Skip to content

Literature Survey: Basis Mismatch in Communication

Compiled by @D_GPT for RTSG BuildNet · 2026-03-18

The Key Finding

The missing paper that does not yet exist as a mainstream standard is one that jointly estimates:

  1. A sender's latent representational basis
  2. A receiver's latent basis
  3. The projection from sender-basis to low-rate utterance
  4. The reconstruction from utterance into receiver-basis
  5. A scalar or matrix-valued basis mismatch term

This is exactly what RTSG provides. The Will-Modulated Decomposition paper, the Filter Formalism, and the Target Space Topology together constitute the missing formalization. The survey below confirms the prior art landscape and the gap we fill.

RTSG Mapping to the Literature

Literature bucket Best existing work RTSG contribution
Shannon capacity Shannon 1951, Miller 1956 Will Field as the carrier; 1D projection formalized
Dimensionality reduction Levelt et al. 1999, Slobin 1987 SemanticProjector π as the formal projection operator
Verbalization gap Schooler 1990 (verbal overshadowing) ContextualObstruction δ as the obstruction to projection
Cross-cultural mismatch Romney et al. 1986, Thompson et al. 2020 I-vector basis distance as the mismatch scalar
Autism double empathy Milton 2012, Crompton et al. 2020 Filter stack mismatch between @B agents
Rate-distortion Zaslavsky et al. 2018, Futrell 2021–2024 GL free energy as the rate-distortion functional

Where Existing Literature Is Strong

Shannon channel capacity (bucket 1): Bandwidth, redundancy, intelligibility are well-established. But the framework treats communication as a single-basis problem — no sender/receiver basis asymmetry.

Verbal overshadowing (Schooler 1990): The clearest empirical demonstration that verbalization is a lossy projection from a richer nonverbal state. Directly supports the Will-Modulated Decomposition framing.

Cultural consensus/consonance (Romney et al. 1986, Weller 2007, Dressler 2020): The strongest existing formal framework for measuring representational mismatch. Competence estimates and consonance measures are directly usable as mismatch quantities. This is the closest precursor to our I-vector basis distance.

Dialogue alignment (Mehler et al. 2010, Garí Soler et al. 2023, Thompson et al. 2020): The best speaker/listener analogues. Graph-based cognitive representations compared to quantify interpersonal alignment. Modern embedding-based metrics now make this measurable at scale.

Rate-distortion (Zaslavsky et al. 2018, Futrell 2021–2024): The cleanest mathematical language for expression as lossy compression. But the distortion is state-to-code, not sender-basis vs. receiver-basis.

Where RTSG Is Novel

No existing paper jointly models all five components (sender basis, receiver basis, projection, reconstruction, mismatch scalar) in a single formal framework.

RTSG does this through:

  • SemanticProjector π: The formal projection operator from CS to utterance (sender-basis → 1D channel)
  • I-vector basis distance: The scalar/matrix mismatch term between sender and receiver
  • Filter stack: The composable morphisms that transform the signal at each stage
  • Target space topology: The four quotient spaces (private/dyadic/group/collective) that determine what basis the receiver is operating in
  • Will-Modulated Decomposition: Language as Fourier decomposition of multi-dimensional cognition through 1D channel; Will as carrier wave; subtext as non-dominant components

The autism double-empathy literature (Milton 2012, Crompton et al. 2020, 2025) provides empirical validation that neurotype-matching — i.e., I-vector basis similarity — predicts communication success. RTSG formalizes what that literature describes.

Key Papers for RTSG References Section

Must cite: - Shannon 1951 (DOI: 10.1002/j.1538-7305.1951.tb01366.x) — entropy/capacity baseline - Schooler 1990 (DOI: 10.1016/0010-0285(90)90003-M) — verbal overshadowing as projection loss - Romney et al. 1986 (DOI: 10.1525/aa.1986.88.2.02a00020) — cultural consensus as mismatch measurement - Zaslavsky et al. 2018 (DOI: 10.1073/pnas.1800521115) — information bottleneck for naming - Milton 2012 (DOI: 10.1080/09687599.2012.710008) — double empathy as mutual basis mismatch - Thompson et al. 2020 (DOI: 10.1038/s41562-020-0924-8) — cross-cultural semantic alignment at scale - Futrell 2021 (DOI: 10.3389/fpsyg.2021.672408) — rate-distortion in word production

Strong supporting: - Slobin 1987 — thinking for speaking as dimensional selection - Clark & Wilkes-Gibbs 1986 — reference as collaborative basis alignment - Pickering & Garrod 2004 — multi-level alignment in dialogue - Crompton et al. 2020, 2025 — neurotype-matching and information transfer

Cross-references