Contribution
A compact neural account of feeling in three layers. At the functional layer, feelings are separated from the drives and emotions (action programmes) that cause them, and given a job: they are valenced proxies of homeostatic state that support learning and anticipation. At the systems layer, the paper argues from lesion and stimulation evidence that feeling is generated in the upper brainstem and only modulated by the insula and somatosensory cortex. That is a direct challenge to Craig’s view that the anterior insula is the “sole platform” of human feeling and awareness (refs 60, 111). At the cellular layer it adds a new speculative proposal: feeling begins in unmyelinated interoceptive fibres.
Key insight
Feeling is interoception made experiential. The body is mapped centrally whether or not anything is felt, and most regulation runs on those maps with no experience at all. A feeling is what such a map is like when it is experienced. Because body states are good or bad for life, the experience comes with a valence already attached, and that valence is the organism’s direct reading of its own biological value. On this view the exteroceptive senses have no valence of their own: they borrow it from the body (concluding remarks).
Assumptions
- Feelings are about the body. Descriptions of feelings “tend to reference states of the body”, and introspective reports from antiquity on are taken as evidence of this (refs 1–5). Experiences of the exteroceptive senses are excluded from “feeling” by definition (Box 1, glossary).
- Feeling requires topographic maps. The search for substrates is restricted to structures with “topographically organized somatic maps”. This is offered as “sensible”, by analogy with vision and hearing, not argued. It is what excludes the amygdala, nucleus accumbens and basal forebrain from a direct role.
- Behaviour “consistent with” feeling is evidence of feeling. It is used for decorticate children (refs 112–113), decorticate mammals, brainstem stimulation in animals (refs 42, 117) and non-human species generally. The authors grant that feeling in non-human species “cannot at present be demonstrated”, and fall back on parsimony.
- Valence is intrinsic to homeostatic deviation. Body states “are necessarily valenced”. This is stated, not derived.
- Synchrony scales cellular to systems phenomena. The model “commonly accepted for cognition” (refs 158–159) is assumed to apply to feeling.
Key results
The paper reports no data of its own. What it argues from:
- Patient B (Fig. 2; ref 114, the authors’ own Cerebral Cortex case report). Complete bilateral destruction of the insula by herpes simplex encephalitis leaves pain, pleasure, itch, tickle, happiness, sadness, apprehension, irritation, caring, compassion, hunger, thirst and bladder and colon distension. Feelings “seem to dominate the mental landscape” of such patients.
- Brainstem lesions. Damage to the posterior half of the upper brainstem gives coma or a vegetative state, in which “feelings and sentience are abolished”. Damage to the ventral half gives locked-in syndrome, with feeling and consciousness spared (refs 17, 115, 116).
- Unilateral insula lesions reduce thermal and nociceptive discrimination (refs 4, 119, 120). Bilateral parietal lesions covering much of SI and SII leave behaviour “clearly indicative of feeling states” (refs 124–125).
- Stimulation. Brainstem stimulation elicits emotional responses with valence, measured by their effect on learning or self-stimulation, in mammals (refs 42, 117), and emotions with feelings in humans (ref 118, a case of acute depression induced by deep-brain stimulation).
- Anatomy of interoception (Fig. 1). The lamina I spinothalamocortical pathway and the vagus converge on the NTS, PBN and PAG and reach the posterior and then the anterior insula, largely through the thalamus. About 80% of the fibres in the main vagal trunk are unmyelinated and most of the rest poorly myelinated (refs 130–132). Interoceptive C and Aδ fibres conduct at 1–8 m/s, against 14–60 m/s for exteroceptive Aα and Aβ fibres (ref 129).
- Receptors along the axon. Some unmyelinated vagal fibres respond to ATP, serotonin, acetylcholine and capsaicin along the axonal trunk, not only at terminals (refs 152–157; Fig. 3).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Feelings are mental experiences of body states and report homeostatic value | definitional, with the introspective literature as support | Box 1, glossary; refs 1–5, 16 |
| C2 | Drives and emotions (action programmes) are distinct from the feelings they cause, and can run without them | moderate | regulation without awareness, e.g. autonomic face recognition in prosopagnosia and the gambling task (refs 33–40) |
| C3 | Felt experience evolved because it makes regulation more flexible, through attention, learning and prediction | hypothesis | an adaptive story; no comparative or experimental test is given |
| C4 | The insula is not necessary for feeling, and its role is modulatory and integrative | strong for “not necessary”, given the case report; inference for “modulatory” | patient B (ref 114); unilateral lesion effects |
| C5 | The upper brainstem (PBN, PAG, deep SC, with NTS) generates feeling | moderate | double dissociation of posterior and ventral brainstem lesions, decorticate cases, stimulation; correlational and partly behavioural |
| C6 | Non-human species with a conserved brainstem very likely feel | argued from parsimony | conserved anatomy plus behaviour “consistent with” feeling; conceded undemonstrable |
| C7 | Unmyelinated interoceptive axons, with ephaptic coupling and ligand binding along the trunk, are the cellular basis of feeling | speculation | the low myelination of interoceptive paths is documented; the link to feeling is “we propose” and “we suggest” |
| C8 | Evolution selected against myelin where access of the membrane to ions and ligands mattered more than conduction speed | speculation | an argument from pleiotropy; it competes with the metabolic-threshold explanation the paper also cites (refs 141–142) and is not tested against it |
| C9 | The number of fibres recruited codes intensity, and only large deviations become felt | conjecture | illustrative model, ref 160 for population coding |
| C10 | ”The advent of feelings was simultaneously the advent of the mind” | assertion | concluding remarks; cites ref 117 |
Concepts
- Action programme. An innate set of actions, visceral, muscular and cognitive, triggered by an internal or external change and aimed at homeostasis. There are two kinds: drives (hunger, thirst, libido, play, care of progeny, attachment) and emotions (disgust, fear, anger, sadness, joy, shame, contempt, pride, compassion, admiration), the latter “mostly triggered” by exteroceptive stimuli.
- Feeling. The mental experience accompanying a body state. The same word (“fear”) names both the action programme and the feeling, and the paper insists they are distinct.
- Interoceptive system. The vagus and the lamina I pathway, the circumventricular organs (the area postrema), and their brainstem and cortical targets.
- Sentience. Used here for the “cellular proto-phenomenon” that, once synchronized across many neurons, becomes a systems-level feeling.
- Ephaptic transmission. Sideways coupling between neurons through extracellular current, which myelin blocks.
Connections
- Seth’s biological naturalism (LIT-135, NOTE-169). Both tie consciousness to the regulation of a living body through interoception. Seth gets there through predictive processing and the free energy principle. This paper gets there through body maps, lesions and evolution, and makes no use of inference or prediction beyond saying that feelings help anticipation. The paper neither cites nor relies on predictive processing.
- Plant sentience (LIT-193, NOTE-120). Terrill and colleagues argue from a graded view of sentience. This paper’s parsimony argument also grades feeling across species, but it anchors feeling in brainstem interoceptive maps, which plants lack. So it is no help to the plant case.
- Roberts on animate bodies (LIT-207). Its claim that affective states need an animate body has here a neural account of why.
- Damasio’s other work filed the same day. Ref 38 is Bechara, Damasio, Tranel and Damasio (1997), cited as decision-making that goes on without a felt experience. The paper’s claim that cognition “borrows” valence from body states restates the somatic marker idea without naming it. The comparison of body maps with visual maps of objects is in the spirit of the retroactivation framework, which is not cited.
- Not held in the record. Craig’s insula papers (refs 4, 60, 111), the authors’ patient-B report (ref 114), Merker on consciousness without a cortex (ref 112), Panksepp (refs 42, 117), and Damasio’s books (refs 5, 17, 41). Searched for in literature.d by name.
Bearing on the record
- With LIT-542 it sources THEORY-045, which records where the two accounts agree (feeling is the experienced regulatory state of the body) and where they disagree (map or prediction; brainstem or agranular cortex). The narrower candidate, “feeling does not need the insula; the upper brainstem suffices”, now sits inside THEORY-090, where it is sourced jointly with Newen & Montemayor (LIT-626); here it still rests on one case report cited second-hand.
- No ML instruction. The bioengineering framing of an action programme (stimulus, detector, executor, halting detector) is description, not advice. Nothing here belongs in the anthology.
Limitations
- Opinion, not review. The evidence is chosen to support the authors’ view. Counter-evidence for an insular or cortical seat of feeling is summarized only as the view being rejected.
- The case against the insula rests on one patient (and an unspecified “some patients” with SI and SII spared) reported by the same authors.
- Behavioural criteria for feeling carry the decorticate and animal arguments. The paper admits that feeling cannot be demonstrated in non-human species.
- The cellular section is untested. No observation links myelination, or ephaptic coupling, to whether something is felt.
- The evolutionary claim is an adaptive story. No alternative, for example that feeling is a by-product of mapping, is considered.
Open questions
- Does the brainstem suffice, or does it only feed a cortex that is needed for some feeling? The double dissociation of brainstem lesions is consistent with both.
- What are the “physiological conditions” beyond mapping that a map must meet to be felt? The paper names intensity and wakefulness and leaves the rest open.
- Is there a test that tells the pleiotropy account of unmyelinated interoception from the metabolic-threshold account?
Corrections
- none to a seeded skim (there was no seed)
- Reference slips in the paper. Ref 3 prints the author as “Morrisj, J. S.”. Ref 22 (Kobatake & Tanaka, J. Neurophysiol. 71, 856–867) is dated 2012, which does not fit volume 71; ref 47 (Behav. Brain Sci. 29) is likewise dated 2012. Neither date was checked against the source, so they are flagged, not corrected.
- An uncited figure. “Depression alone is the leading cause of disease in the United States and the leading cause of non-infectious disease worldwide” carries no reference.