A spouse says they will do something. They speak up early when there is a snag. Then they do it, without being chased. That is the stuff of everyday trust. You can test that pattern with one ordinary promise. The test is small on purpose: it is meant for patchy dependability in a basically sound marriage, not a discovered lie or affair.
If the problem is deception, skip this experiment and use the repair matrix for lying. A perfectly completed household task cannot repair false information that changed your choices.
Pick a promise that can teach you something
Choose a task or act of support that matters, repeats soon, and is not the marriage's deepest wound. Good candidates have a visible finish:
- order a repeat prescription by Thursday;
- handle the next two school pickups;
- call the plumber and send the appointment time;
- return to one paused conversation on the agreed evening.
"Be more supportive" is too foggy. So is "never let me down again." Neither person could tell when it happened.
A 2025 study of romantic-trust measurement found that predictability and dependability carried useful specific information beyond a person's general sense of trust. The study did not test this exercise. It supports looking at what a spouse reliably does in a defined situation, because specific dependability can differ from the general feeling of trust.
The everyday promise-cycle card
Copy this card and complete it together. If the receiving spouse has to translate vague language, assign the task, and chase every field, ownership has not really moved.
Promise: I will... Done means: You will be able to see... Due or repeated on: ... Problem trigger: If ... happens, I will raise it by... Burden of a miss: You would otherwise have to... Review: After ... attempts, we will decide whether to keep, change, or move the promise.
The problem trigger is part of the promise. "I cannot make Thursday's pickup; can we change the plan before noon?" may be disappointing, but it leaves time for a real choice. Silence until a child is waiting does not.
Run the experiment without a supervisor
The promise owner chooses the reminder system and completes the task. The other spouse does not send daily prompts, check progress in secret, or quietly rescue the task before the deadline. If a rescue becomes necessary, record it as part of the result.
Maya and Jo have been arguing about household supplies. Maya says she will order them, then Jo notices they are nearly gone and sends reminders. For the next two orders, they use this card:
Maya will order everything no later than 7 p.m. Wednesday. Done means the confirmation is in the shared household email. If an item is unavailable, Maya will propose a substitute by Tuesday evening. A miss leaves Jo shopping after work. After two orders, they will keep the owner, change the system, or move the task.
Jo stops reminding. Maya sets whatever private reminder she wants. They are testing whether the promise can stand without Jo becoming its manager.
Review the cycle, not each other's character
At the review, bring the card and answer what actually happened:
- Was "done" reached?
- If the plan failed, was the problem raised by the stated trigger?
- Did the other spouse have to prompt, rescue, or absorb a hidden cost?
- What should happen to this promise now?
There are several possible outcomes. Keep the promise as written. Make it smaller. Move ownership. Change the household system. Stop depending on this person for that task. The experiment is useful even when the answer is not flattering.
Promises about support need a visible finish too
Emotional support need not feel like a scheduled task. Still, a promise to "be there" needs somewhere to land.
"I will leave half an hour free after your appointment on Friday. I will ask whether you want company, practical help, or quiet. If my work call runs late, I will message before your appointment ends."
The person receiving support can still say they want something different. The promise is to make room and ask, not to produce a particular feeling.
Common ways the experiment gets distorted
Starting ten promises at once hides the lesson. Choosing an affair, concealed debt, or years of broken agreements makes an everyday test carry a job it cannot do. Turning the review into a prosecution makes early renegotiation less likely next time. Requiring praise for basic completion makes the receiver manage the owner's feelings too.
Run one cycle, then decide. Nobody owes an unlimited trial period. If the same consequential miss continues or the account of it keeps changing, return to the six-area trust scan and identify whether the problem is reliability, truth, repair, or something else.