George Ainslie (2005), Behavioral and Brain Sciences 28(5):635–650 (target article; pp. 650–673 are the commentaries and the author’s response) — DOI-10.1017/S0140525X05000117. A précis of Breakdown of Will (Cambridge University Press, 2001), filed here as LIT-583.
Key takeaways
- The discount curve is hyperbolic, and that alone produces temporary preference. Value falls roughly in proportion to delay (Mazur’s formula, with one free “impatience” parameter), in animals and in people, with real and hypothetical rewards. A smaller, sooner reward is preferred in the period just before it is due and dispreferred from a distance.
- A person is a population of interests, not a unitary maximizer. An interest is the set of processes selected by one kind of reward over one time range: a diet interest and an ice cream interest, not a chocolate and a vanilla one. Each survives by forestalling the others, and “she” is “just the resultant of their activities”. The précis calls this the “bottom up model” and sets it against a governing ego.
- Will is intertemporal bargaining. Treating a current choice as a test case that predicts future choices “bundles” the rewards of the whole series, and bundled hyperbolic curves favour the larger, later reward. A lapse lowers the expectation that the rule will hold, so “personal rules are a recursive mechanism; they continually take their own pulse”. The précis cites bundling experiments in students (Kirby & Guastello 2001) and rats (Ainslie & Monterosso 2003).
- The will has predictable side effects. Rules make choices legalistic. Lapses get fenced into “lapse districts”, which clinicians call symptoms. Both long-range and short-range interests are motivated not to notice a lapse. Rules favour well-marked criteria, which the précis offers as a mechanism for disorders of over-control such as obsessive-compulsive personality disorder.
- Emotions and pain are reward-dependent, not conditioned. A very fast cycle of brief reward followed by longer non-reward would make a process that attracts attention and is avoided at a distance. Positive emotions are limited by “premature satiation”: they stay intense only when their occasions are surprising, which the précis uses to explain belief, the value of other people, and why some goals must be pursued indirectly.
Standing in the record
Filed on 2026-10-03 at the owner’s request, for the record’s coverage of behavioural control and self-control. The owner named the précis as the reading to stand in for the 2001 book. The book is registered beside it as a Deferred seed, LIT-583, because it is the citation others reach for.
What it adds that the record lacked is a mechanism for self-control that is neither a higher faculty nor a single preference. It bears on four things already held:
- Frijda’s account of impulse control (LIT-540) and THEORY-056. Frijda holds that most emotion regulation is one motive state checking another, with no separate regulator. Ainslie reaches a similar architecture from the economics of choice. Successive interests forestall each other, and “stable choice has to be achieved as it is in the kind of markets that don’t have governors”. He adds what Frijda leaves open, a rule for which state wins: the one with the greatest discounted reward at that moment, with bundling added by self-prediction. See NOTE-443.
- The hierarchical accounts of agency in THEORY-029. The précis is not about whose a motive is, so it does not answer Velleman’s dilemma. It does state, against “organ” models of will, that a governing strength “has to be guided by some evaluation process outside of motivation” and asks “What keeps this strength from being co-opted by the bad option?”. That is the same regress, posed about force rather than ownership.
- Self-determination theory (LIT-558, LIT-560). Ainslie’s account of rule-bound, legalistic willpower that loses “the here-and-now” resembles SDT’s controlled and introjected regulation, and his “indirection puzzle” (trying to have fun spoils the fun) parallels the undermining of intrinsic motivation by control. The mechanisms differ. SDT locates the cost in the experience of being controlled. Ainslie locates it in the efficiency that bright-line rules force on appetite.
- Bratman’s planning agency. The précis treats Bratman’s “resolute choice” as a rival model of will and suggests it “may turn out to be another name for intertemporal bargaining”. The Bratman text the précis uses is Faces of Intention (1999, pp. 35–57: the pianist and the toxin). That book is not in the record. Bratman’s planning theory of self-governance is filed as LIT-567, which this entry rivals, and the account built on each is in theory.d: THEORY-067 on this précis, THEORY-076 on Bratman’s essay.
No instruction for machine-learning practice; nothing here belongs in the anthology.