A good hypothesis is a chain of reasoning: from what the field knows, through a gap or a conflict, to a prediction someone could bet against. In most labs that chain lives in one researcher's head. Neutropic makes it an artifact you can inspect, debate, and hand to a study design.
From gaps to predictions
Starting from your literature review, the agent identifies where the evidence is thin, where results conflict, and which moderators remain untested — then proposes candidate hypotheses for each. Every proposal states its expected direction and cites the studies that motivate it, so "why do we predict this?" is answered before it's asked.
Constructs, defined before they're measured
Vague constructs sink studies late, when the measures don't match the claims. Neutropic forces the discipline early:
- Each construct in a hypothesis gets an explicit working definition.
- Established instruments for measuring it are pulled from the literature, with their reliability evidence.
- Where constructs overlap or are commonly conflated, the agent says so.
Rival explanations are part of the map
For every hypothesis, the agent drafts the competing explanations a reviewer would raise — confounds, reverse causality, third variables — and notes which design choices would rule each one out. The hypothesis map isn't just what you believe; it's what you'd have to defend.
Editable, not oracular
Everything the agent proposes is a draft for you to argue with. Rewrite a prediction, reject a construct definition, add a hypothesis of your own — the map keeps the links to the evidence either way, and flows directly into experiment design as the next step.
