7. Multisensory integration: how the nervous system builds an estimate of the body

Knowledge Base – Expert Guide Sensory Flows

Blue Portance Knowledge Base · Non-commercial reproduction permitted with source citation

First of all, the previous chapters presented separately several sources of information available to the nervous system: proprioceptionProprioceptionProprioception is the body’s ability to perceive the position, movement and effort of its various parts without needing to look at them.Read the full entry : Proprioception (B3), interoceptionInteroceptionInteroception is the perception of the body's internal signals: heartbeat, breathing, hunger, thirst, temperature, the state of the internal organs. It informs the nervous system about the physiological state of the body.Read the full entry : Interoception (B4), touch and support (B5), as well as afferents likely to come from different innervated tissues, including certain connective tissues described in B6. None of these sources, taken in isolation, provides a complete description of the state of the body.

More precisely, this chapter deals with how these sources, once available, are combined to produce an estimate the nervous system can use. However, it does not deal with how this combination adjusts over time according to the relative reliability of the sources. That distinct mechanism is presented in B8. Nor does it deal with how movement itself generates new information, presented in B9.

The brain does not receive a complete map of the body. It builds an estimate of its state from partial signals, related to one another and integrated within a given context (Ernst & Banks, 2002 ; Chiba et al., 2016).

1. No source provides a complete copy of the body’s state

To begin with, the previous chapters described several sources of bodily information: mechanical signals transformed into afferent activity (B2), proprioceptive information (B3), interoceptive signals (B4), tactile and haptic contact (B5), and the possible contribution of certain innervated connective tissues (B6). This information does not yet, on its own, constitute bodily perception.

Next, an afferent signal is nerve activity arising from a receptor or a population of receptors. It must then be integrated with other information, compared with the context and related to internal models of the body, according to the chain set out in B2: physical stimulus → tissue deformation → detection by a receptor → afferent activity → central processing → possible perception. Each of these steps has a different function. None should therefore be confused with the next. For example, a pressure map in particular describes a mechanical distribution, not the activity of receptors, their combination, or the sensation or discomfort that may result, as B5 has already established in detail.

This chapter thus extends this chain beyond a single source: afferent signals → central processing and multisensory integration → body representations → sensorimotor responses and, possibly, conscious perceptual experience. This formulation avoids presenting perceptual experience as the necessary, linear outcome of every afferent signal. Indeed, a large part of sensory processing contributes directly to sensorimotor control without ever reaching consciousness.

2. Integrating several sources is not the same as juxtaposing them

In addition, building a bodily estimate draws on several sources — notably visual, vestibular, proprioceptive, tactile and interoceptive — as well as information linked to the motor command. Their integration is not a simple addition. Their contribution depends on context, the task and whether the available signals agree or diverge (Chiba et al., 2016).

For example, some experimental approaches show that combining several sources can reduce uncertainty compared with each of them taken in isolation, notably when integrating visual and haptic information to estimate the properties of an object (Ernst & Banks, 2002). This finding illustrates the general principle of multisensory integration, but it is not a direct demonstration of bodily perception in the sitting position.

Ultimately, the key point is that the signals described separately in the previous chapters do not yet constitute a representation of the body. They are related to the context, the ongoing action and the internal models available. The mechanisms by which the functional weight of the different sources readjusts according to their availability and reliability, as well as the consequences of sensory conflicts, are developed in B8.

Vision, the vestibular system, proprioception, touch and interoception are weighted according to their reliability, the context and the task.
Figure 1 — Integrating does not mean adding Multisensory integration builds a contextualized estimate from partial and uncertain sources. © Blue Portance 2026.

3. Internal models and prediction

Furthermore, sensorimotor control does not rely solely on processing the afferent information received after a movement or a load. It also relies on internal models that allow the nervous system to anticipate the expected sensory consequences of an action, even before the corresponding afferents are available (Franklin & Wolpert, 2011).

Thus, when a person shifts their pelvis, they can anticipate a change in support and in trunk orientation. The afferent information that arrives next can be related to the expected consequences. This organization notably makes it possible to detect a mismatch: the support moves more than expected. The contact may also fail to change in the expected direction. Finally, the support may become less stable than anticipated. The precise mechanisms of this anticipation are, however, still being actively studied and are not developed further here.

Movement command, prediction of sensory consequences, actual feedback and possible prediction error.
Figure 2 — Internal model: expected and observed consequences The nervous system compares the expected consequences of an action with the information actually received. © Blue Portance 2026.

4. The model of the body: the body schema

In particular, the brain has functional representations of the body that make it possible to organize action, orientation and the relationship with the environment. The body schemaBody schemaThe body schema is a sensorimotor representation of the body, largely nonconscious and continuously updated, that makes it possible to locate body segments and organize posture and movement. It is one of the dimensions of body representation.Read the full entry : Body schema can be understood as an operational, relatively dynamic representation of the configuration of the body, its segments and their possibilities for action. It is useful, for example, for orienting a limb toward a target or estimating the position of a segment without looking at it. It also makes it possible to anticipate the consequences of a movement. This representation is not necessarily conscious in the form of a detailed image of the body; it can function as a model that guides action, built from the integration of visual, tactile and proprioceptive information centered on the body parts and the space surrounding them (Maravita, Spence & Driver, 2003).

However, this representation is not perfectly rigid. It can be updated when the body changes position or when sensory information changes. This plasticity does not mean that the brain completely reorganizes the body with every movement, but that its estimates can be adjusted to remain compatible with the information and actions available. A change in support or pelvic position changes the information available about body configuration; how these changes update body representations in sitting nevertheless remains to be established — it would be excessive to conclude that a mobile seat automatically “reprograms” the body schema.

5. Body schema, body image and sense of ownership: concepts to distinguish

Moreover, the body schema should not be confused with body image, which refers to a conscious, subjective and sometimes evaluative representation of the body, nor with the sense of body ownership, the experience of a body part being felt as one’s own. These dimensions are related but not equivalent. Thus, a person can automatically adjust their posture without a precise conscious representation of their body, while a conscious representation can be influenced by emotional and social factors that cannot be reduced to postural controlPostural controlPostural control is the function by which the nervous system controls the body’s position in space, with two objectives: orientation of body segments relative to one another and to the environment, and balance.Read the full entry : Postural control.

Furthermore, the sense of body ownership depends on the integration of visual, tactile, proprioceptive and interoceptive signals; bodily awareness is described as a multidimensional experience that notably includes ownership of the body, self-location and a form of first-person perspective (Blanke, Slater & Serino, 2015). This work is useful for showing that bodily perception results from the integration of several sources. However, it should not be transposed directly to the analysis of a seat. The experimental protocols used in this field — body ownership illusions, specific experimental manipulations — do not demonstrate that an ordinary change of support alters the sense of body ownership.

6. An estimate, not a direct measurement

Indeed, the estimate of body state produced by this integration is not a direct measurement comparable to that of a sensor. It involves uncertainty, can be more or less precise and can remain compatible with several slightly different body states. Some models describe this estimate in probabilistic terms, but this description does not mean that all body representations are necessarily produced by an explicit probabilistic calculation.

Statements to avoid

One should not write that the nervous system “faithfully reconstructs” the position of the body, nor that it has an “exact body map”. The integrated estimate remains a useful, error-prone approximation, not an exact copy of the body’s physical state.

This caution thus extends the warning already given in B5 about pressure mapping. The same logic applies here at a higher level: the estimate integrated by the nervous system is not to be equated with the actual mechanical state of the body either, even when it is a useful approximation of it.

Actual mechanical state, partial sensory signals and bodily estimate with uncertainty.
Figure 3 — A bodily estimate is not a direct measurement The nervous system estimates the state of the body from partial signals; it does not measure it directly. © Blue Portance 2026.

7. Bodily experience is multimodal

In addition, the interoceptive signals described in B4 also take part in this estimate. Breathing, heartbeat, fatigue or the general state of the body can influence how a posture is experienced, through interactions between internal physiological signals, exteroceptive information and the dynamics of brain networks (Candia-Rivera et al., 2024).

For example, a sensation of support can combine skin stimulation, tissue deformation, proprioceptive information, an interoceptive change and an attentional context, without the subject necessarily being able to identify their respective contributions. This inability to consciously break down the sensation does not, however, make mechanical measurements useless: they describe one particular dimension of the phenomenon, not the whole of it.

8. What this chapter does not cover

By contrast, this chapter presents the general principle of multisensory integration and of building a representation of the body. It deliberately leaves two questions to later chapters. How the relative weight of the sources readjusts over time — for example when a source becomes less reliable or changes in availability — is the subject of B8. How the subject’s voluntary movement actively produces new information, rather than passively waiting for a change, is the subject of B9.

9. Application to sitting

In the sitting position, the estimate the nervous system forms of pelvic position, trunk orientation or the distribution of support results from this same logic of integration. It can draw on proprioceptive, tactile and interoceptive information, to which afferents from various innervated tissues may contribute, as well as on internally generated expectations. It cannot be inferred from a single variable, such as pelvic angle, mean pressure or symmetry of support.

Furthermore, the shape, firmness and mobility of the seat can change the mechanical conditions of contact and the associated cutaneous afferent activity. When they also change the configuration of the pelvis and trunk or muscle activity, they can also modify certain proprioceptive information likely to contribute to this estimate. A seat that allows support to vary can give the subject the opportunity to explore a new configuration. By contrast, a highly restrictive seat may limit this exploration. These statements describe plausible mechanisms; they do not demonstrate that a given type of seat will produce the same perception in all users.

Finally, the perception of a seat varies between people according to morphology, skin sensitivity, proprioceptive abilities, state of fatigue, previous experience, the task and expectations. A difference in how it feels therefore does not immediately identify a precise mechanical difference. It may reflect an interaction between the mechanical condition and the way each nervous system integrates it — a direct reminder of the distinction made in section 6 between mechanical measurement and bodily estimate.

10. Key points

  • No sensory source provides, on its own, a complete description of the state of the body; a pressure map describes a mechanical distribution, not a perception.
  • The nervous system relates several sensory sources to the context, the task and the internal models available; the mechanisms for weighting and reweighting these sources are developed in B8.
  • Integrating several sources is not the same as juxtaposing them: afferent signals are related to one another to build an estimate of body state, without constituting an exact copy of it.
  • The estimate of body state also relies on predictive internal models and on a dynamic body schema, distinct from body image and the sense of ownership.
  • This estimate remains uncertain and multimodal — including interoceptive — and must be distinguished from a direct mechanical measurement.
  • The reassessment of the weight of sources over time is the subject of B8; the active role of movement in producing new information is the subject of B9.

In sum, the multisensory integration described here is the link that connects the sources presented in B3 to B6 with the sensorimotor loop set out in B1. How the contribution of these sources varies with their reliability, sensory conflicts and task conditions is developed in B8. Finally, how movement actively enriches this estimate is developed in B9.


Scientific references

Multisensory integration and postural control

  1. Ernst, M. O., & Banks, M. S. (2002). Humans integrate visual and haptic information in a statistically optimal fashion. Nature, 415(6870), 429–433. https://doi.org/10.1038/415429a
  2. Chiba, R., Takakusaki, K., Ota, J., Yozu, A., & Haga, N. (2016). Human upright posture control models based on multisensory inputs; in fast and slow dynamics. Neuroscience Research, 104, 96–104. https://doi.org/10.1016/j.neures.2015.12.002

Internal models and predictive control

  1. Franklin, D. W., & Wolpert, D. M. (2011). Computational mechanisms of sensorimotor control. Neuron, 72(3), 425–442. https://doi.org/10.1016/j.neuron.2011.10.006

Body schema and bodily awareness

  1. Maravita, A., Spence, C., & Driver, J. (2003). Multisensory integration and the body schema: close to hand and within reach. Current Biology, 13(13), R531–R539. https://doi.org/10.1016/S0960-9822(03)00449-4
  2. Tsakiris, M. (2017). The multisensory basis of the self: From body to identity to others. Quarterly Journal of Experimental Psychology, 70(4), 597–609. https://doi.org/10.1080/17470218.2016.1181768
  3. Blanke, O., Slater, M., & Serino, A. (2015). Behavioral, neural, and computational principles of bodily self-consciousness. Neuron, 88(1), 145–166. https://doi.org/10.1016/j.neuron.2015.09.029

Interoceptive contribution

  1. Candia-Rivera, D., Engelen, T., Babo-Rebelo, M., & Salamone, P. C. (2024). Interoception, network physiology and the emergence of bodily self-awareness. Neuroscience & Biobehavioral Reviews, 165, 105864. https://doi.org/10.1016/j.neubiorev.2024.105864
Note : this content aims to explain mechanisms. It is neither a medical diagnosis nor a therapeutic prescription.