Memory and Chunking

Memory

The current models of the human memory identify many different interacting subsystems. For the purposes of our discussion, we’ll concentrate on two: the Working Memory (“WM”) and the Long-Term Memory (“LTM”) (Baddeley, 1986).

Working memory is the memory subsystem that handles the pieces of information that the brain is actively and consciously processing in a given moment. The prevailing theories maintain that working memory can hold aroud four items on average (Cowan, 2001), where these items, or “chunks”, can be pieces of information of various nature, like numbers, sounds, or even aggregate concepts. Working memory is not only limited in its capacity, but is also prone to easy “forgetting” – since its purpose is to act as a buffer for active processing, it easily discards the least used chunks to make space for other incoming data. By analogy, it is usually compared to a computer RAM, even if it would be more accurate to compare it to a processor level-1 cache.

On the other end of the spectrum, we have long-term memory, which is a much more reliable storage mechanism, and can accomodate an impressively large amount of information. Our large body of memories are kept in long-term memory for decades.

There is a continuous flow of information between working memory and long-term memory, in both directions. On one hand, the creation of a permament memory happens through a process called “memory consolidation”, that brings the active state held in working memory¹ to long-term memory by modifying synapses in the cortex. On the other hand, a stored memory can then be brought from its inactive state in long-term memory back to active state in working memory by a “reactivation” process. In addition, every time this happens, the stored information is changed using this new active state through a process named “reconsolidation”.

Also stored memories are linked together, and these links are reinforced by the consolidation of new memories, as well as by the reconsolidation of old ones.

Sleep has as particularly important role in consolidation and reconsolidation processes. This is one additional reason why it is important to space our learning efforts in time. Spaced learning will enable more robust reconsolidation of memories and enable the creation of new associations between them through diffuse thinking.

Chunking

In our discussion on memory above, we briefly touched on the kind of information held in working memory, calling chunks the items stored in those four slots of memory. But what are these chunks exactly? Chunks are sets of related pieces of information, bound together through either meaning or use. Figuratively, chunks can be of different lenghts: basic pieces of information may be seen as short chunks (a digit, a short sound), while a more elaborate concept may be seen as a longer chunk (eg. Beethoven Symphony No. 5).

The concept is important, because it can be an aid in counteracting the heavy limitations of working memory. The four slots can be better utilized if loaded with longer chunks, because it enables us to think in terms of elaborate concepts and even better, to create more complex abstractions, which in turn can become new chunks, like when playing with LEGO™ building blocks.

It is important to know the phases that involve the formation of a new chunk:

• focusing our undivided attention to the information/concept we want to learn
• understanding of the concept
• gaining knowledge of the context in which the concept “exists”
• practice (see next chapter)

Having a good “library of chunks” facilitates the ability to combine and relate concepts in different fields (“transfer”), enable the “compaction” of the information in our working memory, and provides “patterns” to the diffuse thinking mode to fuel our creativity.

Continue to read on how to avoid Illusion of Competence.

References

1. Baddeley, A. (1986). Working Memory. Claredon Press - Oxford.
2. Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences, 24(1), 87–114. https://doi.org/10.1017/S0140525X01003922