An interesting article you wrote there, Ken. I've enjoyed reading all the replies here as well. I probably don't have much to add to the discussion, having never ridden an e-bike at this point. That wont stop me from providing some input from an alternate perspective
I don't really know how much I've ridden pedal only, but I estimate I've worn out more than 50 bicycle chains in my life (many of them in severe service.) I say that to say that I have more than a passing interest in riding bicycles. The vast majority of my riding time is purely recreational, with the destination being the departure point.
My homebuilt recumbent tandem is on the original chain at around 6,000 miles. I have been the captain for almost 100% of those miles. When I ride with a not very fit stoker, or on the rare occasions that I've ridden with a stoker who is FAR more fit than I am, I've observed a relevant phenomenon. Bear with me here... The very fit rider tends to put in about 55-60% of the power. This is regardless of whether the disparity in fitness is tiny or huge. I've tried varying the crank length for cadence preference changes (yes the cadence is the same, but crank length still adjusts for the preference, by feel). When the fit rider puts in the same effort as they would on a single bike, the feeling for the more fit rider is that the bike is slow and sluggish, and the feeling for the less fit rider is that the bike is fast and sluggish.
Why do both the fit and less fit rider think it is hard to accelerate? I think it is because when only one person puts in 50% more effort, it is the same effort as putting in 50% more effort, but the same return as putting in 25% more effort.
Why does the more fit rider feel the bike is slow? The cyclocomputer proves out that two riders of equal fitness are substantially faster together on the recumbent tandem on downhills and flats, and slightly slower uphill, compared to separately on upright bikes. My guess is it is that lack of responsiveness again.
A totally different example: For my snow bike, I have taken individual sprockets from 4 different cassettes, and built a 6 speed cassette with even spacing, 14-36t. Sometimes I want to go just a bit faster, but the shift to the next taller gear puts me enough out of my ideal rpm range that I end up going only a tiny fraction faster for what feels like a lot more effort. Inevitably, I'll pick the lower gear. If the slope rises a bit, I might not even downshift, just put in more effort, because it pays a nominally higher return than cranking the next higher gear on flat does.
I propose that these same issues apply to electric assisted pedal bikes (but NOT necessarily to human assisted or human power backup electric bikes). If the benefit from adding pedal power does not FEEL proportionate to the effort, it is much harder to put in that effort. In the tandem example, I at one point gathered some heart rate data. It was vexing to me that I could easily get to 80% max heart rate and hold there for miles on my road bike, but not on the tandem. I thought there was something about the biomechanics of the recumbent position that was preventing my legs from using the heart and lung capacity. No. It turned out that it was just REALLY hard to force myself to work that much harder for the small dividends it paid with a less fit rider on the back. Riding the tandem alone, I had only marginally more difficulty achieving the heart rate I could get to on the road bike. I think part of that was that the tandem is twice the weight of the road bike, and so there is an incentive to accelerate more slowly and so on.
That all matches with Ken's hypothesis that if there is enough power on tap with the electric assist to make the human contribution feel less critical, or to make the return for extra effort feel too small, it is will psychologically difficult to even acknowledge that there is more in reserve in the human power side's controller.
My tandem experiences also reinforce that too little power might be much less encouraging as well. When riding with a less fit stoker, I tend to only slightly exceed their power output 90% of the time, and only pour on the power pulling away from traffic lights or crossing the top of a steep hill.
This is part of how I decided that I don't really want to experience more than 200-250w of output power from an assist. Experimentally, I am also trying another twist. I decided to try to make it so the assist runs out of gearing well below the speed I can and regularly do achieve unassisted on flat, such that I am only able to use the assist for hill climbing, and for getting through deep snow. We shall see