Building mental models

Lately, I’ve been reviewing some of the lectures in an artificial intelligence class at MIT through OpenCourseWare (OCW). As an aside, I highly recommend OCW to anyone interested in furthering their understanding of any of a number of subjects in science, engineering, mathematics, etc. I’ve taken or sampled from many of the courses made available at and intend to work my way through many more.

Back to the AI course… Paraphrasing what Professor Winston tells his students in one of the lectures, MIT doesn’t really teach the facts or even the mathematics associated with a particular problem to solve. Instead, it teaches its students how to create models.

Thinking about this reminded me of my experiences as an undergraduate at MIT: In a strange sort of way, I enjoyed taking midterms and final exams because they were all open-book – which meant I did not have to memorize anything and instead could simply show up and think creatively, which I love to do! If I couldn’t remember a fact or an equation – I could look it up in my books. With good enough foundational models in mind, if I couldn’t remember how something worked – I could go back to first principles and re-derive what needed to be done on the exam. Fun!

I hadn’t previously described my MIT education in terms of “creating models”. Rather, I’d thought in terms of “how to think”, or how to handle large amounts of incoming data and making some kind of sense out of them.   Making sense out of them, of course, is another way to say creating models matching the given data.

Ordinary thinkers, active learners and great thinkers

Whether we‘re aware of it or not, we all use mental models. How we use them, however, varies widely: Most people use them unconsciously and reactively. Frequently this works – and frees us up from having to consciously re-think everything encountered in daily life. What this approach misses is that models are always approximations and therefore inaccurate, and limited in scope of where they are relevant and useful.

Active learners interact with models differently: They create them, use them, and frequently refine them, consciously and sometimes reactively, at other times proactively.

What sets apart truly great and creative thinkers? They create, use and continually refine models – and they are much more likely than others to be both conscious and proactive in their interactions with their mental models.

My mental model of “life”

I have a mental model of my own life that I tend to think of in a combination of control theory and algorithmic terms: From whatever state I find myself in (here and now), things can go in many directions, like forks in a road or branches in a tree. Which branch I go to next depends on the one I’m on now and on what are the control inputs between now and the next time I measure my new branch. Control inputs that act on my model can be either my own choices and subsequent actions or external perturbations, that is, other things happening in the world around me.

Of course, like every other model, this one has limitations, so I’ll refine it: My mental tree is a little unusual among trees, in that it is infinite in size and continues to branch further and further into more choices, directions, and thus other branches. Thank goodness – I’d hate to arrive at a leaf and find my life ending up in a place where I can no longer move forward!

Each time I end up in a new branch, I step back and think about where I am, whether it’s where I expected to end up, and how I got there (what my control inputs were). If I ended up some place unexpected, I either refine my model of the state of my life, or review whether my control inputs (what I did and what external factors influenced me since the previous branching) were in fact what I thought they were (which is just another model).

In practical terms, nothing is totally deterministic, so some of these branches have a higher probability than others. My role then, in creating desired future outcomes is to enhance their likelihood of occurring. So if I end up in an unexpected branch, it’s also possible my models were just fine the way they were and I simply ended up traveling down a lower probability branch.

Even if I may never know exactly which of those reasons landed me where I am here and now, I simply repeat the process – moment by moment, hour by hour, day by day, year by year. Where am I (what’s the current state of my model)? Where do I want to go (desired outcome)? What action has the greatest likelihood of moving me closer to that (what are the control inputs required)?

What does my model mean in real life?

Applying this mental model dramatically enhances my resilience – as I can always benefit from considering what has caused me to arrive at my current state of being and to use it to inform my choices going forward.

Forming, applying and revising mental models seems to be a characteristic shared with some highly successful figures in business, such as Elon Musk and others. I am still learning what parallels exist with similarly successful people operating in other spheres of life.

Regardless, I am convinced (another model!) that success, however it is defined, is largely a matter of a) habit and b) frequent and proactive revision and refinement (or even outright throwing-out-and-starting-over-again) of our mental models. In other words, success has an awful lot to do with making a habit out of continuous awareness and conscious experimentation with a feedback loop to use the outcomes of that experimentation to inform the next batch of experiments.

Consider current life (model) and desired path; take action (input); evaluate. Rinse and repeat. Endlessly.

“I never lose. I either win or learn.”

I believe it was Nelson Mandela who said, “I never lose. I either win or learn.” I take that to mean, in anything he tried, he either was successful in reaching what he was aiming for, or he used the lessons learned from the previous experiment to revise his mental model and then set out to try again. (Yeah, his way of saying it is much more poetic!)

Beyond thinking about the use of mental models in being successful, in whatever form that takes, I’m also thinking a lot about grit and cross-disciplinary applications of foundational models. My intention is to explore that in a future article.

In the meantime, I’m very interested in hearing from you: What mental models do you tend to rely on most heavily? And what opportunities have you had to tear down an existing model and start fresh? Please write a note in the comments section or send me an individual email – I welcome your thoughts.