Attention as Dynamic Routing

What It Is

A phone notification can displace the function you were writing. A thought about another project can compete with the document in front of you. In each case, some information changes which process receives attention.

Context-aware dynamic routing models that selection. The current state determines which information affects which processes, with the connections changing as new inputs arrive. Unlike a discrete conditional branch in a program, attention routes continuously and probabilistically.

This is a functional account of what attention does, not a claim about its neurological implementation. It has been useful in N=1 practice because it replaces “I have bad focus” with a more specific description: competing default patterns are directing attention elsewhere. Its standard is practical usefulness. If the model identifies parts of the arrangement that can be changed, it has done its job.

These are competing destinations for attention in the article's functional model, not a diagram of neural anatomy.

Read this diagram

Current context branches into The chosen task; An interruption; A familiar default.

The Core Insight: Attention is Routing, Not Mystery

Computing has studied deterministic models of attention since the 1970s through parser grammars and production rules. These systems match patterns according to state and use meta-rules to determine which pattern applies. Context-dependent selection is a familiar computational operation.

A conditional branch routes information discretely:

if condition:
    path_A()
else:
    path_B()

A boolean sends execution down one path. Attention mechanisms instead combine several relevant inputs:

for each query:
    relevant_parts = find_relevant_keys(query, all_keys)
    weighted_sum = combine_values(relevant_parts)

Many paths contribute at once, weighted by relevance. The underlying operation is the same: context determines which parts of the system affect other parts. “Context-aware” means the selection depends on the current state and history. “Dynamic” means connections change with inputs. “Routing” names the resulting distribution of influence.

Observable Routing Failures

Five patterns account for nearly every focus failure in this model:

The diagnosis changes the available response. “I can't concentrate” names a failure without explaining what to alter. “The default route has a lower activation cost than the intended route” identifies a competitor that can be blocked. Other failures call for externalizing competing thoughts, defining a target, or removing an interrupt source. Focus failures are routing failures with parts that can be changed.

Orders of Attention

Using attention, observing it, and changing the conditions that direct it are different operations.

First-Order: Deploying Attention

First-order attention is the moment-to-moment allocation of resources: writing a function, reading documentation, or listening in a meeting. The question is what receives focus now. This is what most people mean by attention and the only level at which most operate.

Second-Order: Monitoring Attention

Second-order attention observes where focus goes and why over minutes and hours. You may notice that work has given way to thoughts about lunch, or that you have checked the phone five times in ten minutes. The observation identifies competing patterns and default responses as they interrupt the intended task.

Third-Order: Designing Attention Strategies

Third-order attention changes the conditions that govern those shifts. Over days or weeks, it creates a focus environment, designs an information diet, installs prevention architecture, and removes interrupt sources.

Putting the phone in a drawer changes the available route instead of repeatedly resisting it. The initial cost is high, but the ongoing cost is near zero. Most people try to focus and occasionally notice that they have become distracted; almost no one changes the architecture. That third level provides leverage because the desired behavior can become the default instead of requiring continuous first-order effort.

Observable Patterns in Behavior

Pattern 1: Phone Hijacks Routing (Default Route Dominance)

Phone-checking is an established response with almost no starting cost. A visible phone takes a fraction of a unit to check, while resisting each impulse costs 2-3 units. At fifty-plus notifications a day, resistance alone can exceed the entire willpower budget.

Thermodynamics selects the lowest-cost option, so the phone wins against work every time. Prevention architecture changes the costs: a closed drawer removes the visible cue and makes checking a deliberate physical act. The default phone route is blocked, so work wins without a continuing resistance cost.

Pattern 2: Fragmented Attention (Competing Routes)

A project, another project, a social event, and unanswered emails can all remain active while you try to work. These patterns occupy working memory, whose limit is 4-7 items. A thought about project B interrupts task A, then an email, then project C's deadline. Seven competing patterns prevent one from holding the router.

A braindump records every active concern. Assigning each to a time slot removes its demand to be handled now. One target is then chosen for the current block, reducing seven-plus active patterns to one while the others remain in external storage.

Pattern 3: No Clear Route (Ambiguous Routing)

“Work on project” does not identify which part or which action. It is too abstract to match an existing execution pattern, leaving attention without a target. The result is a freeze or a return to a familiar alternative such as the phone.

Discretization and a bounded question supply the missing target. “What's the next 25-minute chunk?” reduces the project to an action. “Write the test for login validation” reduces “implement feature” to one function that can receive attention.

Pattern 4: Route Hijacking (External Interrupts)

A notification, person, or sound can override the active task. The visible interruption may last a moment, but resuming requires processing the interruption and reloading the working-memory state. Each interruption costs five to ten minutes in that reload.

This is push-based Beta architecture: the world can write to attention whenever it chooses. Alpha architecture makes information available by request instead. Notifications are turned off entirely, rather than silenced; the phone moves to another room; email is checked on a schedule. Information is retrieved when needed, and nothing pushes an interruption into the task.

Higher-Order Structures (Speculative)

The model also suggests further capabilities. These are more speculative than the three orders above.

Meta-Meta Rules: Pattern Discovery Methods

Using a technique and observing whether it works both operate within an existing set of techniques. Pattern discovery develops a new one from observations.

Will's three-hour morning work window came from weeks of tracking focus. The repeated 6am-9am peak suggested the strategy; it was not copied from a productivity book. Second-order observation supplied the data for third-order design. This distinguishes discovering a useful pattern from recognizing an already-known one.

Self-Modifying Attention (Attention on Attention)

Attention can be directed toward its own operation: what it is attending to, why, whether the strategy should change, and how the conditions could be redesigned. Braindumping before work is an everyday instance. Moving pending concerns to external memory changes the set of patterns competing for attention and prepares a single focus target.

Causal Discovery (Not Just Correlation)

“I focus poorly after social media” records a correlation without specifying what to change. “Social media creates competing patterns that fragment attention for thirty-plus minutes” identifies a cause and an intervention: block Instagram during work hours so those patterns are not introduced before the block. Correlation records what occurs together; a causal account identifies the condition to alter.

Practical Applications

Application 1: Debugging Focus Failures

The failure table provides five questions. Is a familiar option such as phone, email, or YouTube cheaper than the intended task? Are several concerns competing? Is the target ambiguous? Is an external signal interrupting? Does the task lack a recognizable entry point?

The corresponding interventions are to block a default, externalize concerns, bound the question, remove push channels, or discretize the task. Every focus failure has a routing diagnosis and an architectural response. More discipline is not required by the protocol.

Application 2: Designing Focus Architecture

Design begins with observation. For a week or two, record when focus breaks, what captures it, which thoughts intrude, and what triggers a return to familiar defaults. Also record when focus happens naturally and the conditions present at those times.

The observations inform three kinds of change. First, remove interruptions: put the phone off and out of sight, disable notifications, and block distracting websites during work. Second, separate competing contexts: use different work and social locations, assign email to 4-5pm and work to 8am-12pm, and give each block one project while externalizing the others. Third, create specific targets: divide work into 25-minute chunks, open each block with a bounded question, and make completed chunks visible.

The architecture is then run for 30 days. Sustained minutes per block measure focus quality, and remaining failures identify adjustments. After thirty days, the new arrangement becomes the default.

Will's N=1 case began with fifty-plus phone interruptions a day, notifications breaking flow, and social media introducing competing thoughts. His changes were a phone turned off and placed in the drawer during 6am-12pm work, deleted social apps, notifications disabled on every device, and email restricted to 4-5pm. With those routes removed, work won by thermodynamics and focus no longer needed continuous resistance.

Application 3: Meta-Awareness Practice

Second-order practice records routing failures for later redesign. When attention shifts from task A to the phone, email, or another thought, the useful observation is what captured it: an external interruption, an internal concern, an automatic default, or an unclear task. The log records that cause, the time, and the intended focus.

After a week or two, recurring failures become visible. A phone interruption at 10am every day suggests locking it away from 8 to 12. Intruding email thoughts suggest a scheduled email window. Repeated escapes from ambiguous work suggest finer discretization. Monitoring produces the evidence; third-order design uses it to change the next block.

Framework Integration

Connection to Focus

Focus is the intended routing pattern remaining stable over time. Distraction occurs when a competitor takes over. A focus environment makes the intended route the lowest-cost option: phone removal blocks the default competitor, disabling notifications prevents interruption, and the environment makes work the thermodynamic default.

Connection to Working Memory

The 4-7-item limit of working memory caps how many patterns can remain active. Once too many compete, none wins cleanly. Willpower cannot make ten simultaneous targets fit a biological capacity limit. Externalizing competitors with a braindump is the only response that respects that constraint.

Connection to Question Theory

Questions initiate searches. “What should I focus on?” opens an infinite possibility space that never finishes searching, producing paralysis or arbitrary selection. “What's the next 25-minute concrete action on the highest-priority task?” opens a small space that can be searched immediately. The bounded question supplies a route attention can follow.

Connection to Information Architecture

Information architecture determines how information enters the system. Push-based Beta channels deliver feeds, alerts, and notifications on their schedule, causing fragmentation and context switches. Pull-based Alpha channels are queried when information is needed for the current task.

The choice is between removing push sources once and resisting each interruption indefinitely. The latter is expensive and fails.

Connection to Prevention Architecture

Prevention blocks a route before it activates. Resistance attempts to stop it after activation. Removing a phone has a one-time cost; leaving it visible creates a cost of a few units per resisted impulse, repeated dozens of times until willpower is depleted.

Connection to The Braindump

Braindumping manages the active patterns directly. Seven-plus concerns compete before externalization; afterward, the others remain on paper while one target occupies active memory. Clearing that competition allows attention to remain on the chosen task.

Common Misunderstandings

Misunderstanding 1: "Attention is Willpower"

Treating attention as willpower makes weak focus evidence of weak will. The routing model instead changes the environment so the desired route becomes the default and needs no effort. Removing the phone makes work win by thermodynamics without repeated resistance.

Misunderstanding 2: "I Have Bad Focus"

Visible phones, active notifications, ambiguous tasks, and seven-plus competing projects describe a specific architecture. Prevention, externalization, and discretization change those conditions. Calling the same arrangement a distractible character gives no comparable intervention.

Misunderstanding 3: "Just Be More Mindful"

Monitoring and intervention perform different jobs. “I notice I'm checking my phone a lot” records a pattern. “The phone goes in the drawer during work blocks” changes the condition producing it. Observation without redesign creates awareness without change; both are necessary.

Misunderstanding 4: "This is Literal Neuroscience"

The model does not assert that neurons implement computer routing. It describes a function in a way that makes interventions visible: block a path, remove competing patterns, or specify a target. Its usefulness depends on finding changes that work, regardless of neurological accuracy.

Related Concepts

• Focus is sustained attention supported by routing design.
• Working Memory limits simultaneous active patterns.
• Question Theory distinguishes bounded from unbounded routing triggers.
• Information Architecture determines push and pull flows.
• Signal Theory names Alpha pull and Beta push arrangements.
• Prevention Architecture blocks routes before activation.
• The Braindump externalizes competing patterns.
• Willpower is spent resisting active routes.
• Discretization creates specific targets through chunking.
• 30x30 Pattern describes strengthening through repetition.
• Causality Programming treats routing as control of causal flow.
• Execution Resolution places attention at the level where intervention is possible.
• State Machines describes routing patterns as transitions.
• Activation Energy determines which routes win.

Key Principle

The routing model turns a focus failure into questions about the available inputs, their costs, and the target of the current task. Monitoring identifies the failure; changing the architecture makes the intended route the default. The model remains functional rather than neurological, and its test is whether those changes improve the behavior.