Empathy
The wiki’s first construct about other people’s feelings. Everywhere else the object of interoception is one’s own body; empathy is the case where the same machinery is turned on someone else — feeling with another, on the strength of running their state on your own hardware. It enters through Lutz et al. (2008), which asks what compassion meditation does to it.
The perception-action model
The tradition the Lutz paper works in treats affective empathy as a perception-action process (Preston & de Waal 2002): observing or imagining another person in an emotional state activates a similar state in the observer. The load-bearing neuroimaging result is Singer et al. (2004): watching a loved one receive pain recruited the affective components of the observer’s own pain matrix — anterior insula and ACC — but not the sensory components (SI, posterior insula). You feel the hurt of their pain, not its sensory quality.
That is why empathy belongs in an interoception wiki. The affective components of pain are the interoceptive/core-affect ones — how a bodily state feels — so empathy is, on this model, the re-use of your own interoceptive-affective representation to stand in for someone else’s. It is embodied selfhood pointed outward: the same as-if body model that lets you feel your own state, run on another’s.
Affective empathy vs. mentalizing
The Lutz paper keeps two circuits distinct, and the distinction is worth carrying:
| system | regions | job |
|---|---|---|
| affective empathy / emotion sharing | anterior insula, ACC (the salience-network core) | re-representing another’s feeling on one’s own interoceptive machinery |
| mentalizing / theory of mind (“mentation”) | TPJ, pSTS, mPFC, precuneus/PCC (overlapping the default-mode-network) | inferring another’s mental states, beliefs, intentions |
Compassion meditation, in the study, up-regulated both — the insula/ACC affective core (especially to distress) and the mentalizing network (especially in experts, and especially right-lateralized) — suggesting the trained state enhances feeling-with and reasoning-about-others together. The pSTS link is notable: its activation predicts self-reported altruism (Tankersley et al. 2007), which is the paper’s slender bridge from neural change toward prosocial behavior.
What compassion training does to it
The empirical claim from Lutz et al. (2008): voluntarily generating a compassionate state amplifies the empathic response to another’s emotional vocalizations — most sharply to distress (a crying woman over a laughing baby) — and the amplification scales with meditative expertise. Because the insula effect survived covarying out pupil-indexed autonomic arousal, the enhancement is specifically empathic rather than a generic increase in arousal. This is the wiki’s one piece of evidence that empathy is a trainable capacity with an interoceptive substrate — the prosocial counterpart to the self-directed insula training in Farb et al. (2010).
A caveat the wiki should keep
The perception-action / shared-representation model is one framework, not settled fact. Reading empathy as “the same interoceptive state, re-run” is exactly the kind of simulation claim that constructionist accounts would complicate — a shared affective activation at voxel scale need not mean a shared feeling, the same resolution worry the amygdala page records for salience detection. The Lutz study infers empathy from regional overlap with prior empathy studies, not from any independent measure of what the practitioner felt about the other person. Held here as the field’s working model, with that gap flagged.