Patrick Hamilton's play opened in London in 1938 under the title Gas Light. A husband is systematically convincing his wife that she is losing her mind: objects go missing and he denies moving them, he tells her she said things she did not say, and every evening the gas lamps in the house dim, which he assures her is her imagination. They dim because he is in the sealed upper floor searching for jewels, and lighting the lamps up there draws down the pressure below.
The detail that made the play famous is not the deception. It is that the wife has direct sensory evidence — she can see the light change — and is talked out of it.
The clinical literature took the word much later, and it has since been overused to the point of near-uselessness; a great deal of what is now called gaslighting is ordinary disagreement, or lying, or someone remembering an event differently. The distinction that keeps the concept meaningful is this: an ordinary lie asks you to believe something false about the world. Gaslighting asks you to disbelieve your own faculty for finding out.
That is why the pattern targets perception and memory specifically. Denying that an event happened at all. Insisting on a different sequence with great confidence and no evidence. Attributing your account to tiredness, stress, hormones, drinking, jealousy, or a diagnosis. Producing the counter-account immediately and without hesitation, so that your own hesitation looks like uncertainty. And doing all of this repeatedly, over months, about matters small enough that raising them would seem absurd.
Repetition is doing something specific. Elizabeth Loftus's work on memory, covered later in this guide, establishes that recall is reconstructive rather than stored: each retrieval rebuilds the memory from fragments and is vulnerable to information introduced afterwards. Confident authoritative contradiction, delivered repeatedly, is exactly the kind of post-event information that degrades confidence in accurate recall. The victim's uncertainty is not weakness of character. It is the predictable output of a memory system operating as designed under a specific attack.
The sociologist Paige Sweet added the piece that individual-psychology accounts miss. In her 2019 analysis based on interviews with domestic violence survivors, gaslighting is effective not merely because of private confusion but because it mobilizes existing credibility hierarchies. The abuser does not have to persuade the victim alone; he has to persuade her against a background in which she is already the less credible party — because she is a woman, because she has a psychiatric history he arranged to have documented, because he is respected at work, because she is now visibly distressed and he is calm. Third parties then confirm the distortion, and the confirmation is the most powerful part.
The observable effects on the target are consistent enough to function as a signature. Chronic self-doubt about matters of fact. Compulsive evidence-keeping — screenshots, saved messages, journals — which is a rational response to an environment where the record is contested. Rehearsing conversations in advance. Apologizing preemptively. Asking other people whether an interaction actually happened the way it seemed.
If you recognize the evidence-keeping, note what it means: some part of you correctly identified the situation and started building a defense before the rest of you agreed. The countermeasure is to make that record external, contemporaneous, and held by someone else, because the one thing this technique cannot revise is a dated document written before the argument about it began.
Defense: Your Memory Is the Evidence: Detecting Gaslighting in Real Time