Self, Others and Context: Reframing Clinical Reasoning as a Socio‐Cognitive Practice Authors: Thierry Pelaccia, Peter Wyer, Jonathan Sherbino, Geoff Norman Published: 2026 ABSTRACT ABSTRACTRationale, Aims and ObjectivesClinical reasoning is often studied as if it occurred inside the clinician's head. Yet, in practice, clinicians think, decide and act through constant interaction with patients, colleagues, technologies, routines and local constraints. This article challenges predominantly individual and decontextualised accounts of clinical reasoning and reframes reasoning as a situated socio‐cognitive practice. We aim to show how explicit and tacit knowledge interact with the social, material and organizational environments in which clinical practice unfolds.MethodsWe develop a conceptual synthesis drawing on research in clinical reasoning, situated and distributed cognition, ecological psychology, workplace learning and knowledge translation. Tacit knowledge, particularly know‐how and know‐who, is used as a bridge between individual cognitive processes and the healthcare ecosystem. Distributed cognition, affordances, communities of practice and mindlines are mobilised as complementary analytical lenses.ResultsThis synthesis conceptualises clinical reasoning as an emergent capability arising from interactions among Self, Others and Context. Expertise depends not only on what clinicians know, but also on their ability to perceive what matters, mobilise the right resources, judge whom to trust, and adapt action to the situation at hand. This perspective helps explain why expert performance may deteriorate across unfamiliar settings and why artificial intelligence should be understood not as a substitute for reasoning, but as a cognitive artefact whose value depends on clinicians' capacity to critically interpret and contextualise its outputs. It also highlights the importance of authentic participation, cognitive apprenticeship and socially supported reflection in developing clinical expertise.ConclusionsClinical reasoning is neither an isolated mental process nor a simple response to context. It is an accomplishment distributed across clinicians, other people, artefacts and environments. Reframing it in this way shifts attention from how clinicians think in isolation to how reasoning actually works in practice, with important implications for clinical work, education, artificial intelligence and research.Keywords: artificial intelligence, clinical reasoning, cognition, medical education Introduction For more than half a century, the understanding of clinical reasoning within the field of cognitive science and medical education has been largely shaped by models derived from cognitive psychology, which describe it as a primarily individual and relatively decontextualised mental activity [1, 2]. Classical approaches have given particular prominence to dual‐process theory, distinguishing between a fast, intuitive System 1 and a slower, analytical, and deliberative System 2 [3]. Within this framework, the clinician is conceptualised as an information processor whose performance depends chiefly on the retrieval and application of stored knowledge held in long‐term memory [1, 4]. These models have undeniably advanced our understanding of core cognitive mechanisms. However, they have often relegated the environmental and social dimensions of practice to peripheral variables, viewed mainly as potential sources of diagnostic error or performance distortion [1, 5]. From a rationalist and objectivist standpoint, context is treated as noise to be minimised or controlled in order to optimise decision‐making. Such a ‘vacuum‐sealed’ view of clinical reasoning overlooks the systemic complexity within which medical decisions are actually made in authentic clinical settings. From a broader perspective, particularly from the perspective of the social sciences, clinical reasoning cannot be separated from the healthcare ecosystem in which it unfolds. This ecosystem is woven from human interactions, organizational constraints and technological mediations that substantially shape decision‐making [1, 5]. Theories of situated cognition suggest that reasoning—like learning—is not confined to an individual's brain, but an emergent property of the ongoing, dynamic interaction between the individual (‘Self’), those with whom they engage (‘Others’) and their surrounding environment (‘Context’) [6, 7]. As Penner et al. note [5], ‘This interdependence between cognition and context creates a bidirectional relationship: context shapes the content of a case and cognition, which goes on to further shape the context, which then influences the subsequent content and cognition’. Contextual and social factors thus cease to be mere sources of variability and instead become constitutive elements of reasoning and decision‐making. Viewed through this lens, the issue is no longer solely how a physician thinks in isolation, but how reasoning is constructed in and through situated action [6, 7]. Therefore, clinical reasoning is not simply an individual cognitive performance, but a meaning‐making process shaped by the environment in which it is enacted. This shift grounds clinical reasoning in a socio‐cognitive epistemological framework, acknowledging that mental processes are inextricably bound to the physical and social fabric of the clinical encounter. Sociologists, anthropologists, historians of medicine and scholars working within social constructivist traditions have long examined medical knowledge and practice as socially constituted [8, 9]. Likewise, an important body of work in health professions education has drawn on situated cognition, distributed cognition, ecological psychology and sociocultural theory to illuminate the contextual nature of clinical reasoning [7, 10, 11, 12]. In this article, we seek to bring these latter perspectives into dialogue with the clinical reasoning literature by showing how tacit knowledge—i.e. knowledge acquired through experience and embodied in action [13, 14]—may provide a practical bridge between individual knowledge structures and the collective, material and organizational conditions in which reasoning occurs. Our own perspective is, necessarily, grounded in our background as cognitive‐science‐oriented researchers and educators. We hope to inspire others with similar backgrounds and orientations to broaden and enrich their perspectives in these ways. Along the way, we use distributed cognition, affordances, communities of practice and mindlines as complementary lenses rather than as an exhaustive or definitive theoretical taxonomy. Together, they help explain how clinical knowledge is enacted, shared, transformed and stabilised in practice. Tacit Knowledge as the Foundation of Situated Clinical Expertise From Explicit Knowledge to Know‐How and Know‐Who Tacit knowledge is traditionally contrasted with explicit knowledge, defined as objectifiable, formalised, and ‘transmissible’ knowledge recorded in textbooks, research articles, databases, clinical guidelines or procedural manuals. Tacit knowledge, by contrast, refers to a body of experientially‐grounded knowledge that is difficult to codify or articulate, and inseparable from the situations in which it acquires meaning [15]. Its etymological origin, from the Latin tacitus (‘silent’ or ‘unspoken’), underscores its implicit nature. As Michael Polanyi famously observed, ‘we know more than we can tell’ [15]. This form of knowing, enacted in practice, rests on representations, beliefs, values and habits shaped through engagement in authentic contexts and apprenticeship [13]. Expertise relies on the integration of both explicit and tacit forms of knowledge. Explicit knowledge is often captured by the concepts of know‐what and know‐why. Know‐what refers to factual knowledge, such as recognising the symptoms of a disease. Know‐why involves understanding explanatory mechanisms and causal relationships, such as the pathophysiological basis of a condition or the mechanism of action of a medication. Tacit knowledge is expressed through know‐how and know‐who. Know‐how refers to the capacity to act effectively in a given situation. It is not reducible to the mechanical application of an algorithm. Rather, it involves the ability to perceive what is salient, to recognise when and under what circumstances action is warranted, to adjust one's actions in real time, and to respond appropriately to contextual constraints and situational nuances [13]. Drawing upon explicit knowledge, it transcends it by embedding it within lived experience. It is expressed, for example, in the way a clinician adapts a therapeutic education consultation to a patient's reactions, or modulates the execution of a clinical procedure in response to evolving conditions. Know‐who, in turn, introduces an explicitly relational dimension, since it refers to a clinician's knowledge about their colleagues [16]. It informs decision‐making and action by incorporating judgements about others' competencies, working styles and perceived reliability into clinical reasoning. This knowledge is therefore inseparable from the organizational context and human resources within which clinical work unfolds. Tacit Knowledge as a Resource for Clinical Expertise Clinical reasoning relies heavily on tacit knowledge [17]. Research conducted in domains beyond healthcare suggests that such knowledge may even constitute a central marker of expertise, sometimes predicting performance more effectively than explicit knowledge alone [18]. Far from being contextual noise that distorts cognitive purity, tacit knowledge acts as a powerful modulator of data interpretation and action, shaping performance in meaningful ways. Although empirical research in healthcare remains limited, studies show, for instance, that expert emergency physicians rely extensively on know‐who to weigh and filter information communicated by other health professionals based on perceived reliability, and to optimise team‐based patient management [16]. Similarly, emergency triage nurses draw on their knowledge of colleagues to assign patients, tailor handovers and anticipate care needs [19]. Community pharmacists have also been shown to consider their knowledge of prescribers when dispensing antibiotics [20]. These findings illustrate how tacit knowledge operates as a distributed resource, embedded within social interactions and organizational routines. Tacit Knowledge, Recognition‐Primed Decision Making and Dual‐Process Theory This view of tacit knowledge as shaping judgements resonates with the naturalistic decision‐making tradition, particularly Klein's recognition‐primed decision model [21, 22]. In this perspective, experienced professionals do not necessarily generate and compare multiple options before acting. Rather, they recognise a situation as familiar, drawing on accumulated experience and using their tacit knowledge to identify relevant cues, plausible goals, expectancies, and a likely course of action. This option may then be mentally modified if its anticipated consequences reveal a mismatch with the evolving situation. Klein explicitly described this process as combining an intuitive, recognition‐based component with a more conscious and deliberate component of mental simulation, thereby echoing the distinction commonly drawn in dual‐process theories between System 1 and System 2. The relationship between tacit knowledge and dual‐process theory warrants clarification to avoid a common conceptual conflation. Because it shares several attributes with descriptions of System 1 (intuitive) processing—being implicit, contextualised and grounded in experience—it would naturally be tempting to associate tacit knowledge with System 1 and explicit knowledge with System 2. Yet, these constructs belong to different epistemological registers. Tacit knowledge contributes to the formation of the ‘fuel’ of reasoning by shaping the prototypes, scripts and exemplars upon which reasoning draws; System 1 and System 2 constitute the cognitive processes that serve as the ‘engine’ through which this fuel is used [23]. Know‐how and know‐who may inform both intuitive recognition and analytical verification, rather than belonging exclusively to one mode of processing. For instance, an emergency physician may intuitively prioritise a patient after a nurse simply remarks ‘he doesn't look right’, drawing on prior experience of that nurse's clinical judgement (know‐who), or narrow the focus of the clinical assessment based on subtle behavioural or situational cues (know‐how). At later stages, the same forms of knowledge may be re‐engaged to analytically weigh the reliability of information provided by different team members when reconciling conflicting data (know‐who), or systematically adapt the sequence and interpretation of clinical findings in light of contextual constraints, such as time pressure or patient cooperation (know‐how). Tacit knowledge therefore shapes the repertoire of cues, expectations and relational judgements available to clinicians across different modes of reasoning. It may support rapid recognition, but it may also be mobilised during slower processes of verification. We therefore use dual‐process terminology cautiously. It helps describe cognitive operations involved in clinical reasoning, but it does not, by itself, explain how clinicians acquire, organise and use the knowledge required to act effectively in particular settings. A socio‐cognitive account complements and qualifies dual‐process models by shifting attention from processing modes alone to the conditions under which those modes become effective. Contextual Adroitness and the Situated Nature of Expertise This context‐dependence of clinical reasoning has observable consequences for expert performance. An expert clinician may experience a decline in diagnostic accuracy when moving to a new facility, even if the medical problems remain the same [5]. This is not due to a lack of medical knowledge, but to reduced ‘contextual adroitness’, i.e. a lack of background familiarity with the new environment's specific routines and physical configurations [24]. Benner's adaptation of the Dreyfus model is useful in this respect, highlighting a shift from reliance on context‐free rules toward a more situated appreciation of recurrent patterns and salient cues [25]. In a familiar setting, the clinician is deeply integrated into the social and physical fabric of the ward, allowing situational cues to operate in the background of action. In an unfamiliar context, these same elements instead demand conscious, focal attention, thereby increasing cognitive load and disrupting the rapid, intuitive flow of System 1 reasoning. Tacit knowledge thus enables clinicians to navigate the complexity of healthcare settings not by mechanically applying abstract rules, but by ‘dwelling’ within a network of embodied reference points so familiar that they operate in the background of action [15]. In the next section, we examine how socio‐cognitive theories of clinical reasoning help account for the social and material conditions through which such knowledge is developed and enacted. Theoretical Frameworks Supporting Reasoning as a Socio‐cognitive Act To move beyond viewing clinical reasoning as an isolated mental act, we ground it in theoretical foundations that conceive knowledge as distributed across physical, organizational and social environments [26, 27]. Distributed Cognition In distributed cognition, the unit of analysis in clinical reasoning shifts from the individual clinician's mind to the functional cognitive system in which reasoning is accomplished [28, 29]. Such a system is defined by the functional relationships between its components. Cognitive processes emerge from the dynamic interplay among actors (physicians, other health professionals and patients) and their surroundings (the clinical setting, time constraints, workload, institutional norms, etc.) [1, 28, 30]. This includes a complex interplay with cognitive artefacts—tools, technologies, guidelines, procedures—as well as spatial arrangements and institutional routines [27, 28]. From this standpoint, knowledge is neither purely internal nor purely external. It is a flow that circulates across actors and material supports through processes of coordination. This perspective recognises that clinical knowledge cannot be fully contained within a single individual. Reasoning is therefore accomplished through the coordinated relations among clinicians, patients, artefacts, representations, routines and local organizational arrangements. External resources help configure what information becomes available, salient, communicable and actionable in a given situation [31, 32]. To understand the specific mechanisms through which this social distribution and contextual integration occur in clinical practice, additional theoretical constructs are particularly helpful. Affordances Originally conceptualised by James J. Gibson and later applied to workplace learning by Stephen Billett, affordances refer to the action possibilities offered by an environment to those who inhabit it [33, 34, 35, 36]. In healthcare settings, affordances may include salient clinical cues, access to diagnostic tests, opportunities to consult experienced colleagues, a collaborative culture or simply adequate time. From a distributed cognition perspective, such affordances also include cognitive artefacts, i.e. electronic health records, diagnostic algorithms, clinical scoring systems, imaging technologies and guidelines. These opportunities, however, do not automatically produce effects. Their impact depends on the practitioner's ability to perceive, interpret, and act upon them, drawing on knowledge, values, prior experience, interests and agency [34, 37]. This ability is called effectivity [37, 38, 39]. In clinical practice, effectivity may be expressed in recognising meaningful signs, interpreting test results, making productive use of medical records or engaging a colleague at the appropriate moment. From this perspective, tacit knowledge represents the clinical expression of this capacity for engagement. Know‐how and know‐who can be understood as complementary manifestations of effectivity. Know‐how concerns the ability to perceive relevant clinical and technical affordances and to mobilise them appropriately in context (e.g. adjusting resuscitative treatment based on patient response or structuring a consultation). Know‐who shapes access to relational and organizational affordances by influencing how information from others is weighted and how available human resources are engaged (e.g. knowing whom to consult, calibrating the credibility of a colleague's opinion, anticipating a team member's response or allocating responsibilities). Through these tacit forms of engagement, reasoning unfolds at the interface between individual dispositions and environmental possibilities. This dynamic—reducible neither to pure cognition nor to mere context—supports the emergence of coherent care trajectories across time and contexts [26, 34, 37]. Communities of Practice The work of Jean Lave and Etienne Wenger adds a crucial collective dimension. Clinical reasoning is not a skill acquired in isolation; it is co‐constructed through social participation in a community of practice [40]. Novices learn to generate hypotheses, prioritise data and justify decisions by observing, imitating and dialoguing with more experienced members of the group. Lave's legitimate peripheral participation describes the process through which newcomers gradually enter a community of practice, initially at the margins, with limited responsibility, and progressively moving toward more central roles. Participation is ‘legitimate’ because it is recognised and authorised by the group, and ‘peripheral’ because it begins within bounded spaces of responsibility. Wenger subsequently developed this perspective by emphasizing learning as participation in social practice and the interdependence of practice, meaning, community, and identity. As involvement deepens, reasoning becomes aligned with the shared languages, norms, priorities and interpretive frameworks of the community. A compelling example of this transition is seen in the supported independence of residents navigating clinical uncertainty. Ilgen et al. found that while operating in a liminal space at the borders of their abilities, novices move toward central participation by ‘broadcasting’ their evolving thinking and ‘borrowing comfort’ from their supervisors [41]. This negotiated partnership functions as a clinical safety net, allowing learners to engage with the community's tacit knowledge and situational complexities while gradually internalising the professional standards required for expert practice. This gradual participation transforms both competence and professional identity [6, 35, 42]. Immersion in the community's habits, thinking, discourses and evaluative standards teaches the novice not only what to think, but how to think as a clinician. Clinical reasoning thus becomes inseparable from professional socialization: it is shaped by collective practice even as it reinforces belonging to the group. Tacit knowledge acquires particular meaning within this collective dynamic. It develops through participation in shared practices. Know‐who, especially, presupposes a nuanced understanding of colleagues and their contributions to distributed reasoning. It reveals the intrinsically relational character of clinical judgement, incorporating perceptions of reliability, competence and habitual practice as constitutive elements of reasoning itself. Clinical reasoning therefore emerges from relational processes—case discussions, handovers, multidisciplinary meetings and the collaborative process of shared sense‐making. This ‘thinking together’ is a genuine social phenomenon, described by some as ‘social breathing’, which corresponds to the collective responsibility and action through which health professionals continually adjust their reasoning during patient care to make sense of uncertain situations [43]. It is within this living coordination—responsive to context and interaction—that the complexity of clinical reasoning unfolds, well beyond the individual application of rules or isolated knowledge. Mindlines The concept of mindlines, introduced by John Gabbay and Andrée le May, offers a framework for understanding how knowledge is constructed, shared and mobilised in clinical practice [24, 44]. Formal guidelines provide explicit, evidence‐based recommendations that are designed to be applied across settings, but are often experienced as rigid or insufficiently attuned to local realities. By contrast, mindlines refer to dynamically constructed, socially shared and partly internalised norms of practice that guide action pragmatically [24, 45]. Individually and collectively, they serve to integrate available evidence with experiential and contextual knowledge, as well as with local constraints, making them directly usable within specific professional settings. They are continually reshaped through interaction, experience, negotiation and engagement with formal and informal sources of knowledge. Clinicians rarely apply research findings and scientific recommendations mechanistically. Over time, fragments of formal guidelines, local protocols, research summaries, decision‐making tools and educational materials are appropriated and blended with experiential and contextual knowledge drawn from interactions with peers, shared reflection, clinical anecdotes, patient encounters and dialogue, case narratives, and local habits and specificities, such as the socioeconomic characteristics of the patient population, access to local diagnostic resources, and the practical realities of the clinical setting [44, 46, 47]. Mindlines are the outcome of this transformation. Anchored in context and socially situated, they function as socially stabilised clinical scripts that guide everyday decision‐making in environments characterised by uncertainty, time pressure and complexity. They may therefore be understood as a dynamic social process through which tacit and explicit forms of knowledge are continuously brought into play within communities of practice [24]. Mindlines integrate formal knowledge and guidance with know‐how—in the form of experience‐stabilised ways of acting—and know‐who—in the form of knowledge about others, incorporated into clinical judgement. They sustain everyday reasoning as a relational practice in which, for example, the reliability of a colleague's input is weighed alongside clinical data in diagnostic inference. In this sense, mindlines constitute a collective epistemic layer through which encountered affordances and shared experiences are translated into locally meaningful ways of interpreting situations and guiding action. Questions concerning the nature of knowledge are central to the mindlines model. From its inception, the model challenged the assumption that clinically useful knowledge can be detached from social practice [24, 44]. Synthesising the Theoretical Constructs The four concepts developed in this section operate at different, yet mutually supportive, levels of analysis. They offer complementary lenses on the same phenomenon: how clinical knowledge becomes usable through participation in social and material systems of practice. Distributed cognition provides the broadest analytical frame, i.e. it conceives clinical reasoning as an accomplishment of a functional system extending across clinicians, patients, artefacts, representations, routines and time. Communities of practice provide the social and developmental conditions within which clinicians participate in such systems, acquire shared norms, develop professional identities and progressively cultivate a sense of what counts as relevant, credible or appropriate in a particular setting. Within these distributed systems, affordances refer to the possibilities for perception and action made available by the physical, technological, organizational and interpersonal environment. Their practical significance depends on clinicians' effectivities, including the tacit capacities captured by know‐how and know‐who. Know‐how supports the recognition and use of clinical, technical and organizational affordances; know‐who supports the appraisal and mobilization of relational affordances, including judgements about whose input to seek, trust, question or integrate. Through repeated participation in these distributed systems, clinicians collectively develop mindlines, i.e. dynamic, internalised and socially shared practical understandings that bring together formal evidence, experience, local routines and relationships. They are both products of collective practice and resources that subsequently orient future participation within it. These concepts should not be treated as interchangeable theories. They differ in their perspectival emphases, units of analysis and epistemological and ontological assumptions. Our use of these perspectives is therefore selective and aims to provide complementary accounts of the social and material conditions of reasoning, and to help specify how clinicians access, evaluate, and enact knowledge within those conditions. Implications for Practice, Education and Research Redefining clinical reasoning as a situated and socially distributed practice calls for a critical reexamination of the models that currently shape clinical work, clinical‐reasoning education and research in this domain. Implications for Practice: The Age of Artificial Intelligence Contemporary medical practice is increasingly shaped by the rapid integration of artificial intelligence (AI). Recent research highlights a clinically important performance gap: although large language models (LLMs) may perform impressively on decontextualised medical benchmarks and standardised diagnostic tasks, this performance does not automatically translate into improved reasoning or decision making in situated practice [48, 49]. LLMs are highly effective at identifying patterns in large datasets and generating probabilistic predictions based on the information available to them. However, this information is, by definition, restricted to what is explicitly provided, either through the prompt or as embedded in their training data. The tacit, relational and context‐dependent information that often underpins clinical reasoning may not be reliably represented in the data available to such systems. Their performance may therefore be less dependable in situations where the interpretation of information depends heavily on local relationships, evolving circumstances or forms of practical knowledge that are difficult to formalise. We believe that AI tools should be understood as a particular class of cognitive artefact, thus constituting new affordances within the healthcare environment. While they share the distributed nature of conventional decision‐support systems, electronic health records (EHRs) or checklists, AI tools fundamentally differ by acting as probabilistic, generative artefacts rather than static, rule‐based procedural guides. Within a sociotechnical framework, they can create dynamic human‐machine symbioses where the value of AI is conditional upon the clinician's effectivity—specifically, the capacity to critically appraise, contextualise and actively negotiate algorithmic outputs [49, 50]. Without such expert mediation, AI may encourage forms of cognitive off‐loading and a superficial, ‘recipe‐based’ approach to care, whereby users delegate aspects of information retrieval, interpretation or judgement to an external system. This may be beneficial when it reduces unnecessary cognitive burden while preserving active clinical engagement. However, concerns have been raised that repeated and uncritical reliance on generative AI could produce a form of ‘cognitive debt’, in which immediate gains in efficiency are achieved at the expense of the development, rehearsal, and maintenance of independent judgement and cognitive skills [51]. The empirical basis for this concern remains preliminary and largely non‐clinical; it should therefore be understood as a plausible educational and professional risk rather than an established consequence of AI use. The issue may be especially salient for novice clinicians. If early‐career practitioners rely prematurely on AI outputs without first internalising domain‐specific knowledge, they may fail to develop the expertise required for safe and effective practice. A similar issue was observed in the early days of evidence‐based medicine, when some clinicians misapplied research findings due to insufficient background knowledge. The phenomenon can also be observed with algorithm‐assisted ECG interpretation. While such systems can reliably detect well‐defined patterns—such as ST‐segment elevation based on predefined criteria—they may fail to identify more subtle or atypical presentations that require contextual interpretation. Experienced clinicians can recognise these patterns by integrating tacit and contextual knowledge, whereas novice practitioners may over‐rely on algorithmic outputs. Furthermore, AI is currently incapable of modelling ‘social breathing’, as previously defined in the context of communities of practice, nor can it reproduce the relational depth through which clinical meaning is constructed. It does not participate in the collective dynamics through which decisions acquire their meaning and significance. Accordingly, practice must evolve toward models in which AI is incorporated into distributed‐reasoning processes, under the supervision of reflective clinicians capable of translating algorithmic outputs through their socially constructed and context‐sensitive mindlines. Moreover, as Wyer et al. argue [52], even the most robust clinical recommendations remain conditional upon social norms and local constraints. Implications for Education: From Schooling to Guided Participation in Context We have long argued that many errors arise not primarily from flawed cognitive processing (System 1 vs. System 2), but from insufficient access to the situation‐specific experiential knowledge required to act effectively in context [3, 23, 53, 54]. Education in clinical reasoning must therefore move beyond the myth of transferable, decontextualised thinking skills and instead be anchored in authentic clinical situations, where knowledge becomes meaningful through engagement with particular patients, colleagues, artefacts, routines and organizational constraints [1, 3, 23, 53]. This perspective shifts the educational goal from exposure to cases toward progressively fuller participation in the distributed systems through which care is accomplished. Mere presence in the workplace is insufficient: learners must be granted legitimate opportunities to observe, contribute, assume responsibility and engage with the clinical and relational affordances available to them, in order to recognise the social and relational foundations of clinical reasoning. Guided participation and cognitive apprenticeship can support this process through modelling, coaching, scaffolding, articulation, reflection and progressively reduced supervision [26, 41, 55, 56]. Supervisors can make selected dimensions of practice accessible by explaining how they interpreted information from particular colleagues, used or questioned available artefacts, responded to organizational constraints and adjusted their judgement as the situation evolved. Such exchanges may help learners develop illness scripts, know‐how, know‐who, and context‐sensitive mindlines required to participate effectively in clinical practice. Reflection embedded in clinical supervision can help learners derive greater benefit from lived experience by creating opportunities to reconstruct and reinterpret significant encounters [57, 58]. Following such an encounter, supervisors may invite learners to revisit how the situation unfolded, articulate the hypotheses they considered, compare alternative interpretations, and examine the findings that supported or contradicted each possibility. Structured or deliberate reflection offers one possible framework for organising this diagnostic component of the conversation [59, 60]. Supervisors may also draw attention to how interactions with patients and colleagues, available artefacts, temporal pressures and local routines influenced what became salient and actionable. In this sense, reflection becomes a socially supported process of revisiting participation rather than an isolated exercise in introspection. It may bring selected aspects of tacit practice into focal awareness, connect new encounters with previous ones, and contribute to the progressive refinement of illness scripts and mindlines. Simulation can play an important complementary role in this educational agenda. Full‐scale simulation, high‐tech simulation and virtual reality can provide repeated, structured and relatively safe opportunities to engage with uncertainty and improve clinical reasoning [61, 62]. However, these environments serve as a bridge to—rather than a replacement for—the complex socialization, contextual dexterity and tacit learning that can only be fully developed through authentic participation within real‐world communities of practice. Implications for Research: From In Vitro to In Vivo Research on clinical reasoning will benefit from and be enriched by a shift toward studying reasoning in vivo [1]. Importantly, several of the perspectives mobilised in this article have themselves been informed by ethnographic enquiry into professional practice: communities of practice emerged from studies of learning through participation in social practice, including Wenger's ethnographic study of claims processing in a medical insurance company, while the concept of mindlines was developed through ethnographic study of knowledge use in primary care and hospital setting [40]. Experimental paradigms have contributed substantially to our understanding of cognitive mechanisms over recent decades [2], and doubtless will continue to provide the foundation for interpreting the results of other types of empirical research. Ultimately, artificially isolating contextual variables and focusing predominantly on individual cognitive performance risks oversimplifying a phenomenon that is inherently relational and situated [1]. As Olson et al. argue [1]: ‘studies must no longer try to remove individuals from context to seek “truth” – instead, we must seek to observe and understand what happens in the often “messy” place where it happens; that is, “reality”’. Embracing in vivo research does not imply abandoning the theoretical contributions of experimental cognitive science. Decades of work on dual‐process theories and cognitive load have generated robust explanatory models of individual cognitive mechanisms. These frameworks remain indispensable, not as competing paradigms, but as complementary lenses for interpreting how distributed reasoning operates within real clinical systems. Methodologically, this entails complementing in vitro approaches with in vivo investigations capable of capturing collective dynamics, affordances and the processes through which mindlines stabilise in practice. Experimental findings can illuminate the cognitive processes mobilised within distributed systems, while in vivo research reveals how these processes are shaped, scaffolded and transformed by social and material environments. Such integration preserves theoretical rigour while enhancing ecological validity, moving the field toward a more comprehensive account of clinical expertise. Methodologies from the social sciences offer promising prospects in this regard [1, 63, 64]. More specifically, ethnography is a qualitative methodology based on sustained engagement with a field setting, aimed at understanding practices as they unfold in everyday life and from participants' perspectives [32]. It combines complementary sources of data—such as observation, interviews, documents and artefacts—with iterative interpretation and researcher reflexivity. Applied to clinical reasoning, ethnography can trace how information and decisions circulate across clinicians, patients, artefacts, routines and time. It is particularly suited to examining reasoning as a socially and materially mediated accomplishment, rather than as a purely individual mental process. In this perspective, researchers must define the boundaries of the cognitive system under study, determine which actors and artefacts to follow, account for their own role within the setting, and avoid treating context as an undifferentiated residual category [27, 32]. Future studies should therefore specify which social, material, temporal and organizational processes are hypothesised to shape reasoning, while combining close‐grained observation with theoretically informed analysis. Conclusion In this article, we have proposed redefining clinical reasoning as a situated and socially distributed practice, continuously shaped by the interaction among individuals, environments and forms of knowledge. By drawing on situated cognition, affordances, communities of practice and mindlines, we argue that expertise depends not only on the accumulation of explicit knowledge, but also on its integration with tacit, collectively‐developed, understandings that enable authentic practice. Clinical reasoning thus stands as an emergent socio‐cognitive capability, inseparable from the collective, contextual and practice‐based conditions that make it possible. The implications are substantial. For clinical practice, the rise of artificial intelligence reinforces the need for human judgement capable of interpreting data and supervising its application within context. For education, it calls for learning environments grounded in authentic clinical situations, supported by progressive participation in communities of practice and cognitive apprenticeship. For research, it encourages the development of in vivo approaches capable of capturing the relational and contextual dynamics of reasoning in practice. Understanding clinical reasoning as a situated phenomenon does not weaken scientific rigour. Rather, it repositions rigour within the concrete conditions of its application. Author Contributions All authors jointly discussed the structure and content of the article. Thierry Pelaccia wrote a first version of the article, which was critically reviewed by the other team members. All authors approved the final version of the manuscript. Funding The authors have nothing to report. Ethics Statement The authors have nothing to report. Conflicts of Interest The authors declare no conflicts of interest. Acknowledgements Open access publication funding provided by COUPERIN CY26. Data Availability Statement The authors have nothing to report. References Associated Data Data Availability Statement The authors have nothing to report.