Core Primitive
Some chains need conditional branches — if X then chain A else chain B.
The chain that only works five days a week
For two months, Nadia's weekday morning chain ran like a machine. The alarm sounded at 6:00. Her feet hit the floor and landed in the running shoes she had placed there the night before. She walked to the bathroom, splashed water on her face, filled her water bottle, and was out the front door by 6:10. A three-block walk warmed her into a twenty-five-minute run along the route she had memorized so thoroughly that her legs followed it without her brain's involvement. She returned home, stretched on the porch, showered, made coffee, and sat at her desk by 7:15. Not once in eight weeks did the chain misfire on a Tuesday or a Thursday. Not once did she have to negotiate with herself about whether to run.
Then Saturday arrived, as it does, and the chain fell apart.
The alarm sounded at 6:00 — same cue. Her feet hit the floor — same first link. She put on her running shoes — same second link. And then she stalled. Her running route passed the elementary school, which on weekdays was quiet at 6:10 but on Saturdays hosted a farmers market that clogged the sidewalks by 7:00. Her partner was still asleep and would not appreciate the front door opening and closing. The kids would be awake by 7:30 wanting breakfast, and her weekday post-run schedule — shower, coffee, desk — did not account for scrambled eggs and cartoons. She stood in the hallway in her running shoes, fully triggered, with nowhere to go. The chain had fired but every link after the second one assumed a context that did not exist on Saturday.
What happened next was predictable. She tried to improvise — maybe a home workout? She opened YouTube, scrolled through options, realized she did not have a yoga mat downstairs, went looking for one, got distracted by the laundry she had meant to fold, and by 7:15 had done neither exercise nor laundry. The chain had not just stalled. It had disintegrated into open-ended deliberation, which consumed the very time and energy the chain was supposed to protect.
Nadia did not have a motivation problem. She had an architecture problem. Her chain was linear — a single path designed for a single context. And real life is not a single context. The solution was not to build more discipline for Saturdays. It was to build a branch.
Why linear chains break in variable contexts
Every chain covered so far in this phase — morning chains (Morning chains), work startup chains (Work startup chains), shutdown chains (Shutdown chains), exercise chains (Exercise chains) — has been presented as a linear sequence: cue triggers link one, link one triggers link two, and so on until the chain completes. This linearity is a feature when the context is stable. On a weekday morning where the same alarm fires in the same room with the same schedule ahead, a linear chain is beautifully efficient. It converts a dozen micro-decisions into a single automated flow.
But linearity becomes a liability when the context changes. Mornings differ between weekdays and weekends. Workdays differ between office days and remote days. Exercise differs between gym-available days and traveling days. Energy levels differ between well-rested mornings and post-insomnia mornings. A linear chain that assumes one context will misfire in another — not because the chain is poorly designed, but because it was designed for a world with less variability than the one you actually inhabit.
The instinct, when a chain fails in a new context, is to build a completely separate chain for that context. A weekday morning chain and a weekend morning chain, designed independently, with different triggers, different links, and different endpoints. This sometimes works, but it doubles the cognitive infrastructure you need to maintain, and it creates a new problem at the junction: which chain do I fire? If the two chains share nothing, the moment of choosing between them becomes its own decision point — the very thing chains are designed to eliminate.
The better architecture is a branching chain. A branching chain shares a common trigger and common anchors (more on anchors in Chain anchors) but includes a decision node — a single, pre-committed if-then that routes the sequence down one of two or three pre-designed paths. The key word is pre-designed. Each branch is fully specified before the chain fires. The decision node is not a moment of open deliberation ("What should I do this morning?") but a simple conditional check ("Is it a weekday?"). The answer determines the path, and the path runs automatically from there.
This is programming logic applied to behavior. In software, a conditional branch is trivial: if (condition) { pathA() } else { pathB() }. The program does not deliberate. It checks a variable, selects a path, and executes. A branching chain works identically. The decision variable is observable and unambiguous — day of the week, physical location, amount of time available. The branch paths are pre-loaded, fully specified, requiring no further choices. The only cognitive cost is the check itself, and when the variable is obvious (you know what day it is), even that cost approaches zero.
The research behind conditional behavior
The most direct research foundation for branching chains comes from Peter Gollwitzer's work on implementation intentions. Gollwitzer, a psychologist at New York University, introduced the concept in 1999 and has spent over two decades refining it. An implementation intention takes the form "If situation X, then I will do behavior Y" — a pre-committed conditional that links a specific situational cue to a specific behavioral response (Gollwitzer, 1999). The critical finding, replicated across hundreds of studies, is that implementation intentions dramatically increase the likelihood of follow-through compared to mere goal intentions ("I want to exercise more"). The mechanism is specificity: by pre-deciding the response to a particular situation, you offload the decision from the moment of action to the moment of planning.
What makes Gollwitzer's framework directly relevant to branching chains is that implementation intentions are inherently conditional. They do not say "I will exercise." They say "If it is Tuesday and I am at home, I will do the bodyweight circuit" and "If it is Tuesday and I am traveling, I will do the hotel-room routine." Each if-then pair is a branch. A branching chain is, in effect, a set of implementation intentions that share a common trigger and diverge based on a situational variable. Gollwitzer and Brandstatter (1997) showed that participants who formed specific if-then plans completed difficult goals at roughly twice the rate of those who formed only goal intentions. The branching structure is what allows the if-then format to handle real-world variability rather than collapsing at the first contextual deviation.
A deeper theoretical frame comes from Roger Schank and Robert Abelson's script theory, introduced in their 1977 book Scripts, Plans, Goals, and Understanding. Schank and Abelson proposed that much of human behavior is governed by "scripts" — pre-learned sequences of actions associated with specific situations. You have a restaurant script (enter, wait to be seated, read menu, order, eat, pay, leave) and a grocery store script (get cart, walk aisles, select items, checkout, bag, leave). These scripts are context-specific: the restaurant script does not fire in the grocery store. But many life situations are ambiguous enough to require script selection — a process Schank and Abelson called "script activation." When you walk into a place that could be a cafe or a bar, your brain briefly assesses which script to activate. A branching chain formalizes this process. Instead of leaving script selection to an unconscious, sometimes unreliable assessment, you explicitly define the activation conditions: if the variable reads this value, activate this script.
Wendy Wood's research on context-dependent habits provides the third leg of the empirical foundation. Wood has demonstrated repeatedly that habits are not just behavior patterns — they are behavior patterns bound to specific contexts (Wood & Neal, 2007). Change the context, and the habit fails to fire. This is why people who move to new cities often lose their exercise habits even when their motivation remains unchanged: the contextual cues that triggered the behavior no longer exist. A branching chain addresses this vulnerability directly. Instead of building a habit bound to a single context and hoping that context remains stable, you build a habit structure that explicitly accounts for the two or three contexts you actually encounter. The trigger fires in all contexts. The branch node routes to the appropriate path. The habit survives the context change because the context change was anticipated and designed for.
Designing a branching chain
Building a branching chain requires four steps, each one critical. Skip any of them and you end up with either an over-complicated decision tree or a set of vague alternatives that reintroduce the deliberation the chain was supposed to eliminate.
The first step is to identify the variable. This is the single observable factor that determines which branch to take. The most common variables are day of the week (weekday versus weekend), location (office versus home versus travel), available time (full morning versus compressed morning), and energy level (rested versus depleted). The variable must be unambiguous — something you can assess in under two seconds without deliberation. "Day of the week" is unambiguous. "How motivated do I feel?" is not, because it invites self-analysis rather than a quick check. If you find yourself needing more than a moment to evaluate the variable, you have chosen the wrong one. Simplify until the check is instant.
The second step is to design each branch path completely. This is where most people fail. They specify the default branch in detail — the weekday morning chain with every link defined — and then describe the alternate branch vaguely: "On weekends I'll do something lighter." That vagueness is fatal. When Saturday morning arrives and you reach the branch node, "something lighter" requires exactly the kind of open-ended deliberation that chains exist to prevent. Each branch must be specified link by link, with the same level of detail as a linear chain. If your weekday morning chain has seven links, your weekend morning chain needs seven links too — different links, perhaps, but each one named, sequenced, and requiring no in-the-moment decisions.
The third step is to limit the number of branches. Two is ideal. Three is acceptable. Four or more is a decision tree, not a branching chain, and decision trees require the kind of deliberative processing that drains the very cognitive resources you are trying to conserve. If you find yourself designing more than three branches, you have likely conflated two different variables. For example, you might be combining "day of week" and "energy level" into a single branch point, creating paths for "weekday-rested," "weekday-tired," "weekend-rested," and "weekend-tired." The solution is to separate them: first branch on day of week, then within each path, include a secondary branch on energy level — or, better yet, design each path to accommodate variable energy by making the baseline effort achievable on your worst day, as Exercise chains established for exercise chains.
The fourth step is to practice each branch independently before connecting them. This is counterintuitive — you would think the branching chain should be practiced as a whole from the start. But each branch is functionally a separate chain, and chains need repetition to become automatic. If you practice only the weekday branch five days a week and then expect the weekend branch to run smoothly on Saturday, you are asking an unpracticed chain to fire in a context where you have no behavioral momentum. Instead, run the weekend branch deliberately for several sessions — even during the week if necessary — until its links feel as automatic as the default path. Only then does the branch node become a true conditional rather than a transition into unfamiliar territory.
Consider how Nadia solved her Saturday problem. She identified the variable: weekday or weekend. She designed the weekday branch (her existing chain, unchanged) and the weekend branch: alarm at 6:30 (later start), feet on floor, movement clothes on (same as weekday — this is a shared anchor), walk to the living room, twenty minutes of yoga following the same recorded session every time (no decision about which video), finish with the same stretch she did after running (another shared element), then start breakfast with the kids. The trigger was nearly identical. The endpoint converged on the same transition — shower, then first focused activity. The middle diverged based on a single binary check: what day is it? After practicing the weekend branch for three consecutive Saturdays, the branch node became automatic. She no longer stood in the hallway deliberating. She checked the day, and the appropriate path activated.
Branching beyond mornings
The branching chain architecture applies to any behavioral domain where context varies predictably. Work chains branch on location: the office startup chain involves commuting, settling into a desk, and greeting colleagues, while the remote startup chain involves walking to the home office, closing the door, and opening the same applications in the same order. Exercise chains branch on equipment availability: the gym branch includes machines and free weights, while the travel branch includes bodyweight movements and a hotel hallway for walking lunges. Shutdown chains branch on what follows: the weekday shutdown chain (Shutdown chains) transitions to evening routine, while the Friday shutdown chain adds a weekly review step before closing out.
In each case, the architecture is identical. A shared trigger fires. A decision node evaluates a single variable. The appropriate branch activates. Each branch runs as a fully pre-committed, linear sequence. The chain ends at a shared anchor point. The variability in your life is not a threat to the chain — it is accounted for by the chain. You are not improvising. You are routing.
This distinction matters psychologically. When a linear chain breaks because the context has shifted, the experience is one of failure — the chain did not work, and now you must scramble. When a branching chain encounters a context shift, the experience is one of selection — the chain is working exactly as designed, and it is sending you down the path built for this situation. The emotional difference between "my routine failed" and "my routine is handling this" is enormous, even when the behavioral output is similar. The branching chain preserves your sense of agency and control precisely because the variability was anticipated.
The Third Brain
An AI assistant is exceptionally well-suited to helping you design branching chains because it can hold multiple complete paths in working memory simultaneously — something that is cognitively taxing for humans to do during the planning phase. Describe your current linear chain to an AI, then describe the two or three contexts in which that chain needs to operate. Ask the AI to generate a complete branch path for each context, maintaining the same trigger and endpoint while adapting the middle links. The AI can ensure that each branch is fully specified, that no link requires an in-the-moment decision, and that the paths share enough common structure to feel like variations of the same chain rather than entirely separate routines.
The AI can also stress-test your branch design by generating edge cases. What happens if the variable is ambiguous — a holiday that falls on a weekday, a work-from-home day that is supposed to be an office day? These edge cases reveal whether your decision variable is truly binary or whether it needs refinement. Ask the AI to identify every scenario in the next month where the branch point might be unclear, and for each scenario, pre-commit to which branch activates. The goal is to eliminate every possible moment of deliberation at the branch node, so that when the trigger fires, the path selection is as automatic as the chain itself.
Finally, use the AI to audit your branches for unnecessary complexity. Describe all your branch paths and ask whether any two are similar enough to merge. Often, what feels like three distinct contexts can be reduced to two once you strip away superficial differences. The AI can identify the structural similarities you might miss and propose a simplified branching architecture that preserves the necessary variability while reducing the cognitive load of maintaining multiple paths.
The anchor principle
Whether your chain is linear or branching, one structural principle governs its reliability more than any other: the strength of its endpoints. Nadia's branching chain worked because her weekday and weekend paths shared the same first link (alarm, feet on floor, movement clothes) and converged on the same final link (shower, first focused activity). These shared endpoints — the anchors — gave both branches a common foundation and a common destination, making the branch in the middle feel like a variation rather than a disruption.
This is not accidental. The first and last links in any chain carry disproportionate weight. The first link determines whether the chain fires at all. The last link determines whether the chain feels complete — whether the brain registers the reward signal that reinforces the entire sequence. In Chain anchors, you will examine this anchor principle directly: why the first and last behaviors in a chain must be the strongest and most reliable links, how to design anchors that hold even when the middle links vary, and what happens when an anchor weakens.
Sources:
- Gollwitzer, P. M. (1999). "Implementation intentions: Strong effects of simple plans." American Psychologist, 54(7), 493-503.
- Gollwitzer, P. M., & Brandstatter, V. (1997). "Implementation intentions and effective goal pursuit." Journal of Personality and Social Psychology, 73(1), 186-199.
- Schank, R. C., & Abelson, R. P. (1977). Scripts, Plans, Goals, and Understanding: An Inquiry into Human Knowledge Structures. Lawrence Erlbaum Associates.
- Wood, W., & Neal, D. T. (2007). "A new look at habits and the habit-goal interface." Psychological Review, 114(4), 843-863.
- Wood, W. (2019). Good Habits, Bad Habits: The Science of Making Positive Changes That Stick. Farrar, Straus and Giroux.
- Adriaanse, M. A., Vinkers, C. D. W., De Ridder, D. T. D., Hox, J. J., & De Wit, J. B. F. (2011). "Do implementation intentions help to eat a healthy diet? A systematic review and meta-analysis of the empirical evidence." Appetite, 56(1), 183-193.
- Webb, T. L., & Sheeran, P. (2006). "Does changing behavioral intentions engender behavior change? A meta-analysis of the experimental evidence." Psychological Bulletin, 132(2), 249-268.
Frequently Asked Questions