Why You End Up Defending Purchases You Regret, and How to Keep a Regret List That Survives It

Say you keep a list of purchases you regret. One rule: when the regret shows up, you write the item, the amount, and one sentence about what you were thinking at the time. It is a sensible habit, and it has a flaw that only appears once the list is old enough to reread. Some of the entries will have been softened, by you, later. Not deleted; amended. A second sentence saying that actually it had been useful for something, or that it had been a reasonable decision given what you knew.

Some of those second sentences will hold up. Some will be a small piece of fiction. The problem is that at the time of writing them, both feel like remembering accurately.

This piece is about why that happens, what the research says about it, and a small change to how the list is kept that makes the rewriting visible instead of invisible. No list of mine is quoted; the mechanism is the point.

Memory takes the side of the choice you made

The tidiest evidence comes from a 2000 paper by Mather, Shafir and Johnson in Psychological Science. Participants chose between options described by a mix of positive and negative features, and were later asked to recall which features had belonged to which option. Their memory errors were not random. They tended to attribute more positive features, both correctly and incorrectly, to the option they had chosen, and sometimes more negative features to the one they had rejected. The authors call this choice-supportive memory distortion, and note that while it probably reduces regret and is good for general well-being, it is “problematic for memory accuracy, for accountability, and for learning from past experience” (abstract; Psychological Science 11(2):132–138, DOI, checked September 18, 2026).

That last clause is the one that matters for a regret list. The distortion does not feel like distortion. It feels like a clearer view of what happened. Having spent the money, you have an ongoing interest in the money having been well spent, and memory obligingly goes and finds the reasons. The name for the broader family of effects is cognitive dissonance reduction, and knowing the name does not protect anyone from it.

Sunk cost pushes the same way

A second mechanism reinforces the first. In the classic field experiment on sunk cost, Arkes and Blumer sold sixty season tickets to a university theatre at one of three randomly assigned prices, full price or one of two discounts, and then tracked attendance. Those who had paid more attended more performances (Arkes & Blumer, Organizational Behavior and Human Decision Processes 35, 1985, DOI). The usual reading is that people who have sunk more into something are more motivated to use it, which for a regretted purchase means: the more it cost, the harder you will work to make it retrospectively worth it.

One caveat, from a later paper that models the effect. Baliga and Ely point out that the ticket-window design is open to selection bias, since a customer who got a discount could tell a friend with a lower willingness to pay to go and line up (Baliga & Ely, “Mnemonomics”, working paper, checked September 18, 2026). The sunk-cost effect is well replicated across many designs; this particular experiment is just less clean than it is usually presented as. I mention it because the point of a sourced table is to be honest about the sources.

Mechanism	What it does to a regret entry	Source
Choice-supportive memory	Recalls more positives for the thing you bought, more negatives for the thing you did not	Mather, Shafir & Johnson 2000
Sunk cost	The higher the price, the stronger the pull to keep using it and to reframe the purchase as sound	Arkes & Blumer 1985
Dissonance reduction	Reconciles “I am a careful buyer” with “I bought this” by revising the second belief, not the first	Festinger’s framework, of which the two studies above are specific instances

What the rewriting costs

The money is already gone, so the cost of defending a purchase is not financial. It is informational. A regretted purchase that stays regretted teaches you something about your own buying: which category, which trigger, which time of day. A regretted purchase you have talked yourself out of teaches you nothing, and it leaves you with a slightly inflated sense of your own judgement, which makes the next one more likely.

That yields a testable prediction, and it is the first thing I would check in a list of your own: the entries you later soften should cluster in the categories where you go on to repeat the mistake. If the softened entries are all gadgets, or all annual plans, or all things bought late at night, the list is telling you where your judgement is least reliable, and the softening is the tell.

The append-only regret list

The fix is structural, not motivational. You cannot decide to remember accurately, but you can make the record resistant to revision.

1. Never edit an entry. If you later think a purchase was fine after all, write that as a new, dated line underneath the original. Both versions survive, and the revision is visible as a revision. A year on, the later lines will still split into genuine updates and defences, but the defences are now legible as defences because they sit under the sentence they contradict.
2. Record on a schedule, not on a feeling. On the first of the month, look at the previous month’s discretionary purchases and ask about each one. Waiting for regret to arrive means only the sharp regrets get recorded, and the sharp ones are the rare, large, obvious mistakes. The persistent low-grade ones, a subscription you keep meaning to use, a tool that is fine and that you did not need, never produce a feeling strong enough to prompt a note, and in aggregate they are the larger sum. This is the same reason tracking every purchase for a month finds things a memory-based review does not.
3. Write the amount and the sentence, nothing else. No score, no category. Categories get assigned at reread, when you can see the whole list, and assigning them in the moment lets the choice-supportive memory in early (“this was an investment”).
4. Reread at a fixed long interval. Six months is long enough for the softening to have happened and short enough to still remember the circumstances. Mark, honestly, which later lines you now think were defences. That count, over time, is the only measure of progress the list gives you.

What the list is for

It is not a spending plan, and it will not on its own reduce what you spend. What it produces is a map of where your post-purchase memory is least trustworthy, which is useful information for the next purchase in that category, and useless if the map has been quietly redrawn by the same process it is supposed to catch.

Two related tools cover the neighbouring ground. The unexplained-purchase bucket catches purchases you cannot reconstruct a decision for at all, which is a different failure from the ones you regret and then defend. And an explicit price on your own time is a partial defence against the most common kind of rationalisation, because “it saves me time” is the sentence most easily written under a regretted entry and the one most easily checked against a number. For subscriptions specifically, the entry that reads “I will use it eventually” is the just-in-case subscription, and it deserves its own line.

This is a habit for keeping a record of your own decisions, financial or otherwise, and the one rule I would carry to any such record is: make it append-only, because the version of you that rereads it is not a neutral party.

Sources checked September 18, 2026: the Mather, Shafir & Johnson abstract (Princeton repository page and SAGE via the DOI, both linked above), the Arkes & Blumer DOI (resolves to the Elsevier page; the experiment details are as described in Baliga & Ely’s working paper, linked, which I read), and Baliga & Ely for the selection-bias caveat. This article is general information, not financial advice.