Probably the most important bit of work I did as a scientist was the malaise theory of depression - http://www.hedweb.com/bgcharlton/depression.html .
I worked on this intermittently for nearly 20 years from about 1980 when I first began to study psychiatry. My motivation was trying to understand how a mood state could apparently be cured by medication.
From what I was being told, it seemed as if 'antidepressants' were supposed to normalize mood while leaving the rest of the mind unchanged. Of course, this isn't really true, but that was what I was trying to understand initially. Or, to put it another way, I was trying to understand what was 'an antidepressant' - since none of the standard explanations made any sense at all.
So, how did I 'solve' this problem (solve at least to my own satisfaction, that is!). Part of it was 'phenomenology' (i.e. mental self observation) - especially to observe my own mood states in response to various illnesses (such as colds and flu) and in response to various medications (including some which I was taking for migraine).
But the best answer is that I did not really solve it myself, but only when I had subordinated my investigations to the work of two great scientists - two 'Masters': the Irish psychiatrist David Healy and the Portugese neuroscientist Antonio R Damasio.
This apprenticeship was almost entirely via the written world - and involved reading, thinking-about and re-reading (and thinking about and re-re-reading some more) key passages from key works of these two scientists. That is accepting these men as my mentors and discerning guides to the vast (and mostly wrong) literature of Psychiatry and Neuroscience.
The lesson that I draw from my experience is that real science (which is both rare and slow) is done and passed-on by a social groups comprising a handful of great scientists and a still small but somewhat larger number of 'disciples' who learn and apply their insights and serve to amplify their impact.
But even the great scientists have themselves mostly served as apprentices to other great scientists (as has often been documented - e.g. by Harriet Zuckerman in Scientific Elite).
So, when thinking about the social structure of real science, it would seem that real scientific work is done (slowly, over a time frame of a few decades) by small groups that are driven by Masters who make the breakthroughs; plus a larger number of 'prentices who learn discernment from the Masters ('discernment' - i.e. the correct making of evaluations - being probably more important than techniques or specific information).
But disciples by themselves are not capable of making breakthroughs, but only capable of incremental extensions or combinations of Master work.
And it is best if the Master can be primarily responsible for training the next generation Master/s to carry on the baton of real science. Disciples can - at best - only train-up more 'prentices with the humility to seek and serve a Master.