta\u0026thinsp;+\u0026thinsp;gamma * c) where sigma(z)\u0026thinsp;=\u0026thinsp;1 / (1\u0026thinsp;+\u0026thinsp;e^-z)\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWith alpha\u0026thinsp;\u0026gt;\u0026thinsp;0 and gamma being real numbers. Higher theta or stronger sanction context c\u0026thinsp;\u0026gt;\u0026thinsp;0 increases the probability of observing harm.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Variational Free Energy and Gradient\u003c/h2\u003e \u003cp\u003eA Gaussian prior is placed on theta: p(theta)\u0026thinsp;=\u0026thinsp;N(m0, s0^2). Belief updating minimises variational free energy F (Eq.\u0026nbsp;2):\u003c/p\u003e \u003cp\u003eF(q, h, c)\u0026thinsp;=\u0026thinsp;KL[q(theta)||p(theta)] - E_q[log p(h|theta,c)]\u003c/p\u003e \u003cp\u003eWith fixed Sigma, F depends only on mu. The gradient is (Eq.\u0026nbsp;3):\u003c/p\u003e \u003cp\u003edF/dmu = (mu - m0)/s0^2 - alpha * E_q[h - sigma(alpha*theta\u0026thinsp;+\u0026thinsp;gamma*c)]\u003c/p\u003e \u003cp\u003eThe second term is a moral prediction error that drives mu upward when harm is unexpectedly frequent.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Drift, Learning Rate, and Epistemic Volatility\u003c/h2\u003e \u003cp\u003eGradient flow on F yields deterministic dynamics which, with added Langevin noise, becomes (Eq.\u0026nbsp;5):\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003ed(mu)/dt\u0026thinsp;=\u0026thinsp;eta * g(mu, c; d)\u0026thinsp;+\u0026thinsp;xi * noise(t) where g(mu, c; d) := - dF/dmu\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWhere eta\u0026thinsp;\u0026gt;\u0026thinsp;0 is the learning rate, xi controls volatility, and d is an asymmetry trait.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Action Selection and Expected Free Energy\u003c/h2\u003e \u003cp\u003eActions are selected by softmax over expected free energies. We use a pragmatic (outcome-dependent) loss function:\u003c/p\u003e \u003cp\u003eL(Action\u0026thinsp;=\u0026thinsp;Antisocial) = -(12\u0026thinsp;+\u0026thinsp;6 * tanh(d * c))\u003c/p\u003e \u003cp\u003ePositive 'd' makes antisocial choices more sensitive to high-sanction contexts. The advantage of Antisocial (A) over Social (S) action is (Eq.\u0026nbsp;6):\u003c/p\u003e \u003cp\u003eDelta Q = [8\u0026thinsp;+\u0026thinsp;6 * tanh(d*c)] * sigma(alpha*mu\u0026thinsp;+\u0026thinsp;gamma*c\u0026thinsp;+\u0026thinsp;d)\u003c/p\u003e \u003cp\u003eThe probability of choosing the antisocial action is given by the sigmoid function with inverse temperature beta (Eq.\u0026nbsp;7):\u003c/p\u003e \u003cp\u003ePr(Action\u0026thinsp;=\u0026thinsp;A)\u0026thinsp;=\u0026thinsp;sigma(beta * Delta Q)\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Deterministic Dynamics \u0026amp; Catastrophes\u003c/h2\u003e \u003cp\u003eIn the deterministic limit (xi\u0026thinsp;=\u0026thinsp;0), fixed points satisfy g(mu)\u0026thinsp;=\u0026thinsp;0. Varying context (c) and trait (d) jointly yields a cusp catastrophe surface (Thom, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e1975\u003c/span\u003e). Hysteresis emerges naturally: gradual increase in sanction strength keeps the agent on the high-antisocial branch until the upper fold vanishes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Critical Slowing Down\u003c/h2\u003e \u003cp\u003eThe corresponding Fokker-Planck equation governs probability flow. Near folds, drift magnitude approaches 0, producing critical slowing down. Synthetic data (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) show the diagnostic signature: slow reaction times peaking at sudden flips.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Evolutionary Lock-In\u003c/h2\u003e \u003cp\u003eTraits evolve on a slower timescale. Slow-fast analysis shows that populations can become trapped in regions of trait space where only antisocial attractors remain accessible, even if institutional change would otherwise permit escape.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Clinical Implications\u003c/h2\u003e \u003cp\u003eLow eta captures reduced sensitivity to punitive feedback seen in psychopathy (Blair, \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). Negative 'd' aligns with callous-unemotional traits that reward norm violation under peer influence (Haidt's harm/care foundation). The model predicts wider hysteresis and pronounced RT peaks in these individuals.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Conclusion and Future Directions\u003c/h2\u003e \u003cp\u003eThis framework offers a novel, mathematically grounded explanation for the rigidity of antisocial phenotypes. By framing moral belief updating as a dynamical system, we account for the 'stickiness' of maladaptive behaviors through hysteresis rather than simple reinforcement failure.\u003c/p\u003e \u003cp\u003eFuture work will extend this minimal scalar model by treating precision dynamics (Sigma) as an evolving state variable rather than a fixed parameter. Furthermore, empirical studies using ramping-sanction moral-dilemma tasks are required to validate the predicted twin-peak reaction time signatures.\u003c/p\u003e \u003c/div\u003e"},{"header":"Statements and Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of interest/Competing interests\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;The author declares that there are no conflicts of interest or competing interests related to this work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and material\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;This article is based on theoretical and computational modelling and does not report empirical datasets. Any simulated data used to illustrate the model are available from the author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode availability\u003c/strong\u003e\u003cbr\u003ePrototype code implementing the active-inference model and catastrophe analysis used in \u003cem\u003e“Catastrophes in Moral Belief: An Active-Inference Account of Hysteresis and Lock-In in Antisocial Behaviour”\u003c/em\u003e is available from the author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors’ contributions\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;The sole author was responsible for the conception and design of the study, development of the active-inference framework, mathematical and computational analysis, interpretation of results, and writing and revision of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;This work is a theoretical/computational modelling study and does not involve human participants, identifiable personal data, or experiments on animals. Therefore, ethics committee approval was not required.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to participate\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;Not applicable. No human participants were involved in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;Not applicable. The manuscript does not contain any individual person’s data in any form (including images, case details, or quotations) that would require consent for publication.\u003c/p\u003e\n\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eAs the sole author, I conducted all aspects of the study. This included formulation of the research question and theoretical framework; design, implementation, and validation of the active inference\u0026ndash;based generative models and diffusion-model pipelines for synthetic multimodal (behavioral, clinical, neuroimaging) data; execution of all benchmarking and model-recovery analyses; statistical evaluation of fidelity between synthetic and empirical datasets; interpretation of results; and preparation of the manuscript in its entirety.The work relied heavily on established methods from computational psychiatry, Bayesian modeling, and generative deep learning literature. No claim of exceptional novelty or technical virtuosity is made; the contribution consists primarily in the systematic integration and rigorous validation of these existing tools for the specific purpose of addressing data bottlenecks in the field.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eCode is available at: https://github.com/regio-abreu/moral-catastrophes-active-inference\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eBlair, R. J. R. (2007). The amygdala and ventromedial prefrontal cortex in morality and psychopathy. Trends in Cognitive Sciences, 11(9), 387-392.\u003c/li\u003e\n \u003cli\u003eCrockett, M. J., et al. (2013). Models of morality. Trends in Cognitive Sciences, 17(8), 363-366.\u003c/li\u003e\n \u003cli\u003eThom, R. (1975). Structural stability and morphogenesis. Reading, MA: WA Benjamin.\u003c/li\u003e\n \u003cli\u003eFriston, K., et al. (2017). Active inference: A process theory. Neural Computation, 29(1), 1-49.\u003c/li\u003e\n \u003cli\u003eScheffer, M., et al. (2009). Early-warning signals for critical transitions. Nature, 461(7260), 53-59.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","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":"","lastPublishedDoi":"10.21203/rs.3.rs-8448048/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8448048/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eMany antisocial phenotypes exhibit abrupt onset, remarkable stability once established, and striking resistance to intervention. Using active inference as a foundational framework, this paper develops a minimal scalar model of moral belief updating in which saddle-node bifurcations and cusp catastrophes naturally emerge from the geometry of variational free energy. The resulting drift field gives rise to path-dependent moral regimes, critical slowing down at individually variable tipping points, and evolutionary lock-in of antisocial traits. All results are derived analytically or obtained from reproducible simulations.\u003c/p\u003e","manuscriptTitle":"Catastrophes in Moral Belief: An Active-Inference Account of Hysteresis and Lock-In in Antisocial Behaviour","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-26 04:39:52","doi":"10.21203/rs.3.rs-8448048/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":"3682d2cb-fade-4f3a-b6f6-99d62b1a8633","owner":[],"postedDate":"December 26th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-11T15:26:08+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-26 04:39:52","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8448048","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8448048","identity":"rs-8448048","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}