A. As an undergraduate, an important experience in my life was the small "humanities" class that Lettvin taught on the history of science with Georgio de Santillana. In that class we had extensive discussions about what it means to know something. There was serious analysis of the scientific achievements of Galileo, taken in the historical context of his interactions with the church. To this day I understand the process of science more clearly because of that class.
B. When I was an undergraduate I used to walk by Jerry Lettvin's lab on the way to Minsky's shop in Tech Square, where I was employed as a "computer hacker." I would often drop in there and listen to Lettvin expand on some aspect of electronics or neurphysiology. One point that he made repeatedly is that almost all cells in the CNS are extremely small, have no obviously distinguished axon, and produce no easily-measured action potential, and that these are undercounted and not well studied because they are so hard to measure.
His hypothesis: Many of the "neurons" in the CNS are not explicitly directional. They sense the states of their neighbors and modulate their behaviors without action potentials. It is not important if this is actually true or not, but it inspired an idea that I have pursued throughout my career: Computation by Propagation of Constraints.
I would often drop into his lab in building 20, just to talk about things. These were not just about science, although that was the most common thing to discuss. However, he also helped me (and many other undergraduates) in the usual emotional crises that affect young people. He once even helped me get over the deaths of my pet gerbils. They were getting very old and sick so I had to kill them, and I felt terrible afterward. Jerry straightened me out that day!
C. More recently, in the 1990s, I needed to make a preamplifier for a ribbon microphone. A ribbon microphone is a very delicate mechanism with an intrinsic impedance of about 0.75 Ohms. It cannot stand any DC current through it, and the preamplifier needs to be extremely low noise. Jerry taught me a number of unusual techniques for obtaining very low-noise behavior, including some remarkable circuit ideas. I have numerous notes of my discussions with him, with hand drawings of his design suggestions, which I still study from time to time.