Bill Seitz of FluxGarden

I realise I’ve been a little glib with my “I reject the premise of mastery learning.” I’ll explain a little here. Long tweet, strap in. After WWII, psychologists rejected the dominant paradigm of psych at the time, which was “behaviourism”. The big idea of behaviourism is that you were not allowed to look into the inner workings of the human mind. The young psychologists of that era felt trapped. It took a young researcher named Noam Chomsky to attack the eminent behaviourist BF Skinner, paving the way for the next paradigm. Chomsky argued that behaviourism wasn’t enough. It couldn’t explain the natural capability of children to learn language. In order to explain that phenomenon, we have to look inside the mind. The young psychologists were finally set free. (The legendary psych George Miller later said he felt elated, leaving one of the early conferences of the revolution). The paradigm that they introduced is one that is still dominant today. This paradigm is called “cognitivism.” But I want you to imagine that you are a young researcher at the time. You were finally set free. You could look into the human brain! But … how? You feel unmoored. You knew that the behaviourists held their views because the psychologists before THEM looked into the human mind and made shit up — these were Freud and Jung, and they were now widely seen as kooks. So you didn’t want to fall into the same trap. What do you do? In this post war period, the technology that was starting to be declassified, that was starting to spread in universities, were computers. What were computers? They were information processors. They took inputs and transformed them into outputs. En massed, the cognitivists turned to computers as their organising frame. The human mind was surely a computer! Turing defined a Turing machine as a man following a set of instructions, after all. Surely the human brain was also an information processor! ALL of our learning research comes from this paradigm. Anders Ericsson’s mentor was A.I. pioneer Herbert Simon. As a young postdoc, Ericsson did a landmark book on thinking aloud protocols with Simon, and then he went off into the wild, seeking tenure. He began to think: “how do people get good?” And armed with his thinking aloud protocol, and in need of tenure, he went to the places where it was easiest to see how people got good: traditional music conservatories. We now know this work as “Deliberate Practice”. What is the core premise of DP? The core premise of DP is that you will do simpler exercises until you can solve most of them (as assessed by a test). Then you move on to more complex material that builds on those simple exercises. As a by product it is inferred that you will create the “right” mental models, automatically. We do not need to probe what, exactly, those mental models are. Why does it matter? You are given a set of inputs, you produced a set of correct outputs. This sure sounds like a neural network, doesn’t it? This is the information processing paradigm at work. Meanwhile, the US military began funding work to figure out how to improve decision making. This later turned into work to accelerate expertise. These researchers were applied psychologists primarily. They were not beholden to information processing as a paradigm. Many of these psychologists (and Gary Klein especially) did not care what paradigm they used, so long as it produced results. What they found was something very interesting. Experts were experts because the mental models they constructed were very different from novices. They were also different from journeymen (those who are good not great). But most importantly, they found that the same training could produce both great and good people. In other words, the mental model you construct when you are doing training MATTERS A GREAT DEAL. This is the major flaw of deliberate practice, and mastery learning. Being of the information processing paradigm, it assumes that everyone, given the same inputs (problem sets) will construct the right mental models to solve maths problems. But this is clearly not the case. Most of us think that maths is a computation problem. Those of us with high ‘g’, or raw intelligence, will do well and speedrun any mastery learning program. Then at some point they will hit a wall where the computations become too hard, and they will crash and burn. But there is another way. A small group of people realise that actually you can create mental representations (or visualisations) that make maths problems solvable using intuition. The limiting factor becomes how creative you are in coming up with these mental representations. If you are very very good, you will come up with representations that allow you to see things and solve things that nobody could before. These people become Fields Medallists. As for the military funded psychologists (whose field is called Naturalistic Decision Making, so let’s just say NDM) what did they learn? They learnt that if you guided the construction of the “right” mental model directly — that is, whatever mental model the experts in the domain have — then you can skip the high volume of practice demanded by DP/mastery learning, and get to journeyman level at an accelerated pace. This is what is meant by accelerated expertise. This is what the US military uses. There are many many many more implications of this paradigm. It throws into question the usefulness of IQ as a construct. It throws into question the possibility of ‘far transfer’. And it (incongruously) explains why DP retards the development of adaptive expertise, and any method that improves adaptive expertise will also accelerate overall expertise (as the NBA is now beginning to learn, given their adoption of Constraints Led Approach as a replacement for Deliberate Practice style drills). I’ll write a series of essays covering the implications later. But make no mistake: this is a paradigm shift almost as large as that of the behaviourists -> the cognitivists.

In fact it challenges the very premises of IQ as a construct. (Something Bessis has tackled: davidbessis.substack.com/p/a…) So you can see why I’ll need to write a series of essays. The difference is more fundamental than first meets the eye.

I’ll need to finish a series of essays first. Much of education research is built around an information processing worldview. (Ericsson’s mentor was Herbert Simon). All the accelerated expertise work funded by the US military is not. This explains the bulk of the differences.

Describe dreams?! WHY? Bessis says it's a huge part of the 'mathematical experience': the ability to describe something only you can see in your head. Bessis is laying the groundwork to TALK ABOUT THE REPRESENTATIONS in his son's head, when he starts learning maths.