NSFC Scientific Hypothesis Builder

Problem

In an NSFC application, a scientific hypothesis should not merely restate correlations from the literature. It needs to compress a locked research question into one causal chain: what condition, through what mechanism, produces what observable outcome. If the statement stays at “A affects B” or “A may act through B or C,” reviewers cannot judge whether the hypothesis is testable, specific, or meaningfully bounded.

How it works

The skill constrains generation with a three-layer pyramid. It extracts upstream conditions, core mechanisms, and downstream outcomes, then writes a dense hypothesis paragraph in the form condition → mechanism → phenotype:
- Causal direction: prefers “A regulates C through B” over vague “related to” or “affects.”
- Mechanism specificity: anchors the claim in proteins, pathways, cells, or system-level mediators.
- Boundary clarity: states the situation, trigger, or constraint where the claim holds.
- Falsifiability: the hypothesis should imply an experimental result that could disprove it.
The workflow includes element extraction, chain construction, hypothesis statement, and self-check, with core three-part templates and variants for interaction networks, boundary conditions, and temporal dimensions.

Boundaries

It works best after the scientific question is already narrowed, such as after literature review, candidate mechanism screening, or experimental scope selection. If the input is still a broad field or lacks a defined question, the output is mainly a structured scaffold rather than a complete proposal. It does not provide experimental data, statistical validation, or official NSFC formatting rules, and the final hypothesis still needs calibration against field-specific evidence and a feasible experimental plan.