Block Annotations & Memory

Chunking at the structural level.

George Miller's 1956 paper "The Magical Number Seven" established that working memory can hold roughly 7 ± 2 items simultaneously. But what counts as an "item" is not fixed — through a process called chunking, multiple pieces of information can be grouped into a single unit, effectively expanding the capacity of working memory.

CLN block annotations are a chunking mechanism at the document level. A block with a header // 03 · Risk Analysis becomes a single retrievable unit — not five separate lines. The number (03) gives it ordinal position in a sequence. The title gives it semantic identity. The visual separation gives it spatial distinctness. Together, they make the block a mnemonic anchor that organises recall at multiple levels simultaneously.

CLN v2.0

CONCEPT: Block Annotations as Memory Structures

how CLN // NN · Title blocks activate chunking, spatial, and episodic memory

#chunking

#episodic-memory

#spatial-memory

// 01 · Three Memory Systems Activated by Block Structure

Working Memory — Chunking

a block with a header is processed as one chunk, not N lines

reader can hold "block 03 — Risk Analysis" in working memory, not all its lines

cognitive load reduced: navigate at block level, drill into line level only when needed

see: Cognitive Load

Spatial Memory — Block as Location

each block occupies a distinct spatial region on the page

numbered headers create ordinal positions: "block 2 is above the middle"

place cell networks encode the block's position — retrievable as location

see: Spatial Memory

Episodic Memory — Block as Event

each block was written at a moment in time — linked to the authoring episode

"I wrote that risk block after the meeting with X" — episodic tag

episodic context aids retrieval: "when was I thinking about this?"

// 02 · Anatomy of a Memorable Block Header

the full format: // NN · Semantic Title

Component: //

universal "comment" signal — visually distinct, universally recognised

pre-attentive: processed before reading the title

Component: NN — Ordinal Number

encodes position in sequence: "this is the third thing"

enables navigation by number: "go to block 4"

numbers are processed by a different neural pathway than words — double encoding

Component: · — Middle Dot Separator

visual pause between number and title — aids parsing at a glance

consistent pattern makes the entire format pre-attentively recognisable

Component: Semantic Title

the semantic hook: "Risk Analysis", "Architecture", "Open Questions"

activates semantic memory network associated with that concept

3+ memory systems fire simultaneously on a single line

// 03 · Visual Separation and Episodic Boundary Effects

Event Segmentation Theory (Zacks, 2007)

the brain segments continuous experience into discrete events

event boundaries (when something changes) are stored as memory landmarks

more boundaries → more landmarks → better episodic memory for the sequence

CLN Blocks as Artificial Event Boundaries

the whitespace and header between blocks is an event boundary

each block transition triggers an episodic landmark in memory

reader builds a richer episodic trace when reading CLN vs continuous prose

too many small blocks can create noise — optimal block count: 4–8 per document section

// 04 · Practical Recommendations

For Maximum Memorability

keep block titles semantically distinctive — "Risk Analysis" beats "Section 3"

keep blocks 5–12 lines — short enough to chunk, long enough to be meaningful

use consistent block sequencing across documents — same topic = same block position

e.g. // 01 · Context, // 02 · Analysis, // 03 · Risks, // 04 · Actions — always

consistent templates reduce cognitive load and build retrieval habits

For AI Processing

numbered blocks give AI an unambiguous navigation target: "check block 03"

labelled blocks enable cross-document citation without ambiguity

The room remembers what happened in it.

Memory researchers have long known that context is integral to memory — not a tag added after the fact, but part of the memory itself. The method of loci works because spatial context becomes entangled with content during encoding. CLN block annotations create this entanglement deliberately: number, title, visual position, and episodic boundary all become part of the memory trace for every block's content.

This is the deeper function of the // annotation: it is not metadata. It is the door to a room in a palace you are continuously building as you write. Every block you annotate is another room you can return to.

When you write // 03 · Risk Analysis, you are not labelling content. You are building an address that your memory will use to find it.

← Spatial Memory Pre-attentive Processing →