Operant hoarding: a new paradigm for the study of self-control.
Reinforcement can create 'saving' behavior, but its strength flips with the learner's state of deprivation.
01Research in Context
What this study did
Dunn (1990) built a new rat model of saving. The animals could press a lever to 'deposit' food pellets into a bank. Three setups were tested: plain saving, a locked savings account, and an account that paid 'interest' after a delay.
The team varied hunger level. Sometimes the rats were full, sometimes food-restricted. The goal was to see how reinforcement rules and body state interact to create hoarding or self-control.
What they found
When each deposit earned extra pellets, rats saved more. That part was simple. The twist came with interest. If the rat had free access to food before the interest arrived, it stopped saving. If it stayed hungry, it kept saving to earn the bonus.
The same contingency had opposite effects depending on the animal's level of satiation. Self-control, in this model, was not a fixed trait; it flipped with internal motivation.
How this fits with other research
Dunkel-Jackson et al. (2016) and Cullinan et al. (2001) later showed that people with autism or ADHD can learn to wait days for a bigger reinforcer. They used gradual delay increases, not hoarding, but the shared theme is that careful contingencies build self-control.
Zhou et al. (2023) turned the idea toward food stealing. Children with autism were taught to say 'I will wait' and then wait. The verbal rule plus tiny delay steps eliminated stealing, echoing the rat finding that environmental cues and body state together guide choice.
Logue et al. (1986) looked like a contradiction at first. Adult women in a lab almost always picked the larger later reward, seeming ultra-self-controlled. The difference is procedure: humans in a quiet room with money face no immediate biological need, while hungry rats work for food. The hoarding paradigm reveals contingency effects that standard delay tasks can miss.
Why it matters
You can use the hoarding insight today. First, check the client's 'satiation level.' A child who just ate may skip a token board; the same board works wonders before lunch. Second, link savings to visible interest—bonus tokens, extra screen time—to keep the delay worthwhile. Third, write a simple rule the client states aloud; Zhou et al. (2023) showed this bridges the wait. Self-control procedures need to fit moment-to-moment motivation, not just the clock.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Before a token session, note when the client last ate; run the program pre-meal and add a visible 'interest' bonus for every five tokens saved.
02At a glance
03Original abstract
In the first of four experiments, rats were exposed to a modified multiple continuous reinforcement-extinction schedule during 15-min daily sessions. In one condition (saves condition) with the cuelight on, a single lever press produced a food pellet, briefly extinguished the cuelight, and started a clock. Saves (additional lever presses with interresponse times less than 1 s) produced an additional food pellet, briefly extinguished the cuelight, and restarted the interresponse time clock. The cuelight was extinguished 1 s after the last lever press and remained off during a 10-s period of extinction, during which no food pellets were delivered. In the other condition (savings account condition), the contingencies were the same except that the cuelight was extinguished and was not reilluminated after the initial lever press, and the delivery of all food pellets in the reinforcement component was delayed until the onset of extinction. In both conditions, rats made saves, but mean saves (total saves divided by the number of reinforcement components) were slightly reduced in the savings account condition. In Experiment 2, using six equally spaced 15-min sessions per day on alternate days, saves were either followed immediately with food and brief cuelight offset (saves condition) or were not reinforced at all. Mean saves were much greater when saves were reinforced. In Experiment 3, during 5-min daily sessions, saves earned a single pellet (savings account condition) or a number of pellets equal to the ordinal number of the lever press (interest condition). Rats made fewer mean saves, with little change in the food rate, when saves earned interest. In Experiment 4, the rats earned all their food in the operant situation during 24 daily 5-min sessions, these separated by 55-min intersession intervals during which no food was available; otherwise, the conditions were the same as in Experiment 3. In Experiment 4, the shift to interest for saves led to an increase in mean daily mean saves (total daily mean saves divided by the number of daily sessions) as well as to an increase in the number of food pellets delivered in each session. The results are discussed in terms of self-control and behavioral economics.
Journal of the experimental analysis of behavior, 1990 · doi:10.1901/jeab.1990.53-247