Mind Maps Tested: Porphyry's Trees, Tony Buzan and 20,000 Digital Maps

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

A true mind map is a radial hierarchy built around one central concept, with major ideas branching from the center and subtopics branching from those. That single center separates it from concept maps, which can have several central concepts and label the links between them, and from UML diagrams, whose shapes carry fixed technical meanings. Mind maps instead rely on personal associations, so a purple elephant can stand for a date in the French Revolution. Branching diagrams are old: in the third century Porphyry of Tyre drew Aristotle's categories as a tree, and Ramon Llull used similar techniques in the 13th and 14th centuries. British popular psychology author Tony Buzan made the term famous, promoting it from around 1974 with the BBC series Use Your Head.

The research gives a mixed picture. In a 2005 study by Cunningham, 80 percent of high school biology students said mind mapping helped them understand science concepts, while a study by Holland and colleagues found 62.5 percent of art and design students rated mind mapping software effective against 34 percent of computing students. In a 2002 study by Farrand, Hussain and Hennessy, undergraduates who used spider diagrams on a 600-word text improved recall by 10 percent, against 6 percent for those using their usual methods, and the gain held a week later, but their motivation dropped. A meta-analysis by Nesbit and Adesope found mapping more effective than reading text, attending lectures or class discussion, and suggested lower-ability students may benefit most. Joeran Beel and Stefan Langer then analyzed nearly 20,000 digital mind maps from more than 11,000 users.

• The average user in the Beel and Langer data created about 2.7 maps in total.
• The typical map held 31 nodes, with a median of about three words per node.
• One user built more than 200 maps, and the largest single map contained over 50,000 nodes.
• The most text-heavy single node held roughly 7,500 words.
• The episode argues that tools which generate maps automatically skip the effort that produces the memory benefit.