Assumptions about learning

[matti]

I thought the following might make for an interesting discussion about learning:

If you want to learn something, is it better to assume some knowledge in the person you’re trying to teach (this could also be yourself) and go back to basics if they don’t understand something or assume no knowledge at all and start with the basics no matter what?

At first, this seems to be dependent on if the person actually has some knowledge, e. g. in a similar domain, but I’m no sure. Maybe it’s always better to assume more knowledge and let the cognitive dissonance do the rest. On the other hand, when I restarted my free code camp journey recently, there were a lot of cases where I thought I know the answer, but in the end didn’t (or “knew” a wrong answer).

My new way of thinking about this is: How to shorten feedback loops? Both options might, depending on the context, shorten the feedback loop, between my expectations about what I have learned and what I have actually learned. And how to find a way to make sure that I have an appropriate amount of “feedback loop coverage”. In a sense it comes back to an unknown unknowns problem: How to find out (test for) unknown unknowns in my learning efforts?

[narthur]

Fascinating question.

My first thought is that both approaches are useful. At different times I’ve found taking a deep dive into the basics, really trying to master them, to be extremely productive. On the other hand, stretching to achieve something just out of my reach has taught me a lot, too.

Maybe another way of looking at it: Can you push yourself to the edge of your cognitive ability for a sustained amount of time focusing on the basics verses stretching to understand something beyond your current knowledge? I think the answer to that is sometimes A, B, or all of the above.

In my work as a web developer, there are times when a lack of basic knowledge prevents me from interacting efficiently with a problem. At other times, focusing on the basics is a distraction, and what creates results is stepping away from the computer and instead thinking about the abstract problem independent of the implementation details.

[byorgey]

As a learner, I echo @narthur. As a teacher, I find there is often another very important dimension when choosing the approach to use with a particular student, which has to do with their confidence, or their perception of how they fit into the community of knowledge. In particular, when they encounter things they don’t know or can’t do, are they likely to take it as (a) evidence that they don’t belong/are too stupid? or (b) a challenge to overcome? With some students I have to start more with basics just to build up their confidence. Of course I would like to move all my students in the direction of (b), but by definition you can’t do that just by throwing them in the deep end.

[matti]

As we all said, it has lots to do with context. It seems also that no approach works forever and interleaving approaches is important, probably even necessary. I’ll try to list out the things we came up with as regards to this:

• How to shorten feedback loops?
• How to ensure feedback loop coverage?
• When is approach A (assuming some knowledge) or approach B (assuming no knowledge) appropriate? When are both equally well suited?
• How to decide which knowledge to pursue?

I would love to find a way to answer these questions! Can we think up a way to do so? On what are these answers dependent on?

One idea would be to presuppose a fading of effectivity of approaches. Which is why interleaving is important.

I like the idea of doubting oneself as an observational technique, too: Knowing and learning to know is a contingent practice. Put another way: The way in which we know is just one way of knowing. The pattern of information around a concept (a chunk) could have been in a different shape. It might be distorted, simplified, reductive or incomplete in its current form. We could think that the chunk swims in a sea of its own possibilities.

I think that our “how” and “when” questions in that little list up there, also really point towards something like a target that is needed. It gets easier to answer these questions if we know more precisely what are we working towards. Instead of “getting better at coding”, or even “strengthen my JS knowledge”, something like “understanding React hooks well enough to be able to rewrite just this one class component in my app, by the end of the month”, makes those questions a little bit more answerable, imho.

I also snuck in another frame of reference in my example: Time. Let’s assume we would have to face a big test on a sufficiently defined learning goal in two days, two weeks, two months and two years. The further away the due date is, the more things we can consider to take into account. At the same time we lose specificity again, because to much becomes eligible as possible information.

In general, it seems that mobilizing as much frames of reference as possible seems like a good idea. Define the goal, set a due date, make yourself accountable, define the steps to get to the goal, get on the same page with your teachers, peers, etc. (learn to differentiate between right and wrong, good and bad, fair and unfair, as they do, etc.). Nail down as much of the context as possible. Remove as much noise as possible. Even though we still can’t control the shape of knowledge that will be coming out of this framing completely, we still will shape it a good amount.

[matti]

One more thought: If we say the answer to this is dependent on the context, then that context itself is dependent on its context, etc. The further we go up the chain of contexts, the harder it gets to articulate the influence on the topic at hand. I live on earth - but what kind of influence does this have on the question if learning the basics or assuming some knowledge is better? The inverse is also true: We can construct “inner contexts” through a practice of framing and produce context, in a way. Between a more general context of e.g. constructing one’s own identity and its place in a community in general and a more specific, like “Does Prof. XYZ seem to understand me/do I understand him?”, there is a difference in answerability, imho.

The smaller the target, the better this construction of intermediate contexts can be. Instead of “does this work for all people in all circumstances?” we might want to go with “does this work for me, right now?”.