Antonio Damasio, Gil B. Carvalho (2013), Nature Reviews Neuroscience 14(2):143–152, Perspectives: Opinion — DOI 10.1038/nrn3403

Key takeaways

  • A feeling is the mental experience of a body state. Feelings are kept apart from drives and emotions, which are “action programmes”: innate, largely stereotyped sets of bodily and cognitive actions that maintain or restore homeostasis (Box 1). An action programme changes the body; the change is mapped interoceptively; the map may or may not be felt.
  • Body states are valenced by construction (good or bad for homeostasis), so a felt map is a proxy for biological value. The evolutionary hypothesis is that felt experience was added to non-conscious body mapping because it makes regulation more flexible: it forces attention and supports learning and prediction.
  • The upper brainstem is the first place interoceptive signals form a whole-body map, and the authors place feeling there as well as in cortex. The evidence is that a patient with both insulae destroyed still feels (their own patient B, reported in a 2012 Cerebral Cortex paper they cite), that children born without cortex behave as if they feel, and that damage to the posterior upper brainstem abolishes sentience while damage to the ventral half (locked-in syndrome) does not. The insula and SI/SII are given a modulatory role: finer discrimination, and access to memory, language and reasoning.
  • Feelings are therefore very likely not exclusive to humans. Since the brainstem is conserved, “the parsimonious assumption should be that feelings are present” in species that have its structures and behave as if they feel.
  • A cellular proposal, offered as such: interoceptive pathways are mostly unmyelinated (about 80% of vagal fibres), and myelin would block ephaptic coupling and ligand binding along the axon. So unmyelinated membranes open to the milieu may be the cellular root of sentience, with feelings arising as that activity is synchronized up to the systems level.

Standing in the record

Filed on 2026-10-01 at the owner’s request for “some Antonio Damasio”. It is held here and not in the anthology: it carries no instruction for machine-learning practice and no anthology topic holds it. Its primary subject is emotion and feeling, which ADR-014 added emotion-and-affect to say.

NOTE-330 is the close reading of 2026-10-01, and it placed the work: Active. It is a ten-page Opinion, and as an account of how feeling arises it is a programme, not a result. Its value is as the compact, citable form of a view the rest of the record argues with, and its lesion evidence against the insula as the seat of feeling is the part most worth keeping.

How it bears on what the record holds:

  • Biological naturalism. Seth’s case that consciousness belongs to living systems (LIT-135) runs through interoception and the regulation of the body, as this paper does. The routes differ. Seth’s goes through predictive processing and the free energy principle. This paper’s goes through topographic body maps and their evolution, and says nothing about prediction or inference beyond the claim that feelings aid anticipation. Read together, they show that the “life and mind” line in the record has a neuroanatomical route that does not depend on the free energy principle.
  • Who has feelings. The parsimony argument for animal feeling is the kind of graded-sentience premise that Terrill and colleagues’ plant paper (LIT-193) builds on, but it ties feeling to a nervous system with interoceptive maps and a brainstem. On its own terms it gives no support to plant sentience. The Damasios’ own criticism of plant sentience, which Terrill and colleagues cite and NOTE-120 reports, is a different work: a 2023 Animal Sentience commentary, “Sensing is a far cry from sentience” (DOI 10.51291/2377-7478.1805), not held here and not read.
  • Bodies and machines. Roberts’ argument that interoceptive and affective states need an animate body (LIT-207) can take this paper as its neuroscience. The paper already frames the action programme “from a bioengineering standpoint”, as four elements: a stimulus, a detector, an executor, and a detector that halts the correction. Man and Damasio’s 2019 proposal for machines that regulate a vulnerable body, filed beside this one on the same day but not read, starts by its abstract from the same view of feeling as the mental expression of homeostasis.
  • Damasio’s earlier work, filed the same day. The paper cites the Iowa gambling-task result of Bechara, Damasio, Tranel and Damasio (1997) as an example of decision-making that goes on without a felt experience; read with it, that experiment becomes evidence that body signals can steer behaviour below awareness. The somatic marker hypothesis is not cited by name, but its idea, that body states mark options with value, is restated here as the claim that higher cognition “borrows” the valence first built for homeostatic regulation. The paper’s account of mapping, which compares body maps with visual maps of objects, is a cousin of the retroactivation framework but does not cite it.

Note of 2026-11-03

The second appearance of ephaptic coupling in the record is now read: Anastassiou et al. 2011 (LIT-1953, NOTE-1516). Fields of the LFP’s size shift spike timing and synchronize nearby neurons without changing spike counts, explained by circuit physics. It shows the mechanism exists at physiological size and says nothing about sentience.