On a Tuesday in March you sit in your car after a session and feel something loosen. You said the true thing. Your therapist nodded. You drove home lighter than you've been in months. Three weeks later, on a grey Sunday night, you try to recall that feeling and find nothing there. Worse, you find a verdict: therapy isn't doing anything. I'm the same person I was a year ago.

You aren't lying to yourself on that Sunday, and you aren't lying in the car either. Both are sincere. They just come from two different minds, because your mood is quietly editing the archive.

The librarian who only fetches what matches

Psychologists have studied this for decades under the name mood-congruent memory. When you are in a particular emotional state, you tend to retrieve memories that match it. In the early 1980s, Gordon Bower proposed that emotions work like nodes in a network: activate sadness, and related sad material becomes easier to reach. Experiments since then have repeatedly found that people in low moods recall and attend to negative material more readily, and that this bias is part of how depression sustains itself.

The effect is strongest for what is already emotionally charged and ambiguous, which is a fair description of therapy. A session is rarely one clean event. It holds a moment of breakthrough, a stretch of boredom, a clumsy comment, a silence that felt either kind or cold. Which of those you pull up later depends heavily on how you feel when you go looking.

A related finding comes from research on depression and autobiographical memory. Mark Williams and colleagues described overgeneral memory: people who are depressed or have a history of trauma often recall the past in broad strokes ("things were always bad") rather than specific moments ("on that Thursday, I said no to my sister and it was fine"). Specific memories are the ones that serve as evidence. When they go missing, the general feeling is all that remains, and on a low day the general feeling is a harsh judge.

One caution on accuracy: a close cousin of this idea, state-dependent memory, the claim that you recall best when your mood at retrieval matches your mood at learning, has had a more mixed record in the lab. Mood-congruent bias is the sturdier finding. The practical lesson is the same either way. The self that evaluates your therapy is not a neutral observer.

Why therapy is especially vulnerable

Most of what therapy changes is invisible from the inside. You don't feel a new habit of pausing before you snap. You don't notice the argument that didn't happen. Progress mostly shows up as an absence, which is the kind of evidence a bad mood is happy to overlook.

Add the structure of the work itself. You often arrive at sessions when things are hard and leave when they have shifted. Then the shift fades, the ordinary week resumes, and the next hard moment arrives with no vivid memory of the last relief. If you only rate your therapy while you're struggling, you will almost always conclude it isn't working, simply because that's when you're asking.

This can have real consequences. People quit therapy on a low week, not because of anything the therapist did, but because the low week handed down a ruling on the previous six months. They then lose the very support that would have helped with the low week.

The two-witness problem

It helps to think of yourself as two witnesses to your own healing. The first is the person in the car on Tuesday: warm, specific, and reliable about what just happened. The second is the person on the grey Sunday, who is reliable about how things feel now but unreliable about what was true then.

Most of us let the second witness testify about the first. Better to let the first witness leave a written statement, in their own words, while the memory is still intact. Then, when the second witness takes the stand, there is something to cross-examine them with.

This is not about forcing positivity. A record kept on a good day can be wrong in the other direction, because good moods also color recall. The aim is to capture details and let the facts, not the mood, set the terms.

What makes a record that survives a bad mood

A vague note fails. "Good session, felt hopeful" is exactly the kind of general impression a low mood can wave away. What holds up is the specific:

  • Concrete events, not verdicts: "I told her about the lie I've carried since college and she didn't flinch."
  • What you actually did differently: "Texted back within an hour instead of letting it sit for three days."
  • Your own words at the time, even if they are clumsy. A sentence you wrote while it was true has a different authority than one you reconstruct later.
  • The mood you were in when you wrote it, so future you can weigh the source.

That last item matters more than it looks. A note reading "Mood 3/10, rough week, still: this session cracked something open" is more credible than either half alone. It shows the insight survived a bad day.

Your next moves

  • Within an hour of your next session, write three specific sentences. One thing that was said, one thing you felt in your body, one thing you want to try. Do it in the car, or on a bench outside, before the day swallows it.
  • Put a number and a date on every entry. Rate your mood from 1 to 10 and add the date. In a month you'll be able to see which entries were written in which weather.
  • Start an "evidence" page. Whenever you do something you couldn't have done a year ago, such as setting a boundary, pausing before reacting, or sitting through a feeling, write one factual line. No adjectives, just what happened.
  • On a low day, read the page before you judge. If you catch yourself concluding "nothing has changed," give yourself a rule: read your last five entries first, out loud if you can stand it. You can still feel bad afterward. You just won't be ruling on a case you haven't read.
  • Bring the mismatch to your therapist. If your Sunday verdict contradicts your Tuesday notes, say so in the session. "I felt hopeful after last week and by Sunday I'd erased it" is useful, workable material, not a failure.

What to expect

The record will not argue you out of your mood. Mood-congruent bias is not a flaw you can think your way past; it is how memory works in people with ordinary brains. What a record does is give the bias something to bump into. Over time, many people find the bad-day verdicts lose some of their authority, because they can see the pattern: the hopeless Sundays have always come, and they have always passed, and the progress was still there on Monday.

It also changes how you use your sessions. When you can see the arc across weeks, you can tell a genuine plateau from a rough patch, and you can walk in with something better than "I don't know, it's been a bad week."

Where Sesh fits

This is the gap Sesh was built for: what happens in therapy shouldn't stay in therapy, and it shouldn't be left to whichever mood you happen to be in when you try to remember it. It's a private place to capture a session while it's still true, with the date and your own words, so that the version of you on the hard days has something honest to read. If that sounds useful, you can take a look at https://sesh.lumenlabs.works. And if it doesn't, a paper notebook and the five steps above will do the same job.