Learning Agency

The override circuit from the last chapter doesn’t run on insight. It runs on evidence. Every time you try something and it works, the circuit gets stronger. Every time nothing works, it gets quieter (Maier & Seligman, 2016).

Here’s how to train it.

How the skill grows

The override circuit (mPFC) is trained by a simple loop:

  1. Try something.
  2. Observe the result.
  3. Update the model: “this had an effect.”
  4. Generalize: “effort here produces results.”

When this loop runs often, the override circuit gets louder. It scans aggressively for control and dampens the shutdown circuit.

When the loop fails repeatedly, it quiets down. Default passivity returns.

Why “small wins” actually work

The popular advice (start small, get easy wins) has a real brain mechanism. Small successes are not just confidence boosts. They are training data for the override circuit:

  • Many tiny acts of control train the circuit that variables are controllable
  • Few large risks can’t do the same job because feedback is too rare
  • Recent, frequent reinforcement matters more than distant big wins

This is why people often get unstuck through micro-actions, not resolving big life questions.

What blocks the learning

  • Chronic stress depletes the override circuit
  • Repeated uncontrollable events teach the circuit to stop scanning
  • Self-talk that frames everything as out-of-control suppresses detection
  • Big leaps without small steps rarely produce training data

You can’t think your way to agency. You have to give the override circuit something to detect.

Domain specificity

The circuit trains per domain. You can have:

  • Agency at work, helplessness at home
  • Agency in fitness, helplessness in finances
  • Agency in conflict, helplessness in solitude

Each domain trains its own pathway. Generalization happens, but slowly.

Why this matters

  • Agency is not a personality trait. It is a skill the circuit learns.
  • The skill grows with repetition, not with insight.
  • Small wins are not motivational fluff. They are the training data of agency.
  • You did not “lose your motivation.” Your override circuit stopped getting reinforcement.
  • Domains don’t transfer automatically. Build agency in one corner and use it as the foothold.

Find one thing that is actually controllable. Do it. Notice you did. Repeat.