Global emotional moment (Craig’s model of awareness)

The theoretical capstone of Craig (2009): a structural model of how the interoceptive hierarchy could instantiate awareness. See insular-cortex, ad-craig.

The proposal

Salience across all relevant conditions — homeostatic, environmental, hedonic, motivational, social, cognitive — is integrated in the posterior→anterior insular progression, culminating in a unified meta-representation of “the material me” at one moment in time (“now”) near the AIC/frontal-operculum junction. Salience is ultimately grounded in homeostasis (significance for maintaining the body/brain), so the model is built on the interoceptive template: feelings are the “computational common currency of awareness.”

From a moment to a stream

An anatomical repetition of this unit, indexed by an endogenous timebase, yields a sequence of global emotional moments spanning a finite present — a “cinemascopic,” continuous subjective awareness (a “meta-memory”). Storage buffers enable comparison of past/present/future feelings, which instantiates a reflexive “observer” (a comparator that “cannot see itself”). von-economo-neurons are the speculative substrate for engendering and interconnecting these moments across AIC and ACC.

What it explains (and doesn’t)

• Explains: subjective time dilation (during intense emotion, moments “fill up” faster, so the timebase effectively speeds up and time seems to slow); music as an emergent property of awareness (a rhythmic temporal progression of emotional moments); the inclusion of the active agent (ACC) — filling a gap in the somatic-marker-hypothesis and answering the James–Lange criticism about internally generated emotion.
• Doesn’t (Craig’s own admission): how a feeling is constructed, the nature of the timebase, the salience metric, how emotions are differentiated, or how the two hemispheres integrate.

Relation to predictive accounts

The model has a comparator and represents “predictions of future feelings” — so it edges toward the predictive framing of interoceptive-inference — but it remains an ascending integration architecture with the comparator at the apex (AIC), not a descending generative model. This is the nuance at stake in feedforward-vs-predictive-interoception.