When someone reaches for a cup, they do not ordinarily calculate where their hand ends or ask whether the moving arm belongs to them. Their experience presents a body, a point of view, and a possible action as belonging together. A self-model is one proposed way to explain that organization: the nervous system maintains and updates a representation of the organism in relation to its surroundings. The term names a theoretical model, not a little person inside the head, and not a complete explanation of what a person is.
A model of oneself, not a second self
In cognitive science, a model is a structured representation that helps an organism interpret incoming information and guide action. A self-model may include bodily location, ownership, action possibilities, internal states, goals, and expectations about other people. Different theories include different elements. Some focus on the immediate embodied point of view; others address the autobiographical person who can describe a life across time. The concept is useful only when the author specifies which level is being modeled and what evidence would bear on it.
Metzinger’s self-model theory of subjectivity is a prominent philosophical proposal. It argues that the experience of being a self depends on a representational model that is ordinarily transparent: people experience the apparent world and themselves in it, not the construction of a model. This is a framework for interpreting consciousness, not an established laboratory fact that settles metaphysics. One can use the term self-model more modestly to describe testable representations of body and action without adopting every claim of that theory.
A self-model should not be confused with self-concept. Self-concept often refers to beliefs and descriptions such as “I am patient” or “I am a teacher.” A body-related model can organize reaching, balance, and ownership without any verbal description. It is also distinct from narrative identity, the evolving story by which a person connects past and future. These levels can interact. A person who becomes disabled may need to update bodily expectations, daily plans, and life story, but changes in one level do not prove identical changes in the others.
Gallagher’s distinction between minimal and narrative self helps mark the boundary. Minimal self concerns the immediate first-person character of experience; narrative self concerns continuity and interpretation over time. A self-model can be proposed at either scale, but the evidence used to study one does not automatically explain the other. A brief laboratory illusion of hand ownership is an observation about bodily experience. It cannot tell us how someone understands their decades of relationships or the moral commitments they choose to keep.
Body ownership and multisensory evidence
Botvinick and Cohen’s rubber-hand experiment showed that coordinated visual and tactile stimulation could induce a feeling that a visible artificial hand belonged to the participant. This finding matters because it reveals that body ownership is not simply a fixed reading of anatomy. The brain integrates signals about what is seen and felt under conditions of timing and spatial alignment. A self-model account can interpret the illusion as an update in the represented boundary of the body.
The illusion has limits. Participants did not acquire an actual new hand or lose their ordinary identity. A reported feeling of ownership is not the same as every aspect of bodily control or selfhood. Experimental variants differ in procedures and measures, and a laboratory induction does not show that people in daily life are confused about their bodies. The careful inference is that the experienced body can be altered in a controlled context, providing evidence about the flexibility of one component of self-representation.
Body ownership also differs from agency. Ownership asks whether a sensation or body part feels like mine; agency asks whether I experience myself as causing an action. A person can feel that their hand belongs to them while being moved passively by someone else. They can also act through a tool or avatar and feel some control without treating it as literally part of their body. Conflating these components makes experiments hard to interpret. A theory of self-modeling should say whether it predicts ownership, agency, self-location, or some combination.
In ordinary activity, these representations are rarely isolated. The weight of a cup, the visible position of the fingers, and the expected sensory consequences of reaching all help guide correction. If the cup is heavier than expected, the action changes. The adjustment does not require a conscious sentence about the self. This makes self-modeling relevant to action, but one should not infer from an example of sensorimotor correction that a specific grand theory of consciousness has been confirmed. Different mechanisms can produce adaptive movement.
Internal signals and predictive accounts
Interoception concerns the processing of bodily signals such as heartbeat, breathing, temperature, and visceral change. Seth proposed an interoceptive-inference account in which predictions and incoming bodily signals contribute to emotion and embodied self-experience. This is a theoretically rich proposal linking body and feeling. It is not evidence that every emotion is a simple prediction error or that selfhood can be reduced to one bodily variable. It should be read as a model with testable implications and acknowledged alternatives.
A person entering a public speaking event might notice a racing heart and trembling hands. Those sensations become part of their experience of being in that situation. They may interpret them as fear, excitement, exertion, or illness depending on context. The example illustrates why a self-model may need to integrate bodily state with goals and surroundings. It does not make a single interpretation true. Physiological measures, self-reports, and observed action each capture different pieces of the episode.
Predictive accounts often describe perception and action as a cycle of expectation and correction. A model can help anticipate what a movement will feel like, then update when the actual signal differs. This is a useful explanatory vocabulary, but its breadth can become a problem: if every possible outcome is redescribed as a prediction update, the account risks becoming difficult to falsify. The scientific value comes from specifying the predicted contrast between conditions, the measure, and the alternative explanation that the experiment can exclude.
Changes in attention to bodily signals can also change what is reported. Someone trained to notice the breath may describe more detail, but more detail is not automatically greater objective accuracy. Mindfulness and interoception studies do not directly establish a general self-model theory. A meditation-related change in self-report could reflect attention, vocabulary, demand characteristics, emotional appraisal, or some combination. Such a finding should not be presented as proof that practice transforms a unified self-model.
What the theory helps us ask
The self-model concept can organize questions across several levels. What sensory evidence supports the experience that a limb is mine? What signals contribute to the feeling that I initiated an action? How is bodily state integrated with a goal? How do people update expectations after injury, prosthetic use, or virtual reality? These are separate research questions. A broad model can suggest connections among them, but each connection needs its own operational definition and evidence.
Virtual avatars and prosthetic devices make the distinction practical. A person may control an avatar accurately without feeling that the avatar’s body is their own. Another may experience some ownership under synchronized visual and tactile cues while still recognizing it as a simulation. Designers should measure usability, agency, comfort, and ownership separately instead of treating a single “presence” score as the entire experience. The rubber-hand finding motivates research questions, but does not by itself validate a device or predict an individual user response.
Clinical situations require even more caution. Changes in bodily experience can occur with neurological illness, pain, injury, medication, or psychiatric distress. A self-model framework may help formulate research, but it is not a diagnostic instrument. A person’s report should be taken seriously without forcing it into one theory. Assessment needs clinical context, differential explanation, and the person’s own goals. The presence of a body illusion in an experiment does not mean ordinary changes in self-experience are pathological.
There are philosophical disagreements too. Some accounts treat the self as a constructed representation; others argue that first-person experience cannot be captured fully as an internal model. Some emphasize embodied interaction rather than representation, and some separate the practical organism from the phenomenology of selfhood. These disagreements are not merely verbal. They influence what researchers measure and how they interpret the same result. The term self-model should identify a proposal under examination, not announce that all competing accounts have been defeated.
Contemplative claims and their limits
Dahl, Lutz, and Davidson discuss meditation practices that may reconstruct or deconstruct aspects of self-related processing. Their framework is useful for distinguishing attentional, constructive, and deconstructive approaches, but it is not evidence that any practice literally removes a self or reliably produces a specific metaphysical insight. Meditation traditions differ, and scientific tasks often isolate only a narrow outcome. The reported experience of reduced self-focus should be described on its own terms rather than converted into a claim about the permanent structure of personhood.
A person may notice that a recurring self-critical thought is a thought rather than an unquestionable fact. That change can be meaningful without proving that the body model, narrative identity, and first-person perspective have all changed together. A trial measuring self-report, attention, or symptoms would support claims about those measures, not a total transformation of self. Public communication should preserve the distinction between phenomenological description, theoretical interpretation, and demonstrated outcome.
The Nordic and European research context can be relevant when specific laboratories, cohorts, or clinical systems contribute evidence. Geography itself is not a mechanism. A concept should not receive a regional label merely because contemplative or neuroscience work is conducted there. Translation is also delicate: a Swedish word for self, identity, or body awareness may not preserve the same theoretical distinction. Any translation into Swedish should be checked against disciplinary usage rather than presented as a casual equivalent.
How to evaluate a self-model claim
Ask first what component of self is at issue. A study of an ownership illusion should report the induction, comparison, and outcome measures. A study of agency should distinguish intended action from observed effect. A study of narrative identity should examine stories and time, not assume that a hand illusion indexes life continuity. Then ask what alternative explanation remains: suggestibility, task demand, attention, sensory conflict, or a real change in represented ownership. A single result may support a narrow mechanism while leaving the broad framework open.
Cross-study convergence is valuable when different methods point to the same constrained claim. A body-ownership finding, neural measurement, and behavioral adaptation may together strengthen an account of multisensory integration. They still may not establish the nature of phenomenal consciousness. This layered approach avoids both extremes: dismissing a model because it cannot solve every philosophical question, and declaring it proved because one experiment produces a striking illusion.
For everyday use, the concept can encourage humility about apparently immediate self-knowledge. The body and situation are interpreted through multiple signals, and interpretations can change with new evidence. It can also encourage respect for a person’s lived experience without treating it as a complete physiological explanation. If someone says a prosthesis feels more integrated after training, that report matters; measurements of function and context matter too. A careful self-model account can hold these forms of evidence together without collapsing them.
What the evidence does not show
The evidence does not show that there is a literal miniature self inside the brain, that one neural model explains all identity, or that an ownership illusion proves every philosophical claim about consciousness. Metzinger’s and Seth’s accounts are theories with distinct commitments. A rubber-hand effect concerns a limited bodily experience, while narrative identity concerns a life interpreted over time. Mindfulness or interoception findings do not establish that a unified self-model has been rebuilt or dissolved. Clinical and contemplative claims need their own carefully defined outcomes.
A self-model is best treated as a research tool for asking how an organism represents its body, actions, and point of view. It can connect compelling experiments with broader questions while keeping their evidential limits clear. The person reaching for the cup is not consulting an inner portrait; they are using coordinated signals and expectations to act. Explaining that coordination is a scientific task. Deciding what it means to be a self remains a wider question with several live answers.
Research references
- Metzinger, T. (2007). Self models. Scholarpedia, 2(10), 4174. View source
- Gallagher, S. (2000). Philosophical conceptions of the self: Implications for cognitive science. Trends in Cognitive Sciences, 4(1), 14–21. View source
- Botvinick, M., & Cohen, J. (1998). Rubber hands ‘feel’ touch that eyes see. Nature, 391, 756. View source
- Seth, A. K. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive Sciences, 17(11), 565–573. View source
- Dahl, C. J., Lutz, A., & Davidson, R. J. (2015). Reconstructing and deconstructing the self: Cognitive mechanisms in meditation practice. Trends in Cognitive Sciences, 19(9), 515–523. View source