Attention is cheap, engagement is gold
The value of cognitive effort
Yesterday this video showed up on my YouTube recommendation feed.
Benjamin Keep, PhD, JD – What makes something memorable?
Judging by the title, initially I thought it’s about writing techniques, but apparently it’s specifically about learning. Fortunately, learning is still one topic I’m into and I had some time during a commute, so I decided to give it a listen. The ideas are very well-presented. It goes over multiple methods of knowledge encoding, and breaks down each method in terms of its effectiveness, weaknessess, and caveats.
The gist of it, I take, is about cognitive effort.
We can summarise the video as: The more you engage with a piece of data, the better you will internalise it. The more cognitive effort you exerted in the encoding process, the more effective your learning is (better retention, etc).
This is also something we all instinctively know.
But as it is with everything that we know we should/could do but struggle to do, engaging takes more active and painful effort while attending is easier and more passive (but even then, managing our attention is itself a difficult task that requires some effort).
Watching a short video is “merely” attending. Discussing it with a friend is engaging.
Reading an article is “merely” attending. Summarising it is engaging.
Listening to a podcast is “merely” attending. Paraphrasing in notes is engaging.
“Merely” copy pasting from coding tutorials won’t help you get better at coding.
“Merely” reading a guide on how to swim won’t help you not drown.
Know, do, be, create. Go touch grass, talk to people, build stuff.
Anyway, here are some points I picked out from the LLM-generated note (courtesy: my LLM scripts):
Cognitive effort refers to the extra mental work required to generate or process information, which makes it more memorable.
Strategies that require greater cognitive effort, such as creating inferences while reading or generating sentences, lead to better memory retention.
Different memory encoding techniques, such as imagery, drawing, or acting things out, help create unique memory traces in our mind, making information more memorable.
Low frequency words appear less frequently, making them seem more distinct. People tend to remember low frequency words better due to their rarity.
The distinctiveness of words can also impact shallow and deep processing.
The production effect suggests that producing a word aloud helps memory. This may be due to creating multiple memory traces or engaging in cognitive effort. However, reading ALL words aloud does not result in better memory recall compared to reading silently.
High value words may trigger a biochemical reward in the brain, enhancing encoding. For example, people try harder to remember high value words due to their importance.
Integration involves connecting new information with existing knowledge, making it easier to remember and understand. Research shows that students who integrate new information with their prior knowledge are more likely to retain and apply it effectively.
// leaving a meta-comment here to write about in future posts: the following points are good examples of how a statement becomes more sticky when we attach a question onto it, compared to just immediately stating/asserting something.
What is the role of motivation in encoding? Motivation can influence how deeply and effectively we encode information.
What is transfer appropriate processing and how does it apply to encoding methods? Transfer appropriate processing refers to the idea that an encoding method will be more effective if there is a match between the encoding and remembering situations. Encoding methods that expand the range of applicable situations are particularly effective.
Why does explaining steps help students more than just memorizing the order? Explaining steps makes the encoding more broadly applicable to other situations, helping students identify cues for when to apply a procedure in future problems. Encoding methods are not inherently good or bad but expand the range of possible matching situations.
Originally published to Proses.ID