The Impulse

Why motivation fails and habit succeeds

You know the pattern. A burst of effort — the new year’s resolution, the crash diet, the all-nighter. Then nothing. Then another burst. The sawtooth of motivation.

This derivation proves mathematically what your body already knows: bursts cost more than consistency. Not a little more. Ten times more. The proof uses one inequality — Jensen’s — and the result is decisive.

Claim: A control strategy relying on high-intensity, short-duration effort

(“motivation”) is energetically inferior and structurally destabilising. Stability requires a control signal whose variance approaches zero.

Two Control Strategies

Compare two strategies delivering the same total work.

Strategy A — Pulse Control (Motivation). u_A(t) = Σ_i A·1_{[t_i, t_i+τ]}(t), with large amplitude A, short duration τ, and duty cycle chosen so that 𝔼[u_A] = ū. High amplitude, near-zero effort between bursts, reactive.

Strategy B — Continuous Control (Habit). u_B(t) = u_const. Low amplitude, constant, predictive.

The Cost of Variance

The effort term in the cost function is quadratic: λu². Variance is punished nonlinearly.

Both strategies are taken to be admissible square-integrable functions on [0, T]; idealised impulses (Dirac deltas) are shorthand limits, not the objects inserted directly into the quadratic cost.

The formal mechanism is Jensen’s inequality. Because u² is strictly convex:

𝔼[u²] ≥ (𝔼[u])²

with equality if and only if u is constant (zero variance). Therefore, among all control signals with the same mean input, the minimum effort cost occurs at constant u.

Numerical illustration. Total work required: 10 units.

Impulse: u = 10 for 1 unit of time. Cost: 10² = 100.

Continuous: u = 1 for 10 units of time. Cost: 1² × 10 = 10.

Impulse cost = 10 × continuous cost. Same output. Ten-fold penalty. This is a direct consequence of strict convexity.

State Dynamics

Under impulse control: between impulses, dx/dt = −ax ⇒ x(t) = x₀ e^(−at). After each spike, decay resumes immediately. Result: sawtooth trajectory, high variance, constant recovery mode.

Under continuous control: dx/dt = −ax + u_const. Result: convergence to x, minimal variance, no recovery cycles.

Fatigue and Substrate Damage

Finite substrates exhibit fatigue failure: repeated high-stress cycling causes collapse below ultimate yield. Impulse control drives stress from near-zero to near-maximum repeatedly, inducing cumulative micro-damage. Capacity degrades even if u_max is never exceeded.

Continuous control maintains constant load, permits adaptation, and preserves capacity.

Result

Theorem IV.1 (Impulse). Because u² is strictly convex, Jensen’s inequality implies 𝔼[u²] ≥ (𝔼[u])², with equality iff u is constant.

Among all admissible control functions with equal mean input, the minimum effort cost occurs at zero variance. The Operator minimises variance, not intensity.

Corollary IV.2. Impulse strategies maximise energetic cost, structural volatility, and fatigue accumulation. Habit is the least-action solution under finite substrates.

The steady runner outperforms the sprinter in a marathon. The consistent saver outperforms the gambler over a lifetime. The daily practice outperforms the annual retreat. Not because consistency is virtuous. Because u² is convex.