finding

active

finding:samples-drawn-from-the-trained-self-prior-correspond-to-sticker-free-self-in-diverse-poses

Samples drawn from the trained self-prior correspond to sticker-free self in diverse poses

Demonstrates the self-prior learned the sticker-free body distribution as intended

Source paper

extracted_from

Active Inference with a Self-Prior in the Mirror-Mark Task

(2026) · Dongmin Kim · Hoshinori Kanazawa · Yasuo Kuniyoshi

Neighborhood — ranked by edge-count

Concepts (1)

concept

Density Model of Self-Experience
supports
The self-prior functions as a density model over the agent's familiar multisensory states in latent space

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

Passing a sticker-bearing latent through the self-prior removes the sticker in reconstruction, confirming distribution favors sticker-free statefinding0.797
Shows the self-prior's generative distribution rejects sticker-bearing states
The sticker-removal behavior induced by the self-prior corresponds to stimulus-elicited intention rather than endogenous intention, aligning with the developmental view of early intentional agencyclaim0.787
Connects the model's behavior to Zaadnoordijk and Bayne's taxonomy of intentional agency
"the self-prior can serve as an internal criterion for the mark-directed behavior observed in the mirror test, offering a computational basis for investigating the developmental origins of self-awareness"quote0.755
Load-bearing summary of the paper's central contribution
The self-prior operates as an internal criterion for distinguishing self from non-self, without external reward or explicitly computed sticker locationclaim0.753
Central interpretive claim of the paper, supported by EFE decrease after sticker removal
Zero-shot model stitching without a learned stitching layer is feasible because different text models embed data in remarkably similar waysclaim0.735
Strong evidence for representational alignment across models
The self-prior is functionally analogous to the body schema: it captures cross-modal associations and directly guides action planning toward multisensory mismatch reductionclaim0.733
Theoretical interpretation linking the self-prior to the established body schema concept
Untrained model (0 training steps) shows no clear EFE difference before and after sticker removal (Δ = +1.70)finding0.731
Control showing that the EFE signal is learned, not inherent to the architecture
Integrating the tactile modality into the self-prior model may improve learning efficiency for mirror self-recognitionhypothesis0.728
Forward-looking prediction based on Chinn et al.'s finding that tactile experience promotes earlier MSR in infants