I keep seeing this idea posted in messages, but have never found any documentation. Does anyone have a supportable reference? Thanks!

You may need to be more specific. Are you asking how the ECU works and whether it actually does "learn", or it's very existence? The answer to it's existence can be found by simply keying in "auto ECU" on Yahoo.com or Google, and 1000's of articles will appear to verify it's existence and what it does. The answer to how it works can be found in more extremely technical papers such as those found here:
http://www.ecutek.com/tuning/ignition/
http://www.autoshop101.com/forms/h25.pdf

Wikipedia has a simplified explanation of the ECU-- it's not highly technical and seems to be fairly accurate:
http://en.wikipedia.org/wiki/Engine_control_unit

Some sites will not give you the highly technical material b/c they see it as proprietary information (patented) and may even ask you to get it from the U.S. patent office. Also, some may refer to the ECU as the PCM or Powertarin Control Module (which is a broader term). Hope this helps.

I was actually questioning the frequent claim that the ECU learns the driving style of different drivers and adjusts itself to that -- thus the often-heard claim that you have to drive it hard so the ECU learns that is what you want.
The first of those explains quite well how the ECU adapts to fuels of different octane ratings. I could find nothing about "learning" a driver's "style."

The second also mentioned octane compensation, and told why the real adjustments cannot take place until the engine is up to normal operating temperature and so on. Still nothing about different driving styles, and it is, any case, a Toyota paper -- not Nissan.

The Wikipedia article is good for general background, but all it really tells us is that modern ECUs measure everything in sight that might affect combustion, and adjust accordingly. No real "learning" here, just continuous adjustment based on the environment.

I could see no references to adjustments that "remembered a driver's style" or anything resembling artificial intelligence. With the exception of some storage of relative fuel octane so that the ignition timing gets in the right ballpark, it seems to be just another feedback control system. That is, the ECU parameters are based on feedback from measured existing conditions and not learning or memory, with the exception noted above for octane.

There are some more references here that seem to confirm that.

I can easily understand that such learning is possible -- after all, chess-playing computers learn their opponent's weaknesses -- but I would imagine that if our cars had such a feature it would be a major advertising issue and we would be reading about it all over the place. And that's what led to my question!

Thanks for encouraging me to hit the Web and do some more research on the issue.

I don't believe the engine ECUs adapt to the driver at all, but they are constantly adapting or learning to atmospheric conditions, load, and now even reliability thanks to electronic throttle control.

Now with regards to tranmission ECUs, they most definitely learn driver behavior. Lots of manufactuers claim "fuzzy logic" transmission control which learns the way the driver uses the throttle in various load situations. I think this is where people get confused with regards to ECU "learning".

You guys (Dave B & Dan in St Louis) are very bright and I may have reached the limits of my brain at this point.

There are actually two interconnected "ECU's" on our car- one controls the engine (ECU) and the other the transmission (TCU) and both are often referred to as the Powertrain Control Unit (most people refer to them as the ECU- incorrectly).

I could try and write a dissertation here but instead will refer you to sites that may better explain this. In actuality, the term "learn" may be over-used (and I didn't help matters much by insinuating "artificial intelligence") when it comes to the PCM, although the modules do behave as if they are learning. Enclosed is a much better explanations than i could ever hope to give:

http://en.wikipedia.org/wiki/Transmission_control_unit

I'm sorry I can't be of much more help here. What I do know is that I had a >20% increase in mpg once I did the procedure described in my earlier post. It may simply be a correlation and not necessarily causation.

back to you guys.

I may have to retract my previous statement about getting 16MPG in the city.

I just filled up about a week ago and have been through a half tank and I have only driven 62 MILES!

But those are definatly a lot of 1-2 mile drives in 5 degree weather... So what can you really expect!

Also, if you're on a Chicago expressway bumper to bumper, traveling 20 miles in 1.5 hours at rush hour, your mpg will be extremely low. For me it's extremely hilly and steep terrain with serious potholes throughout the winter. In the city I have the double problem of some bumper to bumper plus up and down steep hills (which is hard on brakes and tires).

1-2 mile trips means your probably running rich before the engine completely warms up. On top of this, the ethanol, which burns fast, evaporates quickly and is generally inefficient as a fuel, is of limited help in the mpg area.

One thought tho, if you can, take your car out for a longer spin (>15 miles) at about 70mph at least weekly. This will help keep the engine clean by removing much of the moisture that's trapped b/c of the short drives and may also help clean up some of the carbon deposits.

braintree Def true, I do take out the car for longer spins on the weekend but the majority of my trips are reeeaaally short..

Sometimes I wish I drove more just beaucase the G rocks!
And I feel bad for her in this harsh weather going on such short trips, really can't be to great on the engine.