Thinking in mental models
I have to confess to something – I’ve never read some of the classics in neurology, like LeDoux. I have, for the longest time, found his “Emotional Brain” simply too hard, too obtuse to read. Damasio I could stomach, although he required my full attention. Ramachandran was a joy. Goldberg was ok. Carter was almost relaxing, and so, for the most part, was Sacks, obviously. But LeDoux? He was my example of “very hard stuff to read”.
Or at least, that was several years ago, when I attempted the book the first couple of times. I’ve recently had another go, expected hell, and instead was like “ok, this isn’t that bad at all… I mean, it could be clearer at points, but still…” And even that “could be clearer” turned out to be less troublesome than I thought (and I’ll get back to why later).
Obviously I could just decide “well, I got better/smarter/etc.” and move on. But it did get me thinking about the whole mental models and working memory concept, as promoted by Anders Ericsson. And that in turn made me realize what a useful thing this perspective is for sorting information and thinking about understanding, education, etc. in general.
One important issue is, it gets rid of the whole “stupid/smart” stigma. We have research showing that high intelligence is not necessairly useful in chess (although it is helpful in picking it up), and that expert GO players actually score bellow average on IQ tests (although this might be a reflection of them quitting their education earlier to pursue playing GO full time), and the mental models concept fits quite well with this data. Raw computing power is undoubtedly useful, but for expertise in any field it just can’t beat pre-trained patterns. This shouldn’t be a surprise with our pathetic system speed – we need cached thoughts to operate anyway, so a better and more effective cached thought system would naturally be a more effective method of domain-specific problem solving.
However, while we are socialized to be extremely sensitive about our intelligence, especially any threats to it, expertise or mental-model development did not recieve a similar treatment. So the option of “it’s not about how smart you are, but how many mental models you have and how advanced they are and how well they describe the process… Well, it gets a lot less threatening and potentially opens space for discussion and education.
And speaking of education, the mental models concept, together with working memory limits is definiately something which should be used to remodel A LOT of what current education does. Not everything, there are places and teachers and authors who take it into account, but wow is there a lot of stuff to consider.
Because, on the most basic level, that’s what happened with me and LeDoux. The first time I read his book, the content literally overflowed my working memory. It required me to process, at the same time, more than what fit in the good old magical number “whatever +/- 2”. Perhaps higher intelligence can expand this limit a bit, make it “whatever +1”, or “whatever -1” (+/-2 😉 ), but even so, if the information chunks – the mental model and their parts – aren’t broad enough, +/-1 , even +/-3 doesn’t change all that much. You’re still adding grains of sand when you could be adding buckets… Or beachfulls.
I could see that attempt at building up mental maps in LeDoux during my recent reading, when he went through the history of different models of emotion, slowly adding to them. However, without at least some of such maps pre-built, the pace at which he progresses, and the lack of a previous, general map introduced, makes it still far too steep a learning curve for a typical reader. This was why my original attempts at the book were such failures, and why my recent one was more of a “huh? I really found this that hard once?” The updated mental models simply allowed me to follow the arguments (and at times point to their errors) at a far better pace.
This approach is immensely useful in designing any educational program – be it a course, a book, or whatever you might consider. What are the basic mental models the participants will be starting with? What further models do they need to integrate. How many of these can they hold in their mind at once, how many will be overkill? What is the pace at which they will be integrating new models, so that we can add further models along the way? This is especially useful for long term courses, where there is actual time for any integration to take place.
All of this is, of course, harder to judge by expert, who’ve grown proficient in the use of their own mental models and have compressed them to a significant level. Untangling that compression and following the thread again, step by step, to map out the specific models and their integration time… Well, that’s extremely hard. Doable, and well worth doing, but hard. If we want to improve teaching, it’s something we need to do, one way or another.
Going even further, it’s a really useful (albeit, at this point at least, still hard and time consuming for me) way of thinking about the people you are discussing with. What mental models are they operating on. How much can they actually fit into their working memory and understand? While it’s tedious, it does offer some better chances of convincing, or at least striking up a conversation, and it does help one avoid the “that guy’s just stupid” tendency. I mean, some people are, and that’s true as well, but in many cases it’s really not the main issue.
Finally, for one’s own development, it’s the question of what models can I further integrate? What can I recode and “compress”, in order to be able to hold more concepts in my working memory at the same time? And how specifically to do so? Such a review might be extremely useful for planning further development, especially when you reach an advanced skill level, where effective learning becomes harder. Being able to focus on further condensing the models we use, so that we can move around more of them at once, appears to be a very promising solution here.