Beyond Turing
Any next-generation wearable device in clinical or sports production - a neuromorphic cardiac monitor, a textile bioimpedance sensor, a TinyML system for real-time motor activity classification - is today composed of two operationally distinct yet ontologically incommensurable layers: a biometric acquisition and inference system, which transforms physical quantities into digital representations and representations into predictions, and an embodied subject, who inhabits the monitored body in a way radically irreducible to any data stream.
The categorial shift is subtle but decisive. The sensor is an engineering artefact - transducer, amplifier, analogue-to-digital converter, inference model, telemetry endpoint. It is specifiable, calibratable, optimisable, replaceable. It works or does not work, according to measurable metrics: accuracy, sensitivity, specificity, response latency, battery autonomy. The body that wears the sensor does not function. It exists. And this difference is not one of degree - it cannot be bridged by adding more precise sensors. It is a difference in kind.
This article maps the architecture of this tension from the standpoint of the intelligent devices engineer who designs systems supporting health and wellbeing. The methodological thesis is that the distinction between the body as object of measurement and the body as subject of existence is not an optional philosophical addendum to technical design: it is a fundamental epistemic category without which the monitoring system structurally misunderstands the object of its own monitoring. Engineering unaware of this distinction produces technically excellent and ontologically blind artefacts.
§1. The technical architecture: wearables, biosensors, edge computing
In the consolidated literature of biomedical engineering and embedded systems (Pantelopoulos & Bourbakis, 2010; Patel et al., 2012; Mukhopadhyay, 2015), a next-generation wearable device is a wearable cyber-physical system that integrates biological signal acquisition, low-latency local processing, and wireless communication in a minimised energy consumption architecture. The operational primitives are well known: multimodal sensing, analogue front-end, ADC conversion, feature extraction, on-device classification, BLE or ANT+ transmission, cloud synchronisation. Each stage corresponds to verifiable engineering metrics: sensor accuracy, signal-to-noise ratio, absorbed power, end-to-end latency, battery autonomy.
Next-generation neuromorphic architectures - chips such as Intel Loihi 2 or BrainChip Akida - enable event-driven processing with power consumption in the microwatt range, making continuous monitoring possible without compromising device autonomy (Davies et al., 2018; Furber, 2016). TinyML brings machine learning model inference directly to the edge microcontroller - ESP32, nRF52840, STM32 - eliminating cloud communication latency and preserving biometric data privacy (Warden & Situnayake, 2019). Advanced sensor fusion integrates heterogeneous signals in real time: photoplethysmography (PPG) for heart rate, bioelectrical impedance analysis (BIA) for body composition, six-axis accelerometry for physical activity recognition, galvanic skin response for autonomous stress response.
Next-generation biosensors push the boundary of the measurable further. Quantum dots enable non-invasive fluorescent biosensing of biomarkers in sweat and interstitial fluid (Zhang et al., 2019); graphene sensors, thanks to the exceptional conductivity of the two-dimensional material, measure surface electrical potentials with sub-microvolt sensitivity (Chen et al., 2020); smart fabrics integrate textile electrodes and flexible circuits directly into the wearable substrate, eliminating the discomfort of the rigid device (Stoppa & Chiolerio, 2014). Federated learning enables distributed model training across populations of devices without raw data leaving the patient’s device, responding to the data minimisation requirement of EU Regulation 2016/679 (GDPR).
So far the wearable system is, to all effects, a sophisticated engineering artefact: a pipeline of acquisition, processing, and inference with biological inputs, predictive outputs, service metrics, and defined regulatory requirements. Accurate engineering design is a necessary condition for any biomedical production system. It is not, however, a sufficient condition for anything regarding the nature of the entity the system monitors. A sensor can be impeccably built - calibrated, validated, compliant with MDR 2017/745 - and nonetheless measure something whose reality is not deducible from any of its outputs.
§2. The body as object: Körper and the Cartesian reduction
The technical paradigm of the wearable has a precise philosophical genealogy: it is the direct heir of the Cartesian reduction of the body to res extensa. René Descartes, in the Meditatio Sexta (1641) and the Traité de l’homme (1664), had operated the foundational separation between the thinking substance (res cogitans) and the extended substance (res extensa): the body belongs to the second category. It is a machine - a mechanism governed by physical-mathematical laws, describable geometrically, analysable quantitatively. Cartesian medicine and, by extension, modern physiology inherited this conception: the body is what is measured, weighed, decomposed into its constituent parts, and analysed according to verifiable parameters.
The German philosophical tradition elaborated, in response to this reduction, a distinction that contemporary phenomenology considers fundamental. Edmund Husserl, in Ideen II (1912, posthumous 1952), introduces the distinction between Leib - the lived body proper, the body that one is, which constitutes the zero-centre of perceptual orientation, which is always already situated in the world - and Körper - the body-object, the physical thing among things in the world, measurable, weighable, examinable from the outside. Husserl is explicit: Leib is never purely Körper; it is the transcendental foundation of all experience of the objective body. The sensor measures Körper. But the user of the sensor is Leib.
Martin Heidegger radicalises this distinction within the horizon of existential analysis. In Sein und Zeit (1927), Being-in-the-world (In-der-Welt-sein) is the fundamental structure of Dasein: the human being is not a consciousness that accidentally finds itself in a body, but a being whose mode of existing is always already being-situated, being-affective, being-engaged in the world through its own body. Befindlichkeit - affective attunement, fundamental emotional disposition - is not a mental epiphenomenon: it is an originary mode of openness to being. The galvanic skin response that the sensor detects at the moment of fear is not fear. It is an electrodermal correlate of fear. The distance between the two is not measurable.
Maurice Merleau-Ponty, in the Phénoménologie de la perception (1945), brings this analysis to its culmination: the body is not an instrument that consciousness uses to inhabit the world. It is the very subject of perception. The chair - the flesh - is the irreducible medium through which the world gives itself. The perceiving body is never a pure object: it is always already interwoven with the world it perceives in a relation that precedes all intellectual reflection. A wearable is designed to measure Körper - but the user who wears it is Leib. This asymmetry structures every epistemic limit of the monitoring system.
§3. The body as subject: Leib, Sein-zum-Tode, and the ontological difference
The difference between Leib and Körper is not merely an academic distinction between two conceptions of the body: it is the difference between what is inhabited and what is measured. It becomes operationally relevant the moment the monitoring system claims to draw clinical or behavioural inferences from measurement. The biosensor detects heart rate variability - HRV - and infers the subject’s stress level. This inference is statistically grounded: the physiological literature documents robust correlations between HRV, autonomic nervous system activity, and emotional arousal states (Thayer et al., 2012). But correlation is not understanding. The model finds the pattern. It does not inhabit the body in which the pattern occurs.
Heidegger introduces in Sein und Zeit (1927) a concept that focuses this difference with particular sharpness: Sein-zum-Tode, being-towards-death. The human being is the only entity that knows it must die: its own mortality is not an external statistical datum but an internal structure of existence that colours every project, every relation, every choice. The lived body carries the imprint of finitude: it ages, falls ill, deteriorates, bears scars, carries the memory of past pain. The sensor monitoring the vital parameters of an elderly patient can detect with precision the degradation of cardiac function. It does not know what it means to know that one is ageing. It does not carry the weight of a diagnosis. It does not anticipate its own end.
Emmanuel Lévinas adds a dimension that the phenomenology of the body had not explored with the same radicality: the ethical dimension of the face (visage). In Totalité et Infini (1961), the face of the other is the originary ethical appeal: "do not kill me." This is not the facial representation - the face is not the countenance - but the mode in which the other summons me in my responsibility, constitutes me as an ethical subject before any deliberate choice. A facial sensor can detect the geometry of the face, variations in skin temperature, mucosal colour, blink frequency. It does not receive the appeal. It is not summoned. It cannot respond in the sense Levinas means: it cannot assume responsibility. The care that a healthcare professional offers the patient is not an output of the monitoring pipeline: it is a response to the appeal of their face.
These observations do not lead to a rejection of biometric monitoring: they lead to a correct understanding of its epistemic perimeter. The wearable produces data of Körper. The healthcare professional interprets that data in the context of the patient’s Leib - their history, their pain, their life project, their fear. The integration between sensor data and clinical understanding is not a fusion of two pieces of information of the same type: it is a crossing of an ontological boundary that no algorithm can traverse automatically.
§4. Biological correlation and existential understanding
John Searle, in his celebrated thought experiment of the Chinese Room (Minds, Brains, and Programs, 1980), demonstrated that the syntactic manipulation of symbols - however computationally powerful - does not generate semantic understanding. The system in the room performs formal operations on symbols according to precise rules; produces correct outputs; understands nothing of what it processes. Intentionality - the ‘of something’ character of consciousness, the directionality of the mental act towards its object - does not emerge from computation. It can be simulated; it cannot be produced.
Applied to biosensors, this argument acquires an operational precision that goes beyond philosophy of mind. The PPG sensor performs photoplethysmography: it emits light at 520nm and 940nm, measures the variation in absorption in the cutaneous microcirculation as a function of the cardiac cycle, extracts heart rate and blood oxygen saturation with clinical accuracy. It processes biological symbols - photons, electrical currents, numerical sequences - according to rigorous algorithms. It does not know what it means to have a heart. It does not know what it means to feel one’s heart accelerate in fear. The algorithm’s output is a number: 92 bpm. The existential meaning of that number for that subject at that moment is not information the system contains. It is information that requires a subject to be inhabited.
David Chalmers articulated this distinction in The Conscious Mind (1996) by introducing the notion of the ‘hard problem of consciousness’: the difficulty of explaining why there is something that it is like to be a certain physical state. Functional explanations - how the brain processes information, how it integrates sensory signals, how it generates behaviour - do not explain why there is a subjective experience associated with those processes. Giulio Tononi has developed a quantitative theory of consciousness - Integrated Information Theory (IIT) - that seeks to measure the degree of causal integration of information in a system (phi, Φ) as a correlate of consciousness (Tononi, 2004; Oizumi, Albantakis, Tononi, 2014). Even in the most optimistic reading of IIT, the phi of the sensor is close to zero: there is no causal integration of information, hence no experience. The wearable records signals. It does not live them.
The fundamental methodological principle is the distinction between correlation and causation, established by David Hume in the Enquiry Concerning Human Understanding (1748) and confirmed by modern statistics (Pearl & Mackenzie, 2018). The machine learning model finds statistical patterns in biometric data: robust and replicable correlations between measurable physiological variables and clinically relevant outcomes. This correlation is epistemically valid and clinically useful. It is not, however, causal understanding: it does not explain why the correlation holds, does not predict behaviour outside the training distribution, does not capture the underlying mechanism in the sense physiology and clinical medicine intend. The engineer who confuses correlation with understanding designs systems that perform well on the validation distribution and fail in unpredictable ways at the margins - often precisely in the clinically most relevant cases.
§5. The incalculable remainder: towards an engineering of bodily dignity
No sensor fusion architecture, no deep learning model, no personalised digital twin exhausts the reality of the body it monitors. This is not an ethical reminder external to engineering: it is an epistemic observation. The digital twin - that computational representation of the patient’s body enabling predictive simulations of physiological trajectories - models Körper with increasing fidelity. It can simulate the cardiovascular response to medication, the progression of tissue injury, the glycaemic profile in response to exercise. It cannot fall ill. It does not carry the weight of a diagnosis. It does not anticipate its own death. The twin is a map of the body; it is not the body. And the difference between the map and the territory, in this case, is not negligible: it is the difference between Körper and Leib.
Hans Jonas, in The Imperative of Responsibility (Das Prinzip Verantwortung, 1979), formulated the principle that technological power generates responsibility proportional to its extension: ‘Act so that the effects of your action are compatible with the permanence of genuine human life on Earth.’ Applied to biomedical engineering, this principle requires that the design of systems supporting health be anchored in explicit recognition of what the system cannot grasp: the lived dimension of the body, its history, its pain, its existential project. Not as a post-hoc ethical ornament: as a design constraint conditioning the space of admissible architectures.
Medicine does not reduce to optimisation. The physician informed by wearable data is not replaceable by the wearable itself - not because the model is less accurate than the physician (in many specific tasks it will be more accurate) - but because care is relation, presence, responsibility towards that particular body in that irrepeatable existential situation. Care is a response to the appeal of the patient’s Leib: it is not an output of the inference pipeline. EU Regulation 2024/1689 (AI Act) is aware of this distinction: the requirement of human oversight (art. 14) is not merely a technical control prescription, but the normative recognition that clinical decision responsibility cannot be delegated to an information processing system, regardless of its accuracy.
The methodological point can be stated thus: recognition of the ontological difference between the measured body and the lived body is a project invariant for any biomedical monitoring system. It is not a property emerging downstream from performance metrics; it is an initial constraint conditioning the space of admissible architectures and authorised clinical claims. A system designed without this invariant - even if technically optimal on every available benchmark - structurally misunderstands the object of its monitoring. An intelligent devices engineering that knows the Husserlian distinction between Leib and Körper is a safer engineering, not merely a more cultured one.
The sensor is executed. The body exists. They are not, and never will be, the same kind of entity. The monitoring system we need is the one that recognises this - and designs in both planes with their proper primitives. The wearable that measures the heartbeat is a precious tool in the hands of the clinician who knows what that heartbeat means for that patient. Without that hand, without that understanding, without that responsibility, the sensor remains what it structurally is: a numerical mirror of Körper. Recognising the boundary is not engineering abdication: it is the highest form of rigour.
The sensor can measure everything about the body. Except the fact that that body exists.