It’s slightly more complicated than that. Upthread, I said this:
The interesting thing is that pretty much any gasoline engine will, at some point, expose not-yet-burned mixture to a temperature well in excess of gasoline’s autoignition temperature. This is true even for engines with modest compression ratio. If you’re running an engine at wide-open throttle (WOT), peak cylinder pressure during combustion can be over 1000 psi. This is before combustion is completed, which means that there is not-yet-burned mixture somewhere in the chamber that has been compressed from ~14.7 psi to ~1000 psi by the already-burned part of the mixture. Assuming adiabatic compression, this bit of not-yet-burned mixture gets heated to about 1300F even before the flame front arrives. This is far above the autoignition temperature of gasoline (~536F), but that mixture patiently waits for the flame front from the spark plug to arrive and light it off. This is because of a thing called ignition delay: basically it takes some time, on the order of a millisecond or so, for the unburned mixture to realize what’s going on. By the time the chemistry starts spooling up, the flame front from the spark plug arrives and burns it all in a nice controlled fashion.
You can see an example of ignition delay in figure 2 of this publication (not a PDF). Researchers used a shock tube to achieve near-instantaneous adiabatic compression of an isooctane-air mixture, suddenly exposing it to a temperature far above its autoignition point. The plot shows that after arriving at this elevated temperature, the mixture took about 0.7 ms to ignite. That’s the ignition delay, and it varies with fuel type and also with temperature. In that plot, the temperature was about 2100F, far higher than what I described in the previous paragraph - which means in an engine, the ignition delay for that unburned mixture will be somewhat longer than 0.7 ms.
So if combustion takes a long time, you’re more likely to have problems with autoignition. Combustion takes a long time in engines with large cylinder bores, crummy in-cylinder flow structures, and low-RPM operation. If you want to shorten combustion time, you switch to an engine with smaller cylinders, shape the intake ports and combustion chamber geometry to set up in-cylinder flow patterns that create turbulence near TDC to help stretch the flame front, and run at higher speed. You can also add more spark plugs per cylinder, if you can fit them in there. It’s like burning a candle at both ends simultaneously: you’ll get the burn done quicker than with a single plug. The redundancy is a handy safety feature for airplanes, and it’s also good for efficiency (pilots, note the idle RPM drop when you do your magneto checks during startup). But getting that burn done quickly also helps eliminate knock. There are a number of mass-produced vehicles out there with two spark plugs per cylinder.
TL,DR: it’s not as simple as high-octane fuel at high compression absolutely not igniting without a spark; it just takes more time to autoignite.
Some important clarification on terms. In the first part of this post, I used the term autoignition. This refers to the spontaneous ignition of a patch of mixture, which starts a flame front that propagates outward in “normal” fashion to meet the other flame front that’s advancing from the spark plug. When this happens after the spark, the term autoignition is the right one to use.
Preignition is when the mixture lights off before the spark plug even sparks, typically due to a hot spot somewhere in the combustion chamber - a spark plug with the wrong heat range, a bit of glowing carbon somewhere on the head/piston, and so on. Preignition gets the entire burn done too early; if it happens at high load, it can expose combustion chamber surfaces to very high temperatures, causing rapid engine damage/failure.
Neither of those phenomena by themselves constitute knock/pinging. They start as “normal” combustion, i.e. “deflagration.” The flame front propagates by thermal means at subsonic speeds: each bit of burned mixture heats the bit of unburned mixture next to it, moving the flame front forward. Preignition may cause rough running, but you won’t hear anything like classic knock/pinging from it. Knock/pinging happens under particular circumstances in which the unburned mixture is at high pressure and temperature that render it unstable; in this case, a pressure wave is enough to propagate the flame front, further amplifying the speed and strength of that pressure wave until it’s moving at supersonic speed through the unburned mixture and bringing the flame front with it. This is no longer deflagration, it’s detonation. This is the exact same phenomenon that blew up 2700 tons of fertilizer in Beirut earlier this month: the rapid release of energy results in a shock wave that generates extremely high pressures and temperatures at its leading edge. This is the thing that nibbles away at the edges of head gaskets and other surfaces over time. The “pinging” sound you hear from the driver’s seat is that shock wave reverberating back and forth across the combustion chamber.