Why the World Looks Different When You Are Hungry

Go into a grocery store an hour past lunch and the aisles rearrange themselves. Packaging seems brighter, smells arrive sooner, and items you walk past every week suddenly ask for attention. Nothing about the store changed, and nothing about your eyes changed either. What shifted was the state of your body, and there is a growing research literature arguing that this is not a quirk at the edges of perception but a clue about how perception works at all.

The starting point is predictive processing, a framework in which the brain is treated as an inference machine rather than a receiver. On this account, perception is the brain's ongoing attempt to work out the most likely causes of the sensory signals reaching it, using prior expectations that get corrected by mismatch, or prediction error (Seth, 2013; Ficco et al., 2021). A meta-analysis of imaging studies by Ficco and colleagues (2021) supported the existence of a distributed predictive network rather than a single prediction module, which matters because it means prediction is not one job handled by one place.

Interoceptive inference applies the same logic inward. Interoception refers to the sensing, integration, and interpretation of signals from inside the body, including autonomic, visceral, hormonal, and immune information (Chen et al., 2021). Extending predictive processing to this stream means the brain is also guessing what its own interior should be doing next, then checking the guess against what the viscera actually report (Seth & Friston, 2016).

Here is the claim that reorganizes the picture. The brain does not appear to construct a picture of the outside world first and then paint bodily feeling onto it afterward. Instead, exteroceptive and interoceptive predictions are updated within what looks like a single generative model, a unified internal model of the world and of the body situated inside that world (Dimakou et al., 2025; Seth & Friston, 2016). Interoceptive predictions are one stream inside broader multimodal predictions about being an embodied self, not a separate department reporting upward.

Traffic in this system runs both ways. Predictions shape how incoming signals are interpreted, while errors either revise the model or prompt action that reduces the mismatch (Barrett & Simmons, 2015; Dimakou et al., 2025). For bodily predictions there is a third option that has no obvious perceptual equivalent: an interoceptive prediction error can be settled by an autonomic reflex, meaning the body changes to match the prediction (Seth & Friston, 2016; Berntson & Khalsa, 2021). Regulation, on this reading, is not downstream of perceiving. It is part of the same operation.

That helps explain several things that otherwise sit awkwardly in perception research. Affective tone, meaning the pleasantness or unpleasantness and the level of arousal running underneath ordinary experience, becomes a structural feature of conscious content rather than a mood layered on top (Barrett & Simmons, 2015; Seth & Friston, 2016). Personal significance follows from the same source, because what gets prioritized is weighted by homeostatic and allostatic relevance, meaning relevance to keeping the body regulated now and anticipating what it will need shortly (Quadt et al., 2018). The hungry shopper is not misperceiving the store. Her model is doing exactly what it is built to do.

Where this happens in the brain is less settled than the theory. Park and Tallon-Baudry (2014) argued against a single convergence point where inner and outer signals finally meet, favoring distributed coordination across networks with recurrent exchange and possibly oscillatory synchrony holding visceral and sensory processing together. The insula and neighboring agranular or visceromotor regions come up repeatedly as hubs, since they help issue interoceptive predictions and appear to influence exteroceptive processing elsewhere (Barrett & Simmons, 2015; Seth, 2013; Ficco et al., 2021). Larger networks tied to allostasis, including default mode and salience systems, may do broader integrative work across sensing, acting, and navigating (Chanes & García-Cabezas, 2026).

Not everything the brain predicts becomes conscious, which raises the harder question of selection. Precision weighting is one proposed answer: predictions and errors estimated as reliable gain temporary influence across networks, and that influence may be what admits content to experience (Clark, 2019). Seth and Friston (2016) suggested that higher-level predictions are likelier candidates for conscious content than raw prediction errors. Park and Tallon-Baudry (2014) proposed that bodily signals supply a subjective frame, an anchoring first-person point of view, while carefully noting it is not sufficient on its own.

The caveats are substantial and worth holding onto. Researchers broadly accept that interoceptive and exteroceptive integration occurs, but where and how it is implemented in humans remains genuinely open (Quigley et al., 2021). Measurement is a real constraint: Desmedt and colleagues (2023) reviewed newer interoceptive accuracy measures and questioned whether accuracy is even the most theoretically useful dimension to keep chasing. Teufel and Fletcher (2020) complicate the architecture further, arguing that predictive information can be embedded in bottom-up streams, so the tidy top-down hierarchy is probably incomplete.

Two more qualifications deserve space. Pennartz and colleagues (2019) emphasized that consciousness likely requires integration, dynamic richness, and stability across body and world, not predictive machinery by itself. Cea (2026) argued that conscious bodily selfhood may demand deeper, context-sensitive control than unconscious allostasis can provide, which pushes back on any assumption that regulation automatically produces experience. Prediction seems necessary in these accounts; nobody in this corpus has shown it is sufficient.

What I take from all this is a small shift with long reach. If experience is built from world-predictions and body-predictions updated together, then the feeling of a room, a conversation, or a decision was never a neutral reading with emotion added later. It arrived already weighted by what your body was doing and expecting. That does not mean feelings are illusions or that hunger explains your preferences, but it does suggest that asking what state you were in is a legitimate question about perception, not a change of subject.

References

Barrett, L. F., & Simmons, K. (2015). Interoceptive predictions in the brain. Nature Reviews Neuroscience, 16, 419-429. https://doi.org/10.1038/nrn3950

Berntson, G., & Khalsa, S. (2021). Neural circuits of interoception. Trends in Neurosciences, 44(1), 17-28. https://doi.org/10.1016/j.tins.2020.09.011

Cea, I. (2026). From insentient allostasis to adaptive bodily selfhood: Conscious vs unconscious instrumental interoceptive inference. Adaptive Behavior. https://doi.org/10.1177/10597123261441534

Chanes, L., & García-Cabezas, M. Á. (2026). The possibilities of conscious experience in light of the dual origin hypothesis of the neocortex. Neuroscience of Consciousness, 2025. https://doi.org/10.1093/nc/niaf058

Chen, W. G., Schloesser, D., Arensdorf, A. M., Simmons, J. M., Cui, C., Valentino, R., Gnadt, J., Nielsen, L., St Hillaire-Clarke, C., Spruance, V. M., Horowitz, T., Vallejo, Y., & Langevin, H. (2021). The emerging science of interoception: Sensing, integrating, interpreting, and regulating signals within the self. Trends in Neurosciences, 44, 3-16. https://doi.org/10.1016/j.tins.2020.10.007

Clark, A. (2019). Consciousness as generative entanglement. The Journal of Philosophy. https://doi.org/10.5840/jphil20191161241

Desmedt, O., Luminet, O., Walentynowicz, M., & Corneille, O. (2023). The new measures of interoceptive accuracy: A systematic review and assessment. Neuroscience and Biobehavioral Reviews, 105388. https://doi.org/10.1016/j.neubiorev.2023.105388

Dimakou, A., Pezzulo, G., Zangrossi, A., & Corbetta, M. (2025). The predictive nature of spontaneous brain activity across scales and species. Neuron. https://doi.org/10.1016/j.neuron.2025.02.009

Ficco, L., Mancuso, L., Manuello, J., Teneggi, A., Liloia, D., Duca, S., Costa, T., Kovács, G., & Cauda, F. (2021). Disentangling predictive processing in the brain: A meta-analytic study in favour of a predictive network. Scientific Reports, 11. https://doi.org/10.1038/s41598-021-95603-5

Park, H.-D., & Tallon-Baudry, C. (2014). The neural subjective frame: From bodily signals to perceptual consciousness. Philosophical Transactions of the Royal Society B: Biological Sciences, 369. https://doi.org/10.1098/rstb.2013.0208

Pennartz, C., Farisco, M., & Evers, K. (2019). Indicators and criteria of consciousness in animals and intelligent machines: An inside-out approach. Frontiers in Systems Neuroscience, 13. https://doi.org/10.3389/fnsys.2019.00025

Quadt, L., Critchley, H., & Garfinkel, S. (2018). The neurobiology of interoception in health and disease. Annals of the New York Academy of Sciences, 1428. https://doi.org/10.1111/nyas.13915

Quigley, K., Kanoski, S., Grill, W., Barrett, L. F., & Tsakiris, M. (2021). Functions of interoception: From energy regulation to experience of the self. Trends in Neurosciences, 44, 29-38. https://doi.org/10.1016/j.tins.2020.09.008

Seth, A. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive Sciences, 17(11), 565-573. https://doi.org/10.1016/j.tics.2013.09.007

Seth, A., & Friston, K. J. (2016). Active interoceptive inference and the emotional brain. Philosophical Transactions of the Royal Society B: Biological Sciences, 371. https://doi.org/10.1098/rstb.2016.0007