I've tried to clear this up for everyone before, and I guess it was missed it or it just didn't sink in or...
'Learning' type computers in cars does not mean they adapt to the driver. "Gee, it looks like Bob wants to be a tightwad today so I'll adjust myself to give him excellent mileage" is not a parameter in the computers coding. A learning computer is one that can adjust it's programming to allow for sensor deviations, etc. For example, if the engine coolant temperature sensor should read .xx volts when the engine has run xx amount of time and has seen the throttle at xx% vs. xxxx rpm with xx ambient air temp and xx, xx, xx ..... and the temp sensor is sending a signal a few .ox volts off from what's expected, the computer can 'learn' the new parameter and adjust for the change. As electrics age they normally tend to increase resistance and thus their expected values will change.
Maybe this explanation isn't as fun as thinking your computer is capable of cognizant thought, but it does allow it to optimize the system and not require constant replacement of sensors, etc. (which is probably the real reason for its presence in the first place - federal law requiring conformity for x years and x miles.)
Computers occasionally, as we all know from obviously being able to read this in the first place, catch a 'glitch' (normally voltage fluctuation induced) and create operational errors. Usually the easiest way to correct the problem is to reset the computer by disconnecting its power supply. The older the car and its sensors are, the bigger the difference felt when/as the computer 'learns' the values over again and corrects for deviations from the original programming PIDs.
Whew! That was a kinda short and fast, hopefully it clears up the misconceptions. I know I need another cold and frosty now....
