inductive coupling

When current flows in a wire it wraps the wire in a magnetic field, like the field around a bar magnet. If that field passes through a nearby loop of wire, a changing current makes a changing field, and a changing field pushes a voltage around the loop, which is exactly how a transformer works, except here it is entirely unwanted. That accidental transformer action between two circuits that happen to share magnetic flux is inductive coupling. It is how magnetic fields inject interference, and it is the usual culprit behind mains hum and switching-supply noise.

The induced noise voltage is proportional to how fast the interfering current changes and to how much of its magnetic flux your victim loop catches, roughly Vnoise = M times dI/dt, where M, the mutual inductance, grows with the area of the victim loop and shrinks with the distance between the two circuits. The practical levers drop straight out of that: shrink the loop area the victim encloses (run a signal and its return wire close together), move farther away from the source, and orient the loop so its plane is parallel to the field lines, since it then catches almost no flux. A big floating wire loop near a transformer is the worst case; a tight twisted pair is the best.

Inductive coupling is driven by CURRENT and magnetic fields, which is why its cures differ from those for capacitive coupling. A thin grounded electrostatic shield does very little against magnetic fields; instead you minimise loop area (a twisted pair, or a ground plane carrying the return right under the signal), increase distance, or use heavy magnetic shielding such as thick steel or mu-metal for severe cases. Honest note: low-frequency magnetic fields are genuinely hard to shield, so it is almost always easier to reduce the loop that catches them than to block the field itself.