Someone sends the same kind of request they sent last month. You have already agreed before finishing the message. That reflex feels like a soft spot in your own character. It is closer to something measurable, and it belongs to the person asking, not to you.
Whose prediction this actually is
A set of controlled experiments found that when the same request goes out to a mixed group of people, women are asked to take it on, and volunteer to take it on, more often than men. The revealing part is what happens when the identical request goes to a group of only one gender instead, the gap disappears almost entirely. Everyone volunteers at close to the same rate. That single result says the difference was never a trait sitting inside the person who said yes. It was a belief, sitting inside the person asking, about who was likely to accept. Strip the framing away and the underlying mechanism still holds, a direct request tends to find whoever the person asking already expects will take it. That is a fact about their prediction, formed before you ever answered, not a fact about your own temperament.
There is a second finding worth keeping in mind, and it also belongs to the person asking. Across a large body of research covering many thousands of requests to strangers, people who asked for help consistently underestimated how likely they were to get a yes, and by a wide margin. The reason was not modesty. It was that the person asking does not feel the discomfort of saying no, only the person being asked does. The colleague who keeps handing you their work most likely has no real sense of what it costs you to take it on. That is not an excuse for them. It is useful information, the imbalance you are feeling is real, and it is not a secret you are missing, it is a blind spot they have.
What actually carries a cost
It is worth being honest here before offering anything. Declining a request out loud, in a way other people can see, does appear to carry a real professional cost, and the research on this is unusually direct, the cost attaches to how visible the divergence is, not to how much work is actually being done. A study following dozens of consultants at a global firm over two years found that the people who quietly restructured what they took on, without ever formally asking for less, were rated identically to the people who did everything asked of them, and none of them were denied a promotion. The people who made their limits visible, through a formal request or an open refusal, were rated meaningfully lower, and close to a third of them had left within the year. The lesson is not never say no. It is that the moves worth making are the ones that change what you take on without making the change itself the topic of conversation.
Where to start
The moment a familiar request lands, in writing, reply immediately with no answer in it. Restate what is being asked, name when you will come back on timing, and stop. You have not said yes and you have not said no. You have moved the decision out of the moment it arrived in, which matters, because the pressure to comply is concentrated almost entirely in live, face to face asking. The same body of research that found the underestimation effect also found that over email the pattern reverses, the person asking tends to overestimate how likely a yes is. Moving your answer into writing moves it into the channel where the pressure is already lower.
When it is the second or later time the same person has asked for the same kind of thing, find the last instance before you answer and look at what it actually cost, the start time and the end time, nothing else. Then answer using that number rather than a guess. Stating a duration is not a complaint. In most settings it reads as competence, not resistance, and it is the one move here with the clearest evidence behind it, an experiment that required people to connect a specific past instance to a current estimate more than doubled how often they hit their own predicted timeline.
Reading which of your own recurring requests are genuinely uneven, versus which ones are reasonable asks that just feel heavy in the moment, is something you will judge more accurately than a general rule can. What this gives you is a standard to check that judgment against, not a decision made for you.
There is a specific private sentence worth having ready for the instant before you answer, one that does not change what you say out loud but changes how the moment feels while you are in it, and it is worth having in place before the next request lands.