Why your brain says yes to the impulse buy

Cost Me Research Desk · May 26, 2026

Imagine someone offers you $50 today, or $55 next week. Most people take the $50. Now imagine the same person offers you $50 in a year, or $55 in a year and a week. Almost everyone switches and waits the extra week.

Same delay. Same reward gap. Different answer. That single inconsistency is the entire engine of impulse spending, and it has a name: hyperbolic discounting.

The textbook model was wrong

For most of the twentieth century, economists assumed humans discount the future at a constant rate. A dollar a year from now is worth, say, 95 cents today; a dollar two years out is worth 90; and so on. Smooth, predictable, exponential.

It made the math tractable. It also turned out to be wrong. In a series of experiments through the 1970s and 80s, psychologists kept catching subjects making the choice described above — preferring sooner-smaller now but later-larger when both options were pushed into the future (Ainslie, 1975).¹

That pattern is impossible under exponential discounting. It is, however, exactly what you would expect if the discount curve is steep in the near term and flat in the distance — a hyperbola.

Laibson's formal model

David Laibson's 1997 paper in the Quarterly Journal of Economics made the idea load-bearing for modern behavioral economics (Laibson, 1997).² His “quasi-hyperbolic” formulation introduced a single extra parameter — usually written as β — that captures the discount you apply to everything that isn't right now.

Empirically, β tends to land somewhere between 0.5 and 0.8 for typical adults. Translated into plain English: the moment a reward stops being immediate, its perceived value drops by roughly 20 to 50 percent.

The moment a reward stops being immediate, its perceived value drops 20 to 50 percent.

This is not a small effect. It is the difference between a $100 purchase feeling like $100 (now) and feeling like $50 (in a week). The same item, the same price tag — only the timing of consumption changes.

How the bias actually shows up in spending

1. The “just this once” problem

In the abstract — Monday morning, looking at a quarterly budget — you intend to save. In the moment — Saturday evening, scrolling a sale — you spend. Neither version of you is lying. They're operating under different discount curves.

Frederick, Loewenstein and O'Donoghue (2002) reviewed decades of experiments on this and concluded that present-bias is the single most robust deviation from the rational-economic-actor model.³ It shows up across cultures, age groups, and stake sizes.

2. Buy-now-pay-later

Klarna, Afterpay, Affirm. The product literally separates the consumption (now) from the payment (later) — which is exactly the structure your hyperbolic discount curve rewards most. You get the full hit of the near-term reward at near-zero perceived cost, because the cost lives in the flat, far-away part of the curve.

Predictable result: average basket sizes on BNPL flows run 20 to 40 percent higher than card checkouts of the same merchants. The bias is being monetized.

3. Subscriptions you forget about

A $14.99 monthly charge feels like $14.99. A $180 annual charge feels like $180. Same money. Different perceived cost. The monthly framing pushes the bulk of the payment into the flat region where your discount function barely registers it.

The commitment-device fix

The reason Laibson's paper landed so hard wasn't the diagnosis — it was the prescription. He showed that rational hyperbolic-discounters will voluntarily bind their future selves, even at a cost. The classic example is the 401(k): you lock the money up specifically because you don't trust the version of you who will exist in three years.

This is the same logic as setting your alarm clock across the room. The cool-headed-you knows the in-the-moment-you is unreliable, and pre-commits accordingly.

Every effective impulse-spending intervention uses some version of this idea:

• A waiting period (the 48-hour rule) that forces the decision into the flatter part of the curve.
• A pre-set savings transfer that moves money before you can spend it.
• Removing saved payment details so each purchase requires a small but real effort.
• A friction that makes the future cost visible at the moment of decision.

Why willpower is not the answer

A common — and counterproductive — response to learning about hyperbolic discounting is to redouble willpower. That misreads the finding. The bias isn't a weakness in any single decision; it's a structural property of how human reward systems weigh time.

Telling yourself to “just be more disciplined” is asking the in-the-moment version of you, the one with the steepest discount curve, to override itself in real time. That fails reliably.

Commitment devices work because they don't require anything from the version of you who is actually making the choice. The cool-headed-you already made it, and the in-the-moment-you can't easily reverse it.

How Cost Me applies this research

Cost Me is, in behavioral-economics terms, an attempt to drag the far-away reward into the near term. That's the entire design.

The core calculator shows you, immediately, what a given purchase would be worth invested at the S&P 500's 30-year average return. The bias works against you when the future is invisible; it works for you when the future is shown alongside the present. Same dollar. Now you see both versions.

The 48hrs button is a literal Laibson commitment device. Tapping it doesn't cancel the purchase — it just defers it past the steep part of the discount curve. By the time the cooldown ends, you're evaluating the item from somewhere flatter on the function, which is when most impulse decisions reverse. We tracked this in the app's analytics: a meaningful share of items that go into 48hrs never come back out.

The Vault button is a different intervention for the same bias. By logging a resisted purchase as “saved,” the app turns an abstract avoidance into a concrete, visible total — which gives the in-the-moment-you a near-term reward (the count goes up) that competes with the near-term reward of the purchase itself.

The Amy coach uses your own pattern data to flag the moments your discount curve is at its steepest — late-evening sessions, repeat lookups of the same item — and surfaces a question before the tap, not after.

None of this requires willpower. It just makes the far-away cost visible at the moment of decision, which is the only intervention the research suggests actually works.

References

1. Ainslie, G. (1975). Specious reward: A behavioral theory of impulsiveness and impulse control. Psychological Bulletin, 82(4), 463–496. https://doi.org/10.1037/h0076860
2. Laibson, D. (1997). Golden eggs and hyperbolic discounting. Quarterly Journal of Economics, 112(2), 443–477. https://doi.org/10.1162/003355397555253
3. Frederick, S., Loewenstein, G., & O'Donoghue, T. (2002). Time discounting and time preference: A critical review. Journal of Economic Literature, 40(2), 351–401. aeaweb.org/articles?id=10.1257/002205102320161311

Want more like this? The neuroscience of why cash hurts more than cards or the adult version of the marshmallow test. Or head back to costme.io.