From a 1938 stage to the staff meeting
The name comes from Patrick Hamilton's 1938 play Gas Light, in which a husband dims the gas lamps and insists to his wife that the light never changed — engineering her collapse of confidence in her own perception. The play gave a name to a specific cruelty: not lying about the world, but lying about what the other person experienced, repeatedly, from a position of trust or power.
Scoped to work, the pattern is quieter but structurally identical: agreements that 'were never made', warnings that 'nobody raised', objections that become 'your misunderstanding', concerns met with 'you're the only one who sees a problem'. One revision is a memory dispute. The pattern — always in one direction, always from power toward less power — is the tactic.
What the research actually says
For years the workplace conversation ran ahead of the science, borrowing clinical language without organizational evidence. That gap is closing: a 2026 paper in Frontiers in Psychology introduced the Gaslighting at Work Questionnaire (GWQ) — a validated 12-item instrument whose factor analysis supports a two-dimensional structure: trivialization (your perceptions are minimized: 'you're overreacting', 'it wasn't like that') and affliction (the active distress of having reality contested).
Two findings matter for prevention. First, power: higher-status actors are positioned to gaslight because their version of events carries institutional default weight. Second, the research frames awareness itself as an intervention — teams that can name the pattern identify it earlier. What the science does not support is diagnosing individuals from single incidents; the construct describes sustained patterns, and this page follows that discipline.
Recognizing the pattern — not the incident
- ✗Systematic denial of the documented — decisions and warnings that were made get 'never happened', even against evidence.
- ✗Trivialization as reflex — every raised concern becomes 'overreacting', 'too sensitive', 'not a big deal' (GWQ factor one).
- ✗Memory blame flows one way — the record-keeper is always the one 'misremembering', and always the less powerful party.
- ✗Isolation moves — 'everyone else is fine with it' (unverifiable) detaches the target from the group's reality-check.
The counter — and its honest limit
The behavioral counter is contemporaneous documentation: decisions, objections and agreements written down when they happen, in shared spaces, not private notes. Not as an act of suspicion — as ordinary hygiene that incidentally makes reality-revision expensive. When 'that never happened' meets a dated, shared record with named participants, the revision fails in public.
An argument-level discussion produces, as a side effect of ordinary use, exactly the artifact gaslighting cannot tolerate: a decision record in which the framed question, the claims, the objections and the outcome all carry authors and timestamps. 'Nobody raised the risk' fails against the risk's own node; 'that's not what was agreed' fails against the agreement's. That is real protection — and here is its honest edge: structure documents reality; it does not repair people. Sustained manipulation is an interpersonal and organizational harm that needs HR, leadership and sometimes exit — not better tooling. A product page that promised otherwise would be gaslighting you about gaslighting.
Where it connects
Gaslighting is the reality-layer sibling of motte-and-bailey — one revises what was claimed, the other what happened — and it pairs naturally with tone policing, where 'you're overreacting' does double duty as trivialization. In the wider family it sits beside DARVO — deny, attack, reverse victim and offender — covered in the field guide with the 2024 PLOS ONE research that first quantified its users. The bias it weaponizes in bystanders is naive realism: the assumption that the confident senior account of events is the neutral one.