Query syntax and logic: how to ask Roam questions with queries

Roam queries turn a sprawling notes database into something you can interrogate: they let users ask for “everything I’ve written that matches these conditions,” then combine those conditions with AND/OR/NOT logic to surface exactly the right blocks. That matters because Roam’s value depends on being able to reuse past thinking—finding related insights, spotting connections, and pulling context into a “conversation” with one’s own database rather than manually hunting through pages.

At the core are three logical operators. AND requires all specified page conditions to be true for a block to appear in results (e.g., blocks mentioning both “page 1” and “page 2”). OR returns blocks that match any of the listed pages (e.g., “page 1” or “page 2”). NOT excludes blocks that match a given page condition. The real power comes from nesting these operators inside each other using parentheses-like grouping, so complex rules can be expressed precisely—similar to how math uses parentheses to control order of operations. With that structure, a query can demand multiple ideas at once while explicitly filtering out noise.

The transcript then walks through practical patterns. One common use is broad retrieval for preparation: if someone wants to prepare for a talk about “gamification,” they might query for any blocks related to “difficulty matching,” “failure states,” or “user goal.” Because those are treated with OR, the results pull together dispersed notes even if they were never written in one place.

Another pattern handles synonyms and aliases. “Perceptual control theory” and “PCT” are treated as equivalent labels, so a query can be placed on either page and still retrieve all relevant notes without forcing the user to remember which term they used when writing. The transcript also highlights a cleanliness tactic: excluding generic query pages (via NOT) so earlier query definitions don’t flood results.

A further refinement targets a specific author. To find only what Warren Mansell has written about perceptual control theory (or PCT), the query combines the OR group for the theory terms with an AND condition requiring Warren Mansell to be present. This yields a focused slice of the database rather than a broad thematic sweep.

Finally, the transcript shows how to connect two concepts by grouping their multiple aliases. For example, it groups “reward prediction error” with its alias “RPE” and connects it to “perceptual control theory” with “PCT,” also incorporating relevant author names like Wolfram Schultz and Warren Mansell. Using nested OR groups inside an AND structure, the query surfaces blocks where terms from both concept groups appear together—effectively mapping how two ideas relate across the database.

Overall, the takeaway is that once query logic becomes fluent, Roam’s chaos becomes navigable. Queries provide a reusable way to resurface notes based on meaning (tags/page names and aliases), not on where the notes were originally placed, enabling faster synthesis and stronger idea connections.