Analogy; the unique cognitive mechanism for learning

Douglas Hofstadter and Emmanuel Sander have recently co-authored a book, Surfaces and Essences: Analogy as the fuel and fire of thinking (no, I haven’t read it). Hofstadter famously won a Pulitzer Prize in 1979 for Godel Escher Bach, which I reviewed in February 2009, and is professor of cognitive and computer science at Indiana University, Bloomington, while Sander is professor of psychology at the University of Paris.

They’ve summarised their philosophy and insights in a 4 page article in last week’s New Scientist (4 May 2013, pp. 30-33) titled The forgotten fuel of our minds. Basically, they claim that analogy is the fundamental engine behind our supra-natural cognitive abilities (relative to other species) and their argument resonates with views I’ve expressed numerous times myself. But they go further and claim that we use analogies all the time, without thinking, in our everyday social interactions and activities.

Personally, I think there are 2 aspects to this, so I will discuss them separately before bringing them together. To take the last point first, in psychology one learns about ‘schemas’ and ‘scripts’, and I think they’re very relevant to this topic. To quote from Vaughan and Hogg (professor of psychology, University of Auckland and professor of psychology, University of Queensland, respectively) in their Introduction to Social Psychology, a schema is a ‘Cognitive structure that represents knowledge about a concept or type of stimulus, including its attributes and the relations among those attributes’ (Fiske and Taylor, 1991) and a script is ‘A schema about an event.’

Effectively, a schema is what we bring to every new interaction that we experience and, not surprisingly, it is based on what we’ve experienced before. We even have a schema for the self, which we continually evaluate and revise dependent on feedback from others and our sense of purpose, not to mention consequential achievements and failures. A ‘script’ is the schema we have for interactions with others and examples include how we behave in a restaurant or in a work place or in the home. The relevance to Hofstadter’s and Sander’s article is that they explain these same psychological phenomena as analogies, and they also make the point that they are dependent on past experiences.

I’ve made the point in other posts, that we only learn new knowledge when we can integrate it into existing knowledge. A good example is when we look up a word in a dictionary – it will only make sense to us if it’s explained using words we already know. Mathematics is another good example because it’s clearly a cumulative epistemological endeavour. One can’t learn anything about calculus if one doesn’t know algebra. This is why the gap between what one is expected to know and what one can acquire gets more impossible in esoteric subjects if one fails to grasp basic concepts. This fundamental cognitive ability, that we use everyday, is something that other species don’t seem to possess. To give a more prosaic example, we all enjoy stories, be it in books or on stage or in movies or TV. A story requires us to continually integrate new knowledge into existing knowledge and yet we do it with little conscious effort. We can even drop it and pick it up later with surprising efficacy.

And this is why analogy is the method of choice when it comes to explaining something new. We all do it and we all expect it. When someone is explaining something - not unlike what I’m doing now - we want examples and analogies, and, when it comes to esoteric topics (like calculus) I do my best to deliver. In other words, analogy allows us to explain (and understand) something new based on something we already know. And this is the relationship with schemas and scripts, because we axiomatically use existing schemas and scripts when we are confronted with a new experience, modifying them to suit as we proceed and learn.

But there is another aspect to analogy, which is not discussed explicitly by Hofstadter and Sander in their article, and that is metaphor (though they use metaphors as examples while still calling them analogies). Metaphor is undoubtedly a uniquely human cognitive trait. And metaphor is analogy in compact form. It’s also one of the things that separates us from AI, thus far. In my own speculative fiction, I’ve played with this idea by creating an exceptional AI, then tripping ‘him’ up (yes, I gave him a gender) using metaphor as cliche.

To be fair to Hofstadter and Sander, there is much more to their discourse than I’ve alluded to above.