Here, I show the output of a protocol to break down information that exceeds typical working memory limits into chunks. The goal is to enhance understanding at first reading, both for the person breaking down the information and for the person reading the result. The original text, an explanation of how RNA interference works, can be found under "Text Of The Original Description." Readers who want to skip straight to the result should go to the section "Result of Chunk-Knitting Procedure." The goal is to produce a systematic way of building a composite understanding from complex atomic information. I call this procedure "Chunk-Knitting."

Note: I also expand some acronyms and bold key terms the first time they're introduced in the Chunk-Knitting output, but not in the original text.

Introduction and epistemic status

How can we deal with working memory limitations when presenting complex topics? Often, there are too many individual parts to remember easily. Yet the fascination of understanding how they fit together depends on first mastering these atomic parts. It is hard to maintain focus. It seems that learning may be bottlenecked by the student's ability to get through this period of internalizing atomic parts and their individual relationships. The payoff comes at the end, when the student understands how they all fit together into a powerful integrated whole. This should motivate us to find new ways to present information that minimize the memory and attention limitations that students face at the beginning of this process.

We know that spaced repetition, breaking information into chunks, and building connections between atomic concepts are all useful ways to improve memory and integrative understanding. Chunk-Knitting is an attempt to leverage these three tools to systematically create pedagogical text that is easier to understand the first time the reader encounters the ideas it is conveying. This post is a first attempt to apply Chunk-Knitting. The technique has not been experimentally studied, and there is no data supporting the hypothesis that Chunk-Knitting makes information easier to understand.

RNA interference is a naturally occurring method of gene regulation that is also widely used as an experimental technique and has therapeutic potential. Although the basic mechanism can be compressed to the space of a paragraph, the number of details is large enough to exceed a typical person's working memory capacity. It therefore becomes difficult to learn how RNA interference works on a mechanistic level by reading such an efficiently compressed paragraph. Many descriptions of the mechanisms of RNA interference take just this approach to presenting the topic. This makes it an appropriate target for presenting it via Chunk-Knitting.

The basic procedure of Chunk-Knitting is as follows:

  1. Break down the information into component pieces of information and their relationships. An example is a process flow diagram.
  2. From that information breakdown, create a "Chunk Block Schematic." A Chunk contains three pieces of information and their connections. A Block contains three Chunks, along with a concluding summary that describes the Blocks as a whole. The goal is that each atomic concept will be presented in several Chunks, always in a new context using different atomic concept neighbors than it was presented with before. Note: this will be illustrated below.
  3. Convert the Chunk Block Schematic into "decompressed" natural language prose, and end with a final recap in which you rewrite the information in a straightforward paragraph form.




Molecular Mechanisms and Biological Functions of siRNA is a well-cited introduction to siRNA, a key component of RNA interference. We will apply the process of Chunk-Knitting to its highly compressed one-paragraph description. I also include a graphical illustration from another source, which will be recopied with each new approach to RNA interference for the sake of convenient reading. It is dense, and the whole point of this experiment is that it might be difficult to follow. The Chunk-Knitting experiment below is intended to present the same information in a way that's easier to digest at first reading.

Text Of The Original Description

The first step of RNAi involves processing and cleavage of longer double-stranded RNA into siRNAs, generally bearing a 2 nucleotide overhang on the 3′ end of each strand. The enzyme responsible for this processing is an RNase III-like enzyme termed Dicer. When formed, siRNAs are bound by a multiprotein component complex referred to as RISC (RNA-induced silencing complex). Within the RISC complex, siRNA strands are separated and the strand with the more stable 5′-end is typically integrated to the active RISC complex. The antisense single-stranded siRNA component then guides and aligns the RISC complex on the target mRNA and through the action of catalytic RISC protein, a member of the argonaute family (Ago2), mRNA is cleaved.


Step 1: break down natural language into an initial schematic

The first step in Chunk-Knitting is to break down the original paragraph into a process flow diagram. It is not necessary to follow exactly how I produced the following schematic from the former paragraph - I just wanted to give an example of what this schematic might look like. You can compress the information in whatever way makes sense to you.

Dicer -> dsRNA -> siRNA -> 2nt difference in complementary strand lengths -> RISC -> siRNA strands separated -> strand with stable 5’ end integrated -> antisense ss-siRNA guides RISC to target mRNA -> Ago 2 cleaves

Step 2: form a Chunk Block Schematic

The next step in Chunk-Knitting is to expand the compressed schematic from step 1 into Chunk Blocks. I start with the components that are most important to the topic, or most likely to be familiar to the reader. Here, I expect that readers interested in siRNA know about mRNA and how it plays into protein production. Since they're reading about siRNA, I select these three pieces of atomic information to form the first Chunk.

Block 1

siRNA -> cleaves mRNA -> prevents protein production.

siRNA -> binds mRNA -> RISC cleaves

RISC -> Ago2 is a component -> Ago2 cleaves mRNA

Summary: siRNA -> binds mRNA -> RISC -> Ago2 is a component -> Ago2 cleaves mRNA

Block 2

Dicer -> dsRNA -> siRNA

siRNA -> binds RISC -> binds and cleaves mRNA

siRNA -> RISC -> Ago2

Summary: Dicer -> dsRNA -> siRNA -> binds RISC -> Ago2 -> binds and cleaves mRNA

Block 3

Dicer -> dsRNA -> 2nt difference in strand length

Stable 5’ end -> RISC -> binds mRNA

Dicer -> RISC -> Ago2

Summary: Dicer -> dsRNA -> 2nt difference in strand length -> stable 5’ end -> binds RISC -> binds mRNA -> Ago2 cleaves mRNA

Step 3: convert to decompressed natural language prose

The following is how we might convert the Chunk Blocks above into natural language prose. This is what we might present to the reader interested in siRNA. The act of producing it can also be very helpful in installing the information in your own mind. The illustrative diagram from above is reproduced for convenience.

Result of Chunk-Knitting Procedure

Small interfering RNA (siRNA) is a special type of RNA that helps interfere with messenger RNA (mRNA) and prevents it from being used for protein production. The role of small interfering RNA is to bind a specific messenger RNA sequence. Then a protein complex called RISC cleaves that messenger RNA molecule. The part of the RISC complex responsible for catalyzing messenger RNA cleavage is called Ago2.

  • Summary 1: siRNA binds mRNA, while a part of the RISC protein complex called Ago2 cleaves the mRNA.

Dicer is another important protein in this process. Its role is to cleave double stranded RNA into small interfering RNA. The small interfering RNA binds the RISC complex, which then allows RISC to bind and cleave a specific type of messenger RNA. So the key components of the active RISC complex are small interfering RNA and Ago2.

  • Summary 2: Dicer cleaves double-stranded RNA into siRNA, which binds RISC, allowing Ago2 to cleave a specific mRNA.

When Dicer cleaves double-stranded RNA, it divides it into smaller pieces of double-stranded RNA. Each piece of the double-stranded RNA that Dicer produces usually has a longer strand and a shorter strand, with a two-nucleotide difference in length. The long and short strands separate within RISC, and whichever of the two has the more stable 5’ end binds RISC. It hybridizes with mRNA for which it is complementary, and guides RISC into close proximity. So Dicer chops up the small interfering RNA and feeds it into RISC, where it directs Ago2 to cleave a specific mRNA sequence.

  • Summary 3: Dicer cleaves double-stranded RNA into pieces, each double-stranded piece having a two nucleotide difference length between its two complementary single strands. A single strand from one of these pieces, having a more stable 5’ end than its complementary partner, binds RISC. This siRNA strand then directs RISC to bind complementary mRNA, which Ago2 cleaves.

Final Recap 

RNA interference is a process for preventing a specific messenger RNA (mRNA) from serving as a template for protein production. This is accomplished by cleaving the messenger RNA. An activated multiprotein complex called RISC performs this function. It contains a single stranded RNA molecule, the small interfering RNA (siRNA), which is complementary to a specific messenger RNA. It also contains the catalytic protein Ago2, which catalyzes messenger RNA cleavage. The small interfering RNA guides RISC, including the Ago2 component, to the mRNA, where Ago2 cleaves it and destroys it. To produce the small interfering RNA, a separate protein called Dicer cleaves longer double-stranded RNA (dsRNA) into smaller double-stranded fragments. Each double-stranded fragment is composed of two complementary single strands, with one usually being two nucleotides longer than the other. Of each pair of complementary single strands from a particular fragment, one will have a more stable 5’ end than the other. The two strands separate within RISC, and the more stable of the two binds RISC to guide it to the target mRNA, so that Ago2 can catalytically cleave and destroy it.

Future Directions

After executing the Chunk-Knitting procedure, it is difficult as the writer to know if the resulting text is really easier to understand or not, because you are no longer reading it for the first time. To explore this method further, it would be necessary to produce more examples and run an experiment to see which version of the text - the original compressed paragraph or the Chunk-Knit expanded version - promoted greater comprehension and focuse in first-time readers. I can only report at this time that the process of Chunk-Knitting seemed to help me build a much greater understanding of the material than I had when I first read it. It is time-consuming, and there may be more efficient ways to build that understanding - although any such alternatives might have their own tradeoffs. Chunk-Knitting is a tool I'm excited to keep exploring and add to my toolkit for writing and researching.


New Comment
1 comment, sorted by Click to highlight new comments since: Today at 9:18 AM

FWIW I wasn't previously familiar with the topic (some background biology knowledge, but not about this in particular) and the chunked version did seem much clearer than the original.

Although I'm uncertain exactly how much of that clarity came specifically from the chunking, versus other changes like including more definitions of terms.

New to LessWrong?