You try to explain it to a friend and hear yourself describing a comment about an email. It sounds like nothing. It sounded like nothing last time too, and the time before, and you have now had this feeling for eight months.
The reason each incident sounds trivial is that each incident is trivial. The thing you are reacting to is not any one of them.
Why you cannot just tell
Bella DePaulo and colleagues' meta-analysis (2003) assembled the evidence on cues to deception. The short version is that the reliable, obvious tells most people believe in are not there, and human accuracy at detecting deception from behaviour in the moment is close to chance.
Reading the room in real time is the wrong instrument for this. The instrument is the record.
This matters because the intuitive approach, watching them more closely for a giveaway, is the approach the research says will not work, and it keeps you inside a frame where each individual moment is deniable.
Ordinary influence is not manipulation
Worth separating, because the categories blur and treating all influence as sinister is its own trap.
There is a well-documented set of influence techniques that work on almost everyone: reciprocity after a small favour (Regan, 1971), the foot-in-the-door effect where a small agreement makes a larger one more likely (Freedman & Fraser, 1966), social proof (Goldstein, Cialdini, & Griskevicius, 2008), scarcity (Worchel, Lee, & Adewole, 1975), and the pull of apparent authority (Bickman, 1974).
These are used honestly all the time. A real deadline is scarcity. A genuine recommendation is social proof. The technique is not the problem. What separates them is whether the thing being signalled is true, and whether it would survive you seeing it plainly.
The move: Pattern, Not Incident
Method at a glance
Pattern, Not Incident
- Log before you confront. Write what happened, when, and in what words, before deciding whether it warrants a conversation.
- Write facts, not verdicts. Date, place, who was present, the actual words used.
- Act on the pattern. One entry is nothing. Four entries with the same shape is evidence.
The log does two things at once. It removes your reliance on memory and impression, which the deception research says is an unreliable foundation. And it forces distance between the sting of an incident and the decision about whether to act.
Write facts, not verdicts
This is the part people get wrong, and it is worth being blunt about.
The moment you write "he is manipulating me" or "this is gaslighting" into the record, you have converted a factual document into an accusatory one, and handed the other side the argument that you decided the conclusion early and collected evidence to fit.
A log of dated facts is very hard to attack. A log of interpretations is easy.
Evidence hygiene
Keep it off company property. Work laptops, work email and work drives belong to the employer, not to you, and access to them is not yours to control. A log that lives on a work device can be read, and it can disappear on the same afternoon your badge stops working.
And on recordings: laws vary by jurisdiction and some require the consent of everyone present. Getting it wrong can cost you the evidence, the case and the job at once. If you are considering it, that is the point where this stops being a communication question and becomes one for a lawyer where you live.
One pattern worth being able to name
Sarah Harsey and Jennifer Freyd (2020) studied a specific sequence they label DARVO: deny, attack, and reverse victim and offender. In their study, exposure to that response pattern reduced how credible observers found the person raising the concern, and increased sympathy for the person responding.
Knowing the shape is useful for a narrow reason. If raising a concern reliably produces denial, then a counter-accusation, then a reframing in which you are the one who has caused harm, that sequence is itself a repeating pattern and belongs in the log as one.
Where this tips into something worse
A log is a tool for establishing whether a pattern exists. It is not a tool for building a case against someone you have already decided about.
If you find yourself recording only the entries that fit, or reading ordinary friction as evidence, the instrument has stopped measuring. The honest test is whether you are logging things that do not fit as well as things that do.
And a caution the chapter makes explicitly: this is a recognition framework in a communication book. It is not a diagnosis of anybody, and terms like narcissism and psychopathy are clinical categories, not descriptions you can apply to a colleague from the outside. Naming a pattern in behaviour is legitimate. Diagnosing a person is not, and it undermines you when you try.
When this is the wrong tool
If what you are describing involves threats, coercion, or fear for your safety, this is not a communication problem and a book is not the right resource. That belongs with a lawyer, a domestic abuse service, or the police, depending on the situation.
Where to start
Start the log tonight, somewhere personal, and write only what was said and when.
Give it six weeks before you decide anything. Either a shape appears, which tells you what you are dealing with, or it does not, which is also worth knowing.
Common questions
How can I tell if someone is manipulating me?
Not from any single incident, and not by watching them more closely for a tell. Meta-analytic evidence finds that behavioural cues to deception are weak and that people detect lies at close to chance (DePaulo et al., 2003). The reliable method is a dated log of what was actually said, reviewed over weeks for a repeating shape.
Why does each individual incident sound trivial?
Because each one is trivial. That is the shape of the thing rather than evidence you are imagining it. A single entry amounts to nothing; four entries following the same pattern are the beginning of something you can actually point to.
What should I write in a log, and where?
Facts only: date, place, who was present, and the words actually used. Writing conclusions such as “this is gaslighting” converts a factual record into an accusatory one and invites the argument that you decided first and collected evidence to fit. Keep it somewhere personal, never on a work device you do not control.
Should I record conversations as evidence?
Check the law where you live before you do anything. Recording rules vary by jurisdiction and some require consent from everyone present. Getting it wrong can cost you the evidence and more, so that question belongs with a lawyer rather than a communication guide.
Sources
DePaulo, B. M., Lindsay, J. J., Malone, B. E., Muhlenbruck, L., Charlton, K., & Cooper, H. (2003). Cues to deception. Psychological Bulletin, 129(1), 74–118. · Harsey, S., & Freyd, J. J. (2020). Deny, attack, and reverse victim and offender (DARVO). Journal of Aggression, Maltreatment & Trauma. · Paulhus, D. L., & Williams, K. M. (2002). The Dark Triad of personality. Journal of Research in Personality, 36(6), 556–563. · Tepper, B. J. (2000). Consequences of abusive supervision. Academy of Management Journal, 43(2), 178–190. · Regan, D. T. (1971). Effects of a favor and liking on compliance. Journal of Experimental Social Psychology, 7(6), 627–639. · Freedman, J. L., & Fraser, S. C. (1966). Compliance without pressure: The foot-in-the-door technique. Journal of Personality and Social Psychology, 4(2), 195–202. · Goldstein, N. J., Cialdini, R. B., & Griskevicius, V. (2008). A room with a viewpoint. Journal of Consumer Research.