Solving Present Bias: Governance across time.

🧠 Free Recall ❓ Quiz 🔍 Cues 🔮 Simulation 🥷 ShadowBox 📐 Encode Cards

1. The naïve-agent model
The naïve agent assumes:
“My future self will value this goal as much as I do now.”
This is usually false. Tomorrow’s self will also prefer comfort, novelty, optionality, and immediate rewards.
This model fails because it treats intention as though it were a durable command. But an intention is more like a proposal submitted to a future decision-maker.
Resolution: sophistication
The sophisticated agent assumes:
“My future self will predictably attempt to renegotiate this.”
Instead of merely setting a goal, the person designs around their expected future weakness. Behavioral-economics models distinguish between naïve agents, who do not anticipate their future present bias, and sophisticated agents, who expect it and may deliberately restrict their future choices. 
This produces the first principle:
Do not make plans that require your future self to become morally superior. Make plans that remain workable when your future self behaves normally.
2. The commitment-device model
A commitment device changes the future choice architecture so that abandoning the goal becomes harder or more costly.
Examples:
automatic payroll deductions,
prepaid classes,
public deadlines,
money forfeited upon nonperformance,
software blockers,
removing tempting alternatives,
scheduling another person who is depending on you.
The classic metaphor is Odysseus tying himself to the mast: present Odysseus governs future Odysseus because he knows future Odysseus will be temporarily compromised. Commitment devices can range from soft reminders to binding financial or institutional constraints. 
The limitation
Commitment devices are strongest when:
the desired behavior is unambiguous,
the future environment is reasonably predictable,
failure is mostly caused by temptation.
They are weaker when legitimate uncertainty exists. A harsh penalty may punish you for illness, changed circumstances, or discovering that the original goal was misguided. Recent research also suggests people may avoid commitment devices because they reasonably fear failing and incurring the penalty—not simply because they are unaware of their present bias. 
Better version: bounded commitment
Commit to the process, not blindly to the final outcome.
For example:
weak: “I must lose 25 pounds.”
stronger: “For 12 weeks, I will train three times weekly and review the plan afterward.”
weak: “I will write a book.”
stronger: “I will protect 7:00–7:45 a.m. for writing Monday through Friday through September.”
This preserves adaptation while preventing daily renegotiation.

3. The principal–agent model
Treat the goal-setting self as a principal and future selves as agents.
The principal has a long-horizon objective. The agents control day-to-day execution, but they have different incentives and better local information.
This immediately reveals why many goals fail:
The principal issues vague instructions.
Agents are not given resources.
Performance is not measured.
Rewards arrive too late.
There is no enforcement.
The principal disappears for six months and then expresses disappointment.
No competent organization would operate this way.
Resolution: an incentive-compatible contract
A well-designed personal goal should specify:
Mandate: What outcome are we pursuing?
Authority: What may the executing self change?
Budget: What time, money, and energy are reserved?
Leading indicator: What action counts as execution?
Review date: When may the commitment be reconsidered?
Exit clause: What evidence justifies stopping?
The crucial distinction is between revision and defection.
Revision happens during a scheduled review, using evidence.
Defection happens in the moment because execution is uncomfortable.
Future selves should have the authority to revise the plan, but not at arbitrary moments.

4. The constitutional model
A constitution does not try to decide every future case. It establishes rules governing how future decisions are made.
A personal constitution might say:
Goals are adopted quarterly, not impulsively.
Every adopted goal must receive calendar time.
Goals cannot be abandoned on a bad day.
They may be changed only during a weekly or monthly review.
Missing once triggers recovery, not cancellation.
Repeated failure triggers redesign before increased effort.
No new major goal may be added without removing or reducing another.
This is stronger than motivation because it governs the procedure of renegotiation.
Central-bank credibility has a similar structure. A declaration is weak when future policymakers retain incentives to reverse it. Credibility comes from rules, institutional constraints, transparency, independent authority, or reputational costs—not the intensity with which the original promise was stated.
The personal analogue is:
Your goals become credible when future deviations have a defined procedure and cost.
5. The implementation-intention model
Many goals fail not because the person changes their mind, but because the intention never becomes attached to a specific trigger.
“I will exercise more” leaves every operational question unresolved.
An implementation intention takes the form:
When situation X occurs, I will perform behavior Y.
For example:
“After I pour my morning coffee on weekdays, I will write for 25 minutes at the kitchen table.”
Implementation intentions are specifically intended to close the intention–behavior gap by linking a recognizable cue to a predetermined action. They work best when the underlying goal is genuinely valued rather than externally imposed. 
The broader principle:
A goal should compile into executable instructions.
A goal that has not been converted into a time, place, trigger, and next physical action is not yet an operational plan.

6. The temporal-motivation model
Temporal Motivation Theory roughly models motivation as increasing with:
the value of the outcome,
the perceived probability of success,
and decreasing with:
delay,
impulsiveness or sensitivity to delay.
A distant, uncertain reward is motivationally weak. As a deadline approaches, the reward or penalty becomes psychologically immediate, which explains last-minute surges of effort. Research using this model describes procrastination as changing dynamically over time rather than as a fixed lack of character. 
Resolution: shorten the motivational distance
Do not rely only on the distant reward. Create proximal rewards and consequences.
Instead of:
“Build a respected body of work over five years.”
Use:
publish one artifact Friday,
receive peer feedback,
update a visible count,
earn a leisure reward afterward,
make progress observable this week.
This is not merely breaking the goal into smaller tasks. It is giving each future self a reason to participate.
The distant goal supplies direction. The near-term cycle supplies energy.

7. The future-self-continuity model
People differ in how strongly they experience their future self as “me” rather than as a distant stranger.
Higher future-self continuity has been associated with less discounting of future rewards, greater saving, and better long-term health behavior. Experiments that make the future self more vivid or connected have sometimes improved subsequent behavior, although the intervention literature is promising rather than conclusive. 
Resolution: preserve continuity
Methods include:
writing letters between present and future selves,
keeping a decision journal explaining why the goal matters,
recording a short video at the moment of commitment,
reviewing cumulative evidence of prior work,
using a consistent narrative: “This is the project I am carrying forward.”
However, future-self visualization alone is insufficient. Emotional continuity should be joined to material continuity:
the same workspace,
a persistent scoreboard,
the next task already chosen,
notes left for the next session,
recurring calendar blocks.
The objective is to reduce the psychological “handoff cost” between selves.

8. The identity-and-reputation model
A declaration can create the illusion of identity without requiring evidence:
“I am going to become a writer.”
The person consumes some of the psychological reward of being a writer before writing anything.
A better model reverses the direction:
Actions are votes that gradually establish identity.
But identity should not be defined so rigidly that one lapse becomes an existential contradiction. A useful identity is procedural:
“I am someone who returns.”
“I maintain promises or formally renegotiate them.”
“I leave evidence for the next version of myself.”
“I do not confuse enthusiasm with commitment.”
This introduces intrapersonal reputation.
Every time present self keeps a promise, future selves gain evidence that personal commitments are meaningful. Every abandoned declaration lowers the internal credibility of the next declaration.
This explains why repeatedly announcing ambitious goals can be actively harmful. You are not merely failing to reach the goal; you are depreciating your own internal currency.
Resolution: protect the credibility of declarations
Make fewer promises.
Classify intentions:
Interest: “This seems valuable.”
Experiment: “I will test this for two weeks.”
Project: “Resources have been committed.”
Rule: “This remains in force until formally changed.”
Most New Year’s resolutions are interests spoken in the language of rules.

9. The control-systems model
A goal is a desired state. A control system continually compares the current state with the desired state and applies correction.
A functioning control loop requires:
a target,
measurement,
frequent feedback,
corrective action,
a tolerance range,
periodic recalibration.
New Year’s resolutions usually contain only a target. They have no sensor, feedback cadence, or correction mechanism.
Resolution: make the goal observable
For a writing goal:
target: publish 24 essays,
leading measure: three writing sessions per week,
state: 2/3 sessions complete,
error: one session behind,
correction: Saturday recovery block,
recalibration: monthly review.
The important insight is:
Deviation is normal. Uncorrected deviation is failure.
This replaces the brittle “perfect streak” model with a resilient return-to-path model.

10. The real-options model
Sometimes abandoning a goal is rational. The present self does not possess perfect information about future circumstances, preferences, or opportunities.
Therefore, permanent commitment can be as irrational as no commitment.
The real-options model says:
make a limited investment,
obtain information,
preserve the option to expand, modify, or terminate.
Examples:
twelve-week fitness block,
six-essay writing season,
one-month product prototype,
three customer interviews before building,
one semester rather than a permanent identity decision.
Resolution: stage commitments
Use a sequence:
Explore
Commit
Execute without daily reconsideration
Review
Renew, revise, or terminate
This avoids both errors:
abandoning worthwhile goals whenever discomfort appears,
continuing obsolete goals merely because a past self declared them.
A unified model: intertemporal alignment
The strongest solution is not one technique. It is an architecture with five layers.
1. Legitimacy
The goal must survive scrutiny before adoption:
Is it genuinely valuable?
What am I willing to displace for it?
Would I still want it without the social reward of announcing it?
2. Translation
Convert aspiration into operations:
recurring behavior,
first action,
trigger,
time and place,
minimum acceptable version.
3. Incentive alignment
Give the executing self something immediate:
visible progress,
satisfaction,
social accountability,
reward,
avoidance of a credible cost.
4. Constraint
Prevent impulsive renegotiation:
automation,
calendar reservations,
environmental design,
prepayment,
external dependency,
restricted alternatives.
5. Governance
Specify how the plan changes:
weekly correction,
monthly review,
explicit renewal date,
evidence required for termination,
recovery protocol after failure.
A compact personal contract could therefore read:
Until September 30, I will write from 7:00–7:30 a.m. on Monday, Wednesday, and Friday. The minimum session is ten minutes. I may alter the method during weekly review, but I may not cancel a session at the moment of execution except for a predefined emergency. Missed sessions move to Saturday morning. On September 30, I will renew, revise, or terminate the project based on completed sessions and produced work.
That contract aligns the selves because:
the choosing self commits resources,
the executing self receives clear instructions,
the tired self has a minimum version,
the disrupted self has a recovery route,
the informed future self retains a legitimate exit,
no self has unilateral power to quietly forget the goal.
The central shift is this:
Goals should not be treated as commands issued by one sovereign self. They should be treated as institutions designed to coordinate a changing coalition of selves over time.

Unstarred

Dunk (exclude from study)

Solving Present Bias: Governance across time.

Review Questions

Todos

Sensemaker