A conceptual framework published in Frontiers in Psychiatry models formal thought disorder not as a collection of static, isolated communicative deficits, but as disruptions occurring across continuous, temporally extended speech trajectories. Authored by researcher Tianlai Vincent Tang of the Nashville Brain Institute, the Hypothesis and Theory paper introduces Dynamic Expressive Flow Theory, abbreviated as DEFT. The framework addresses a long-standing challenge in clinical psychiatry: while clinical practice possesses an extensive descriptive vocabulary for thought, language, and communication abnormalities, the boundaries between clinical signs frequently overlap, and the precise organizational mechanisms generating disordered discourse remain poorly resolved. Rather than evaluating spoken language solely through discrete diagnostic categories or aggregate coherence metrics, DEFT proposes a formal state-space architecture where ongoing expression is treated as an organized pathway navigating through state, transition, and whole-path properties over time.
Formal thought disorder has held a central position in the clinical phenomenology of schizophrenia since early descriptions of disordered association, gaining substantial operational clarity through standardized criteria such as Nancy Andreasen’s Scale for the Assessment of Thought, Language, and Communication. However, empirical investigations demonstrate that these communicative disturbances are transdiagnostic, appearing across bipolar mania, major depression, and other psychiatric presentations rather than remaining confined to schizophrenia alone. Furthermore, the conventional clinical categorization of thought disorder into positive forms, such as derailment, tangentiality, and incoherence, and negative forms, such as poverty of speech and poverty of content, distinguishes polarity of verbal output rather than underlying operations. Highly abundant speech can stem from unconstrained generativity, deficient pruning, or unstable topic switching, whereas impoverished speech can result from diminished generative capacity, excessive constraint, or motor slowing. DEFT approaches this complexity by decomposing surface classifications into underlying dynamic trajectory dimensions.
Cognitive and computational approaches to formal thought disorder have expanded rapidly, yet an explanatory gap persists between fine-grained computational metrics and observable psychopathology. Semantic network theories implicate abnormal spreading activation or weakened conceptual boundaries, while executive models emphasize deficits in working memory, sequencing, and context maintenance. Modern computational psychiatry increasingly applies natural language processing, vector embeddings, and speech-graph algorithms to quantify semantic coherence, structural fragmentation, topic progression, and pause dynamics. Nevertheless, pairwise semantic similarity scores cannot discern whether a departure from a prompt represents creative elaboration, circumstantial overinclusion, or pathological tangentiality, nor can speech graphs alone distinguish adaptive integration from pathological over-linkage without evaluating communicative direction. Tang argues that contemporary psychiatry does not lack empirical data or descriptive terms, but rather an intermediate organizational language that articulates how locally meaningful verbal units form coherent, goal-oriented trajectories that adaptively transition and recover from perturbation.
To construct this intermediate organizational language, DEFT defines a minimal two-dimensional organizational state architecture grounded in systems neuroscience complexity theory. The core state coordinates are differentiation and integration, tracked across discourse time. Differentiation represents the capacity of an expressive communicative system to generate and maintain distinct representational alternatives without prematurely collapsing them into a single state, providing narrative branching, novelty, and elaboration. Integration represents the complementary structural capacity to organize these differentiated elements into coherent relational hierarchies, connecting neighboring clauses into sentences and subordinate ideas into overarching communicative objectives. While complexity theory established differentiation and integration as complementary constraints within distributed neural networks, DEFT adapts these principles specifically to the timescale of connected human discourse. In this context, units of analysis are communicative utterances and topic segments rather than interacting populations of biological neurons.
Metastability operates as a conceptually distinct dynamical property within the DEFT architecture, characterizing the transition regime that governs how organizational states transform over time. In coordination dynamics, metastability denotes a functional operating regime wherein transient assemblies remain sufficiently stable to execute an ongoing communicative task while retaining the critical flexibility to release, reorganize, and switch states. Within expressive discourse, adaptive metastability allows a speaker to maintain focus on an intended conversational topic without succumbing to rigid capture, transition fluently between related conceptual domains, and recover thematic orientation after external interruption. DEFT formalizes stability as the persistence of an active state configuration, transitionability as the operational capacity to vacate that state, and metastability as their balanced coexistence. When metastability is compromised, discourse dynamics can deteriorate into pathological fixation or sudden, uncoordinated communicative collapse.
Beyond the core state properties of differentiation and integration and the transition property of metastability, DEFT posits three higher-order trajectory descriptors: continuity, direction, and prioritization. Continuity represents the temporal preservation of meaningful linkage across successive states, distinguishing temporal narrative coherence from instantaneous structural integration. Direction denotes the target-relative vector alignment of an utterance path, evaluating how directly an expressive sequence progresses toward an intended communicative objective. Prioritization describes the dynamic, context-dependent weighting assigned to competing representational alternatives based on goals, internal states, and environmental context. Within the formal conceptual state-space formulation, the organizational state evolves under external context, cognitive load, internal state, and residual noise, regulated by the metastable transition regime. The resulting expressive flow trajectory exhibits continuity, direction, and prioritization as emergent geometric features rather than redundant primitive coordinates.
This dynamic geometry allows DEFT to formulate precise, partially dissociable quantitative predictions for classical psychiatric signs that are frequently conflated under general labels like disorganization or dysexecutive speech. In tangentiality, the framework hypothesizes a distinct local-global dissociation: local continuity between adjacent utterances remains intact, yet global directional alignment deteriorates, causing speech to drift progressively away from the conversational target. In contrast, circumstantiality is modeled as preserved continuity and ultimate convergence accompanied by impaired directional efficiency, where insufficient pruning produces excessive branching and prolonged trajectory length before the target is reached. Derailment, or loose associations, represents an explicit fragmentation of local continuity, characterized by low adjacent semantic coherence and unstable transition structures. Rather than treating these signs as identical manifestations of cognitive loosening, DEFT predicts that automated trajectory analyses will demonstrate distinct, quantifiable geometric profiles separating them.
The theory extends these geometric distinctions across the broader spectrum of thought, language, and communication pathology. Incoherence is conceptualized as a severe, multiscale integrative failure wherein local semantic linkage, syntactic organization, and overarching discourse structure dissolve simultaneously, requiring multivariate combinations of grammatical, graph-based, and semantic metrics rather than simple pairwise similarity to capture clinician ratings. Flight of ideas is modeled as a state where heightened generative differentiation outruns directional steering constraints, generating rapid branching and switching alongside an unstable goal vector. Conversely, poverty of speech and content occupies the low-differentiation quadrant, characterized by short trajectories, reduced novelty, and sparse branching. For perseveration, DEFT predicts attractor trapping characterized by excessive recurrence, prolonged dwell time, and deficient transition diversity, whereas thought blocking is modeled as an abrupt failure of trajectory persistence and metastable maintenance marked by sudden termination and altered recovery dynamics.
From a neurobiological standpoint, DEFT does not attempt to localize differentiation, integration, or metastability to isolated, dedicated brain structures. Instead, the framework maintains compatibility with systems neuroscience models of large-scale, distributed network coordination. Clinical thought disorder correlates across functional neuroimaging literature with distributed structural and functional alterations spanning frontal, temporal, executive-control, default-mode, and salience networks. DEFT hypothesizes that clinical speech phenotypes reflect time-varying dysconnectivity and coordination failures among generative default-mode networks, goal-maintaining executive networks, and switching-oriented salience networks. For computational operationalization, the framework suggests assembling multiscale feature pipelines using semantic embeddings to calculate prompt distances, graph-theoretic metrics to evaluate recurrence and fragmentation, topic models to assess drift, and acoustic measures to record pauses, avoiding reliance on single, uncontextualized summary statistics.
To illustrate how these theoretical dimensions translate into empirical investigation, the author presents synthetic clause-level language models comparing simulated responses to standard prompts. In a modeled response to why friendship is important, a simulated tangential trajectory demonstrates relatively preserved adjacent continuity alongside progressively rising target distance across successive statements, with re-entry following a distal branch after a neutral examiner interruption. A simulated derailment trajectory exhibits sharp drops in adjacent linkage down to zero point nineteen, reflecting unstable transitions rather than gradual drift. DEFT specifies an initial tier of eight falsifiable empirical predictions testing phenotypic correspondence against validated clinical rating scales, predicting that multidimensional trajectory models will significantly outperform isolated coherence measures, and that open-ended narrative tasks will selectively unmask latent trajectory instability compared to structured conversational interviews.
Crucially, Tang emphasizes that matching clinician ratings on instruments like the Scale for the Assessment of Thought, Language, and Communication demonstrates phenotypic correspondence rather than proving the underlying theoretical architecture. To rigorously evaluate DEFT against competing attractor, executive, or hierarchical generative accounts, the paper proposes two advanced experimental validation strategies. Prediction Nine establishes a within-person experimental perturbation paradigm that manipulates dual-task cognitive load, interruption frequency, and prompt structure independently of psychiatric diagnosis. Under this design, baseline transition-regime indicators measured in separate sessions must predict the magnitude of induced trajectory divergence, dwell time, and re-entry latency in held-out trials beyond working memory, speech rate, and baseline severity. Prediction Ten establishes a cross-task test requiring that differentiation, integration, and transition metrics measured during one conversational task predict independently measured continuity, direction, and prioritization during an entirely different communicative task.
The author explicitly underscores the formal boundaries and methodological limitations of DEFT, stating clearly that it is a conceptual and theoretical framework rather than an implemented speech-production engine, validated software diagnostic pipeline, or commercial rating tool. The publication contains no original patient data, clinical speech recordings, or empirical validations, presenting only prospective synthetic examples and analytical workflows. DEFT must confront the problem of equifinality, wherein indistinguishable expressive phenotypes can emerge from disparate underlying factors, including sedation, linguistic background, general cognitive ability, educational attainment, medication effects, negative symptoms, or cultural discourse conventions. Furthermore, because overt verbal expression reflects social motivation and conversational strategy rather than an unmediated reflection of private thought, trajectory measurements reflect communicative organization rather than internal cognition. Tang concludes that DEFT must undergo rigorous prospective revision or abandonment if simpler semantic models match its out-of-sample predictions or its constructs fail reliability criteria.
Subject of Research: Dynamic Expressive Flow Theory for formal thought disorder
Article Title: Formal thought disorder as disturbed expressive trajectories: a dynamic organizational framework
Article References: Tang, T. V. (2026). Formal thought disorder as disturbed expressive trajectories: a dynamic organizational framework. Frontiers in Psychiatry, 17, Article 1965298. https://doi.org/10.3389/fpsyt.2026.1965298
Image Credits: AI Generated
DOI: 10.3389/fpsyt.2026.1965298
Keywords: formal thought disorder, Dynamic Expressive Flow Theory, computational psychiatry, speech trajectories, differentiation, integration, metastability, tangentiality, circumstantiality, derailment, perseveration, thought blocking, discourse coherence, psychopathology, schizophrenia research, dynamical systems, automated language analysis
News Source: Glenn Wilkins. (November 21, 2026). Theoretical Framework Models Formal Thought Disorder as Disrupted Speech Trajectories. Scienmag.



