Pre-Attentive Processing
The visual system classifies colour, shape, and spatial position in under 200ms — before conscious attention engages. CLN's operator column is architected to exploit this mechanism, turning semantic triage from a serial reading task into a parallel visual scan.
The phenomenon — seeing before looking
Spend a moment with this question before reading further: how long does it take you to find the red item in a field of blue items? The answer is roughly the same whether the field contains ten items or two hundred. The red one pops out. You do not search — you simply see it.
This is pre-attentive processing. The visual cortex runs a massively parallel analysis of the entire visual field before the spotlight of conscious attention has moved to any individual location. Features processed this way — colour, orientation, size, motion, certain shapes — are detected across the whole field simultaneously, with processing time independent of the number of distractors.
Pre-attentive processing — the automatic, parallel analysis of basic visual features (colour, orientation, size, closure, motion) that occurs across the entire visual field in approximately 100–200ms, prior to and independent of serial focused attention.
Contrasted with attentive (serial) processing, which scans the visual field sequentially — examining one location at a time under the control of executive attention. Serial processing time scales linearly with the number of items; pre-attentive processing does not.
The discovery — formalised in Anne Treisman's Feature Integration Theory (1980, with Gelade) — was that the visual system has two distinct modes. The first is fast, parallel, effortless, and bounded to single features. The second is slow, serial, effortful, and required for conjunctions — combinations of features that define specific objects. Finding the red circle among blue circles and red squares requires serial search; finding the red item among blue items does not.
The practical consequence for information design is radical: if you encode the information you most need to find in a pre-attentive feature, finding it becomes essentially free. It pops. If you encode it in the content of a sentence, finding it costs linear time across every sentence until you reach it.
The pre-attentive feature channels
Not all visual properties are pre-attentive. Treisman's original taxonomy and subsequent work by Julesz (1981), Wolfe (1992), and others established which features are processed in the parallel stage and which require attention.
| Feature | Pre-attentive? | Strength | Example |
|---|---|---|---|
| Colour (hue) | ✓ Strong | Pop-out at any field size. Fastest channel. | Red item in blue field detected in ~100ms regardless of N |
| Orientation | ✓ Strong | Tilted line among vertical lines; pop-out if angle > ~18° | Diagonal slash in a field of verticals |
| Size / length | ✓ Moderate | Pop-out degrades as size difference decreases | Long line among short lines |
| Curvature / shape category | ~ Partial | Curved vs straight pop-out; complex shapes require serial search | Circle among squares, but not pentagon among hexagons |
| Spatial position (region) | ✓ Strong | Items in a consistent spatial region are localised instantly | Left column always vs right column — eye knows where to go |
| Motion | ✓ Very strong | Moving item among stationary; fastest pop-out channel | Not applicable in static documents |
| Luminance / brightness | ✓ Moderate | Bright item among dim; context-dependent | White text on dark background in a dark-text field |
| Feature conjunction | ✗ Not pre-attentive | Red circle among red squares and blue circles requires serial search | Bold italic — requires attention, not pre-attentive |
| Alphanumeric identity | ✗ Not pre-attentive | The letter Q among Os requires serial attention unless size/colour differs | Keywords in prose body text — always requires serial scan |
The bottom two rows are the most important for document design. Text content — the actual words — is never pre-attentive. Reading requires serial attentive processing. There is no shortcut. The word "DANGER" embedded in the middle of a paragraph is not detectably faster to find than the word "marginal" — both require reading every word until you reach them. If you want the danger signal to pop, it cannot live in the text.
Why prose fails at urgent information
Standard documentation is almost entirely built on alphanumeric content — the one channel that is definitively not pre-attentive. The sole exception is bold and italic formatting, which adds a mild luminance/size differential, but bold text scattered through prose is a feature conjunction (bold + specific word), which requires serial search. The visual system cannot find it without reading.
The prose version hides the risks as efficiently as if they were deliberately obfuscated. A reader in a hurry — the exact situation when risks matter most — will likely miss at least one. The CLN version makes missing them nearly impossible without closing one's eyes.
CLN's operator column — a pre-attentive display
CLN's operator column is not decorative. It is a pre-attentive display layer that runs parallel to the content layer. The two layers serve different processing stages: the operator column is processed in the parallel pre-attentive stage; the content is processed serially in the attentive stage. A reader approaching a CLN document uses both stages simultaneously — the operator column orients them before their attention has finished processing the first word.
Each operator fires three independent pre-attentive channels simultaneously: colour, shape, and spatial position. This redundancy is not accidental.
A reader in low-contrast conditions (bright ambient light, degraded screen) may lose the colour channel — but the shape and position channels survive. A reader who is colour-deficient retains the shape and position channels. A reader scanning very fast may catch only the positional regularity — enough to know which column holds the operator.
Three channels also produce faster pop-out than one. Each additional distinctive feature increases the salience of the target in a conjunction display. The ! is not just red — it is red, exclamatory, and positioned at the left margin. All three features point at each other.
Search asymmetry — why ! finds you, not the other way round
One of the most robust findings in pre-attentive processing research is search asymmetry (Treisman & Souther, 1985): the ease of finding a feature-rich target among feature-poor distractors is not symmetric with the reverse. Finding a circle with a line through it among plain circles is fast — the line is the distinctive feature that pops. Finding a plain circle among circles-with-lines is slow — the absence of a feature is not pre-attentive.
The implication for CLN design is direct: the ! operator works because it is a presence in the operator column — a distinctive feature-rich symbol against a field of other operators. Bold text in prose works poorly because it is defined by an absence compared to the surrounding text — readers must notice what is not there. The visual system does this badly under time pressure.
Spatial position — the most underappreciated channel
Of the three channels active in CLN's operator display — colour, shape, position — spatial position is the most powerful for sustained reading, as opposed to single-target pop-out. Colour pop-out is fast but fatigues; spatial regularity is fast and does not fatigue.
The reason is that spatial position is processed by a dedicated retinotopic mapping in the early visual cortex, prior to feature binding. The visual system maintains a continuously updated spatial map of the entire field. Regular structure — items always at the same horizontal position — exploits this map. The eye learns after 2–3 lines that operators live at the left margin, and begins pre-positioning its saccades before it has finished reading the current line.
The pre-attentive benefit of CLN's operator column depends entirely on spatial regularity. The operators must always be at the same horizontal position, with consistent indentation for child lines. If operators appear at varying horizontal positions — sometimes after a label, sometimes inline, sometimes preceded by bullet spacing — the spatial channel is broken and the reader must use serial attention to locate each operator.
This is why CLN renders best in a monospaced font and why the operator must always be the first non-whitespace character on its line. The spatial contract between reader and document is: the left margin is always the operator column. That contract enables pre-attentive navigation.
Colour assignment — why the specific hues matter
CLN's operator colours are not arbitrary. They are assigned to maximise discriminability across operators that most need to be distinct from each other, while respecting the hue categories that the pre-attentive system reliably separates.
highest salience · threat detection system pre-calibrated to red
distinct from red · signals unresolved state without threat urgency
cool + distinct from blue · ambiguity hue in most colour systems
go / verified / positive outcome · universal association
neutral · action without emotional valence · workhorse operator
warmth · knowledge and naming · anchor colour
muted · subordinate · context recedes behind content
structural · different from result green · administrative
instrumental · lower salience than action · dependency signal
+ and − use semantic colour associations directly
The critical pairings are ! and ~: both are in the warm spectrum, but ! is clearly red (threat) and ~ is clearly pink/magenta (tension without immediate danger). A reader who distinguishes only by colour — not shape — reliably separates these two. A reader who distinguishes only by shape — the exclamation point versus the tilde — also reliably separates them. The redundancy across channels makes the distinction robust to degraded viewing conditions.
Graceful degradation — when pre-attentive channels break
Pre-attentive display design must account for failure modes: not all readers see colour, not all rendering environments preserve colour, not all contexts allow visual scanning. CLN is designed to degrade gracefully — to preserve meaningful structure even when one or more pre-attentive channels are unavailable.
| Degradation | Channel lost | What breaks | What survives |
|---|---|---|---|
| Monochrome / printed | Colour | Operator discrimination by colour only | Shape channel fully intact · Position channel intact · All 10 operators remain distinguishable by symbol |
| Red-green colour blindness (~8% male) | ! vs → distinction by hue | Red/green discrimination — ! and → may appear similar hue | Shape channel distinguishes ! from → trivially · no semantic confusion survives |
| Plain text / terminal / no font rendering | Colour + shape rendering | Colour · Some symbol rendering (→ may become -> or →) | Position channel intact · Symbol identity intact · Semantic structure readable |
| Very fast scan (1–2 sec total) | Content (serial) | Any information in body text | !! lines detected by position + colour · Operator distribution readable · High-level document shape known |
| Stressed / cognitively loaded reader | Serial attention capacity | Long prose explanations · Nested rationale chains | !! lines pop · Action sequence visible · Critical path extractable without reading * children |
The last two rows are the most important from a safety standpoint. CLN documents are often consulted in exactly the high-stress, time-compressed states where cognitive load is highest and serial attention capacity is lowest. The pre-attentive design is not a nicety for leisurely reading — it is a critical path redundancy for exactly the conditions where the information matters most.
What breaks pop-out — the CLN design violations
Understanding pre-attentive processing makes the most damaging CLN anti-patterns obvious. They are not stylistic mistakes — they are violations of the pre-attentive contract that destroy the document's ability to be scanned.
Connection to dual coding and memory
Pre-attentive processing is the input stage of a sequence that connects to memory encoding. When the operator ! fires the pre-attentive colour and shape channels, it does two things simultaneously: it directs attention to the risk line for serial processing, and it activates the amygdala's threat-detection response through the low-road subcortical pathway.
The amygdala activation is not metaphorical. Colour in the red spectrum activates threat-processing circuitry before the content of the line has been processed. The content of the risk arrives into a neural state slightly primed for threat-relevant processing — which means it consolidates more durably than neutrally-presented information of equal importance.
This is the same mechanism described in the Mnemotechniques article — emotional tagging as a retention amplifier. CLN's ! operator exploits this involuntarily and consistently. The reader does not need to decide to pay attention to risk lines. Their visual system has already flagged them before the decision could be made.
The dual coding connection is equally direct. Each CLN operator line encodes the same semantic content through two independent pathways: the symbol and colour (visual-imagistic encoding) and the text content (verbal-propositional encoding). Allan Paivio's dual coding theory (1971) established that these two pathways are stored independently — either alone can cue recall of both. The operator becomes a retrieval handle that can pull the content even when direct verbal recall fails.
The layered design principle: CLN's operator column is simultaneously a pre-attentive display layer (directing attention in under 200ms), a dual-coding layer (creating independent visual and verbal memory traces), an emotional tagging layer (!! lines activate mild amygdala priming), and a spatial regularity layer (training the eye's saccade planning after 2–3 exposures). These effects stack — a well-structured CLN document is easier to scan, easier to remember, harder to miss critical information in, and faster to navigate than prose on every single dimension simultaneously. None of the effects require reader effort. They are all structural.