Greg, you do have "proof". I'm pretty sure [MENTION=48413]Quentin[/MENTION] knows of what he speaks.
Methinks "learned driving habits" ranks right up there with "Invoice price". Urban myth. Sounds good though, eh?
Here is my experience from digging into a few different modern ECUs that have been fully reverse engineered. I'll start by saying they are all different, but most support the same general concepts. I've been out of it for a while, I have not touched this stuff since 2007 or so, but I don't think the fundamentals have changed.
There are many things that are "learned" that have impact on SOTP performance. A few off the top of my head.
- Timing adjusters based on recorded knock sensing patterns (fuel octane adjusters). This is why you get more power on premium
- Long term fuel trims based on o2 readings in different load cells (there is generally one cell for WOT, but no learning is done there due to narrow-band o2). It gets its adjustment from the last stiochometric commanded load cell
- Transmission shift pressures (these get really different depending on the transmission internals, think shift speed...)
One of the biggest things you will find in the ECU killing power (but not learned) is something called Torque mgt. It is basically a governor to protect the drive-train. This is mainly used during shifts and leaves a lot of performance on the table in favor of reliability.
I've never encountered anything around commanded throttle. There is usually a simple static calibration map going from hardware input (either voltage or frequency) to a commanded %number. It is amazing what tweaking it will do to perceived performance . And yes, sometimes there are multiple maps based on a driving mode.