Externalize your commitments

Your brain is lying about what it will remember

You made a commitment this week that you have already forgotten. Not a vague aspiration — a specific promise. You told someone you would do something, or you told yourself you would do something, and the commitment is gone. Not renegotiated. Not deprioritized. Simply lost in the noise of everything else your working memory tried to hold.

This is not a character flaw. It is an engineering failure. You are storing binding agreements in a system — your biological working memory — that was never designed for reliable long-term storage. Nelson Cowan's research established that your central cognitive workspace holds roughly three to five items at a time. Every new input competes for those slots. A commitment made on Tuesday morning is displaced by Tuesday afternoon's meetings, Wednesday's emergencies, and Thursday's context switches. By Friday, the commitment does not feel broken because it does not feel like anything — it has been overwritten.

The previous lesson asked you to externalize your assumptions — to write down what you believe so you can test it. This lesson applies the same principle to what you have promised. An unwritten commitment is an invitation for your future self to renegotiate without even realizing a negotiation is happening.

The cognitive tax of uncommitted commitments

In 1927, psychologist Bluma Zeigarnik observed something her professor Kurt Lewin had noticed in a Berlin restaurant: a waiter could remember complex unpaid orders with remarkable accuracy, but after the bill was settled, the details vanished. Zeigarnik's subsequent experiments confirmed the pattern — interrupted or incomplete tasks occupied mental resources that completed tasks did not (Zeigarnik, 1927).

David Allen, who built the Getting Things Done methodology on this insight, calls these uncommitted commitments "open loops." Every promise you have made — to a colleague, a friend, yourself — that lacks a concrete next action and a trusted external home becomes an open loop consuming cognitive bandwidth. Allen's central claim is that your mind is for having ideas, not holding them. When you try to use working memory as a commitment tracker, you pay a continuous tax: background anxiety, reduced focus, the vague sense that you are forgetting something important.

The empirical confirmation came from Masicampo and Baumeister (2011), who demonstrated that unfulfilled goals produce intrusive thoughts that interfere with unrelated cognitive tasks. Participants with unfinished goals performed worse on reading comprehension and anagram tasks — not because of the goal's difficulty, but because unresolved commitments commandeered attentional resources. The critical finding: when participants wrote specific plans for their unfulfilled goals, the intrusive thoughts stopped. The plan did not need to be executed — it needed to be externalized. The act of committing the commitment to a concrete, written form released the cognitive resources that the open loop had been consuming.

This is the mechanism behind the feeling of relief you get when you write down a task that has been nagging you. Your brain does not distinguish between "this is handled" and "this is captured in a system I trust." Both close the loop.

Commitment devices: binding your future self

Externalizing commitments is not just about memory. It is about the structural relationship between your present self and your future self — two agents with different incentives, different emotional states, and different information environments.

Thomas Schelling formalized this problem in The Strategy of Conflict (1960). Schelling, who won the Nobel Prize in Economics, argued that a rational agent may voluntarily limit their own future options to make a current commitment credible. The paradigmatic example predates behavioral economics by three thousand years: Odysseus, approaching the Sirens, ordered his crew to bind him to the ship's mast and fill their own ears with wax. He knew his future self — hearing the Sirens' song — would be a different agent with different preferences. The ropes were not a weakness. They were an infrastructure decision.

Jon Elster extended this framework in Ulysses and the Sirens (1979), arguing that precommitment is the rational response to imperfect rationality. You are not one consistent agent across time. You are a series of agents who share a body, and each version of you inherits the environment the previous version left behind. The externalized commitment is an environmental modification: it changes what your future self encounters, making the intended behavior easier and the unintended behavior harder.

Modern commitment devices operationalize this principle. StickK, founded by Yale economists Dean Karlan and Ian Ayres, allows users to create commitment contracts with financial stakes — money goes to a designated charity (or "anti-charity") if the commitment is not met. Research analyzing stickK's data found that financial commitment contracts can up to triple completion rates, with anti-charity stakes (money sent to an organization the user opposes) being the most effective configuration (Royer et al., 2015; Schwartz et al., 2014). Beeminder uses a similar model, tracking quantified goals and charging users automatically when they go off track.

The principle beneath these tools is simple: an externalized commitment with visible consequences is structurally different from an internal intention. The internal intention relies on your future self's motivation remaining constant. The external commitment creates an environment where follow-through is the path of least resistance.

Implementation intentions: the format that doubles follow-through

Not all externalized commitments are equally effective. The format matters. Peter Gollwitzer's research on implementation intentions — spanning three decades and hundreds of studies — demonstrates that the specific structure of "if X happens, then I will do Y" produces dramatically higher follow-through than general goal statements.

A meta-analysis of 94 independent tests found that implementation intentions had a medium-to-large effect (d = .65) on goal attainment (Gollwitzer & Sheeran, 2006). A more recent meta-analysis of 642 tests confirmed the effect across cognitive, affective, and behavioral outcomes, with effect sizes between d = .27 and d = .66 depending on the domain (Keller et al., 2024). The mechanism is not motivational — it is architectural. When you write "If it is 7 AM on Monday, then I will review the architecture doc," you create a strong associative link between a situation and an action. The situation becomes a trigger. The action becomes automatic. You offload the decision from effortful deliberation to pattern recognition.

Compare the formats:

• Vague commitment: "I'll review that doc soon."
• Goal intention: "I will review the architecture doc this week."
• Implementation intention: "If it is 7 AM Monday, then I will open the architecture doc and spend 30 minutes reviewing sections 2 and 3."

The first is barely a commitment. The second is a commitment without infrastructure. The third is a commitment with a trigger, a time, a location, a scope, and an exit condition. It is the difference between wanting to do something and having built the cognitive rails that make doing it the default.

This is why David Allen's "next action" principle works: every commitment must be decomposed to the point where the very next physical action is obvious. "Plan the offsite" is an open loop. "Email Sarah to ask about venue availability on March 15" is a closed specification. The externalized commitment must be concrete enough to execute without further deliberation.

Building your commitment externalization practice

The practice has four components, and each one addresses a different failure mode of internal commitment storage.

Capture within 60 seconds. The moment you make a commitment — verbally, mentally, in a meeting, in a text message — record it in your trusted system. Not "later when I have time." Within 60 seconds. The gap between making and capturing is where commitments die. Your capture tool does not matter (notebook, phone, task app, voice memo). What matters is that it is always with you and that you trust it. Allen calls this ubiquitous capture, and it is the foundation on which everything else rests.

Specify the next action. Every captured commitment must be rewritten as a concrete next physical action with a date or trigger. "Follow up with the client" becomes "Email client with revised scope document by end of day Thursday." If you cannot specify the next action, the commitment is not yet real — it is still a wish. Force yourself to answer: what would someone watching me do see me physically doing to advance this?

Review daily. A commitment system you do not review is a graveyard. Every day, spend five minutes scanning your externalized commitments. Which are on track? Which need renegotiation? Which have become irrelevant? The daily review is the heartbeat of the system — it is what transforms a list into a living accountability structure. Without the review, you have traded one forgetting mechanism (working memory) for another (an ignored app).

Renegotiate explicitly. Not every commitment should be kept. Circumstances change. Priorities shift. But the difference between a broken promise and a renegotiated commitment is communication. When a commitment can no longer be met, externalization gives you the infrastructure to see it before the deadline, communicate proactively, and adjust the agreement. Silently dropping a commitment is the behavior of someone whose commitments live in their head. Renegotiating is the behavior of someone whose commitments live in a system.

AI as commitment infrastructure

Large language models introduce a new layer to commitment externalization — one that previous generations of tools could not provide. A task manager stores your commitments. An AI can interrogate them.

The pattern: at the end of each day or week, present your externalized commitment list to an LLM and ask it to audit. Which commitments are overdue? Which conflict with each other? Which lack a specific next action? Which have been on the list for two weeks without progress — suggesting they are aspirational rather than genuine? The AI is not providing motivation. It is providing the analytical function that your working memory cannot perform while also being the system that holds the commitments.

More specifically, you can use an AI to transform vague commitments into implementation intentions. Feed it "I need to finish the Q2 planning doc" and ask it to generate if-then specifications with triggers, durations, and exit conditions. The LLM becomes a commitment compiler — translating the high-level intention into executable instructions that your future self can act on without deliberation.

Andy Clark and David Chalmers' Extended Mind thesis (1998) argued that when an external resource reliably functions as part of a cognitive process, it is part of the mind. A notebook that stores your commitments extends your memory. An AI that reviews, challenges, and restructures your commitments extends your executive function. The commitment is no longer just stored outside your head — it is actively managed outside your head. You have built infrastructure where there was only intention.

Tools like Beeminder and stickK externalize consequences. AI externalizes the management layer itself: the pattern-matching, conflict-detection, and accountability dialogue that previously required either a human coach or your own unreliable self-monitoring. The combination — externalized commitments, externalized consequences, externalized management — creates a commitment infrastructure that is resilient to exactly the drift and forgetting that internal commitments cannot survive.

The protocol

1. Audit your current commitments. Spend 20 minutes writing down every commitment you currently hold — professional, personal, implied, explicit. Include promises to yourself. The total will likely surprise you.

2. Externalize each one. Move every commitment into a single trusted system. For each, write the specific next action and a date or trigger. If a commitment cannot be specified concretely, either clarify it or drop it.

3. Install a daily review. Block five minutes at the same time each day. Scan every active commitment. Mark progress. Identify what needs renegotiation. This is non-negotiable — the review is what makes the system alive.

4. Use implementation intention format. For your three most important commitments this week, rewrite them as if-then statements: "If [specific trigger], then I will [specific action] for [specific duration]."

5. Run an AI commitment audit. Present your full commitment list to an LLM. Ask: "Which of these commitments lack a concrete next action? Which conflict with each other? Which have I been carrying for more than a week without progress?"

6. Renegotiate before breaking. When you identify a commitment you cannot meet, communicate that fact before the deadline. The practice of explicit renegotiation is what separates a reliable person from a well-intentioned one.

What this makes possible

When your commitments live outside your head — visible, specific, reviewed, and managed — several things change at once.

Your cognitive load drops. Masicampo and Baumeister proved that the mere act of making a plan for an unfulfilled goal eliminates the intrusive thoughts it generates. Every commitment you externalize is a loop you close, freeing working memory for the thinking that actually matters.

Your reliability becomes structural rather than heroic. You stop relying on willpower to remember and start relying on systems that cannot forget. The person who reviews their commitment list daily and renegotiates proactively is not more disciplined than you — they have better infrastructure.

Your relationship with yourself changes. When commitments are visible, you cannot lie to yourself about what you have promised versus what you have delivered. The gap between stated and actual priorities — which the next lesson, Externalize your priorities, will address directly — becomes measurable. And what is measurable can be improved.

Odysseus did not trust his future self. He built an environment that made his commitment irrevocable. You do not need ropes and a mast. You need a capture tool, a daily review, and the willingness to treat your promises as infrastructure rather than intentions.

The commitment you write down is the commitment you keep. The one you leave in your head is the one that quietly disappears — and takes a piece of your self-trust with it.