You have the built-in charge of the gate (doping) against
the charge of the channel (doping) across the insulator.
Depending on the dopings, you could have an enhancement
mode (depleted at zero bias) or intrinsic / "zero-VT" (at
threshold, more or less, at zero bias) or depletion-mode
(an established channel, inverted near the surface, at
zero bias).
Think of the gate doping as being equivalent to a gate
voltage, across the insulator, and attracting an "image
charge" sheet to form under it. P (or positive voltage)
pulls electrons and tends to invert P- making a NMOS
channel. Dope the gate N-type (per usual practice, often
gate poly gets the same shot as S/D) and the electrons
of the body are repelled, until a positive voltage applied
to the gate overcomes that balance and attracts them.
Think back to electrostatics and the gold leaf electrometer.
How a charged object brought near its electrode, induces
an image charge. Across air or across oxide, same
difference. Except oxide "holds onto things" and has more
flaws.
That's the old-timey intuitive explanation, from before
everybody got all fancy with bands and equations and
stuff.