Lauri Nummenmaa

The wiki’s principal contemporary empirical universalist about embodied emotion, and the counterweight to Barrett in a way that Friedman (2010) — a review — is not. Where Friedman defends autonomic signatures argumentatively, Nummenmaa’s group changed the measurement: emBODY maps the felt body at pixel resolution across thousands of people, and finds the emotion-specific structure that peripheral psychophysiology has struggled to produce.

The position

Emotions are physiological survival functions with a strong biological basis, “only modestly influenced by cultural factors” — Ekman’s premise, defended with a new instrument rather than with faces. Three claims stack:

1. Discreteness. Each emotion has a distinct bodily fingerprint; emotions are “discrete patterns of systemic activity” (Nummenmaa & Saarimäki 2017). See bodily-sensation-maps.
2. Constitution. Those fingerprints do not merely accompany feelings — they determine the qualia of subjective emotional experience. This is the Jamesian commitment, and the strongest claim in the programme.
3. Universality. The fingerprints are pan-cultural and sex-invariant, and the universalism extends past Ekman’s six to social emotions like pride, shame and jealousy. See cultural-universality-of-emotion.

The signature move: relocate the evidence from expression to experience

Nummenmaa’s most interesting argument (in volynets-2020-cultural-universality) grants his opponents the terrain they have won. Faces are culturally variable — Jack et al. (2012) stands. His reply is that faces were always the wrong place to look for universality, because they are visible: expressions can be observed, copied, and shaped by display rules, whereas heart rate and muscle tension cannot. So the embodied layer should be the more conserved one, and the field’s universality debate has been conducted on its most culturally plastic component.

Where he is exposed

Two tensions the wiki records rather than resolves:

• The instrument cannot separate feeling from concept. Word-cued self-report of where an emotion is felt is compatible with shared concepts rather than shared physiology — precisely Barrett’s account. The programme’s answer is Nummenmaa et al. (2014)‘s word-cued ≈ induction-cued concordance, which on direct reading is better evidence than this wiki credited (three induction experiments, rs = 0.71–0.82, induction classifying better than words) and narrower than the papers citing it need (Finnish undergraduates; it is cited forward to license word cues across 101 countries). It also leaves the general objection standing, as the paper admits: a vignette is not a word but is still a token of a shared category. See embody. This exposure has since grown teeth: Ferré et al. (2024) show the concept EMOTION is itself largely constituted by interoceptive content, so the emotion word he cues subjects with arrives carrying a bodily expectation. The same result is ambiguous in his favour — a folk concept could be interoceptive because bodies are — but it converts the objection from a possibility into a measured quantity.
• The discussion outruns the limitation section. Volynets et al. argue for a biological basis to emotion while conceding the maps “are thus subjective feeling maps” with no physiological warrant. The constitutive claim (2) needs a link between felt maps and bodily states that his own papers say is weak (citing Mauss et al. 2005 and Critchley et al. 2004). See bodily-sensation-maps.

Notably, he is not in conflict with Scarantino the way he is with Barrett: New BET expects felt/behavioural regularities without expecting fixed autonomic signatures, which is roughly the profile Nummenmaa’s data show.

Independent replication — the programme’s best week in this wiki

Until now every emBODY result here came from Nummenmaa’s own group. Lyons et al. (2021) is a German clinical-psychology lab with no stake in the basic-emotion debate, using picture induction rather than words or movies, and the core result replicated at comparable accuracy. Replication by a disinterested lab with a changed cue is worth more to claim (1) — discreteness — than any further in-house extension, and it partly answers the word-cue objection above: these maps were never cued by an emotion word.

It also strengthens claim (1) from an unexpected direction. Lyons et al. found the maps dissociate two psychiatric populations — undifferentiated in schizophrenia (Torregrossa et al. 2018, on which Glerean and Nummenmaa are co-authors), differentiated but deactivation-dominated in medicated depression. Shared folk concepts of where sadness lives should not fracture by diagnosis.

Claim (2) — constitution — fares worse, and from inside the paradigm. Deactivation is most of the sadness signature in his own maps but “almost no deactivation reached significance” under picture induction; and medicated patients whose felt emotion had turned to deactivation classified most accurately of any group. A fingerprint that changes with the ink, and stays sharp while the feeling reportedly blunts, is not yet a fingerprint that determines qualia. See bodily-sensation-maps, antidepressant-emotional-blunting.

The founding paper supplies a third problem for claim (2), which the programme does not cite: in Nummenmaa et al. (2014)‘s experiment 5, naive observers shown the averaged basic-emotion maps recognised fear at 8% — below the 14% chance level, and happiness at 22%. A classifier separates fear from the other maps; a human being who has felt fear cannot pick it out. If the map constituted the feeling, that gap should not exist. The available defence — reading a heatmap is not having an experience — is also a concession, since it inserts distance between map and feeling exactly where claim (2) needs them fused.

One further exposure, now filed at is-more-interoceptive-awareness-better: his group’s proposal that age-related dampening of felt bodily emotion is beneficial invokes a generic mechanism (less signal → emotion becomes more cognitive → easier to regulate) that, taken at face value, predicts depression and antidepressant blunting should also improve emotion regulation. Neither does.