Reward & Motivation: The Overjustification Effect

Reward Structure

Reward magnitude 40

Reward schedule

Framing (controlling → informational) 0.30

Class settling in…

Live Readouts

Avg. engagement

—

Avg. intrinsic motivation

—

Overjustification index

—

Sim time

0:00

How it works

Each of the 24 students carries two separate drives that sum into observed engagement, following Self-Determination Theory (Deci & Ryan) and the classic overjustification effect (Lepper, Greene & Nisbett, 1973):

E_i = clamp( I_i + X_i · s_i , 0, 1 )

dI_i/dt = k_rec·(I0_i − I_i) − k_over·M·(1 − F)·I_i
dX_i/dt = k_ext·( M·G(sched) − X_i )

I_i = intrinsic motivation, X_i = extrinsic drive, s_i = individual reward sensitivity, M = reward magnitude (0–1), F = framing (0 = controlling/task-noncontingent, 1 = informational/autonomy-supportive), G(sched) = schedule gain.

Overjustification term — a large, controlling (low-F) reward erodes intrinsic motivation I_i over time; the same magnitude delivered with informational framing (high F) barely touches it. This reproduces the original finding that tangible, expected, task-noncontingent rewards undermine intrinsic interest, while informational feedback does not.
Variable vs fixed schedule — a variable-ratio schedule (G ≈ 1.25) sustains a higher extrinsic drive and resists extinction better than a fixed schedule (G = 1.0) of the same average magnitude — the partial-reinforcement extinction effect (Skinner).
Remove Rewards — sets M → 0 so only I_i keeps students engaged; classes built mostly on high-magnitude, low-framing rewards collapse fastest, exposing how much of their engagement was ever intrinsic.
Each student has a random baseline I0_i and sensitivity s_i, so the classroom always shows a spread, not one uniform value.

High engagement Moderate Disengaged

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.