Research Program / Claims / telepathy-5
During thought transmission, copying a thought deforms it; the receiver receives a new thought, not the original.
思想传递时，复制思想会使思想变形，接收者收到的是新思想而非原思想。
Concept telepathy · Domain Extended Perception and Consciousness Transduction · mechanism · interface 3/3 · testability 3/3 · status untested · ★ fundable
This program treats the Seth Material as a source of hypotheses, not as doctrine to be validated. Lead dossiers are AI-compiled from the corpus with a book-and-session citation after every statement; literature reviews are AI web-search summaries whose references were all DOI-verified against Crossref; the agenda draft is an AI synthesis. None of it is Seth's original text, and none of it represents scientific consensus. English fields are machine-translated; the Chinese text is authoritative.
Seth sources (book §session)
Adjudicating experiment
Approximately 300 participant pairs, 80 trials per pair, split into two measurement sessions spaced four weeks apart, over 12 months. Compare conventional linguistic transmission, fully isolated sending, and no-sender conditions. Senders receive randomly generated graphics or short stories with arbitrary attribute bindings; receivers first freely reconstruct and then select. Freeze semantic features, embedding models, and training procedures. Use double held-out partitioning by participant and target, and conditional permutation tests for target information beyond receiver priors. Also test whether a distortion matrix is reproducible across sessions. First demonstrate information existence, then analyze distortion structure.
Suggested paper title
Target-Specific Information and Systematic Reconstruction Bias in a Pre-registered Sender–Receiver Task
What it would overturn
Only if target-specific information appears in the isolated condition would it revise the hypothesis that without an input channel, one cannot reconstruct another person's random thought content. Systematic reconstruction bias alone would not overturn existing cognitive theories.
Candidate labs / funders
Alexander G. Huth; Jerry Tang; Shailee Jain
Specificity of Seth's formulation
It specifies no rules for 'different arrangements' or direction of bias, and is compatible with memory reconstruction, semantic compression, and predictive processing. The information-theoretic version is a newly added research hypothesis.
Note
Machine learning can generate stable bias patterns from pure noise plus common priors, so 'bias is not random' cannot replace the test that reconstructions contain hidden target information.
Evidence (1, DOI-verified via Crossref)
Cite: Future Mind Institute, Seth Material Research Agenda, claim telepathy-5, https://www.seth.org.cn/en/research/claims/telepathy-5