Do human societies learn? If so, how do they do it, and if not, why not? The American activist singer and song writer Pete Seeger took up the first question in the 1950s (Seeger 1955) in a song whose concluding lines circled hauntingly back to its opening and whose refrain – ‘When will they ever learn?’ – gave anti-war protest in the 1960s a musical voice. Seeger's answer was, apparently, ‘never’. Like many a pessimist before and since, Seeger saw human beings as essentially fallible creatures, doomed to repeat history's mistakes. But modern societies cannot afford to stop with that unregenerative answer. The consequences of error in tightly coupled, high-tech worlds could be too dire (Perrow 1984). If we do not learn, then it behoves us to ask the next-order questions. Why do we not? Could we do better?

For social analysts, part of the challenge is to decide where to look for answers. At what level of analysis should such questions be investigated? Who, to begin with, learns? Is it individuals or collectives, and if the latter, then how are knowledge and experience communicated both by and within groups whose membership remains indeterminate or changes over time? Organizational sociologists from Max Weber onwards have provided many insights into why collectives think alike.