Two rewards sit ahead in time: A = $50, arriving after a short delay, and B = $100, arriving later still. As "now" advances (the elapsed-time slider), both delays shrink by the same amount. How much each reward is worth right now depends on the discounting model:
Hyperbolic (Mazur, 1987): V(t) = Amount / (1 + k·t)
Exponential (classic economic model): V(t) = Amount · e^(−k·t)
Exponential discounting is time-consistent: the ratio VA/VB = (50/100)·e^(k·gap) never changes as elapsed time increases, because both delays shrink by the identical amount — whichever reward you prefer at a distance, you still prefer up close. Hyperbolic discounting is not: it falls off steeply for short delays and gently for long ones, so as the smaller reward's remaining delay approaches zero its subjective value shoots up disproportionately, which can flip the preference from "wait for B" to "take A now" — the textbook mechanism behind present bias, procrastination and impulsive choice reversal.
- Hyperbolic / Exponential — switch the discounting function applied to both rewards.
- k — steepness of the discount; higher k devalues delayed rewards faster.
- Delay to A at rest / Extra delay to B — sets the two reward horizons before any time has elapsed.
- Elapsed time — slides the vantage point ("now") forward; drag it or press Auto-advance to watch the markers slide toward the origin and see whether the preferred reward changes.
Real-world relevance: this is the mechanism cited for why people sign up for a gym membership starting "next month" but skip today's session, why savers under-contribute to retirement accounts, and why commitment devices (locking in a choice while it is still far away) can out-perform willpower alone.