Melioration is a theory of decision-making that explains why people and animals repeatedly choose what feels best right now, even when doing so leads to worse outcomes overall. Developed primarily by psychologist Richard Herrnstein and economist Drazen Prelec, the theory proposes that when we face the same kind of choice again and again over time, we tend to gravitate toward whichever option currently delivers the higher local payoff, rather than stepping back to calculate which pattern of choices would leave us best off in total. The result is a systematic drift toward suboptimal behavior that looks irrational from the outside but follows a clear internal logic.
What Meliorating Actually Looks Like
The word “meliorate” comes from the Latin for “to make better,” which is exactly what the decision-maker thinks they are doing at every step. Imagine you have two activities available to you repeatedly. One gives you a quick, reliable payoff each time you choose it. The other gives a smaller immediate payoff but would generate much larger cumulative returns if you chose it more often. A meliorating decision-maker looks at the local rate of return from each option and shifts toward whichever one is currently paying off better, choice by choice. They are not ignoring the alternatives; they are comparing them and picking the winner. The problem is that the comparison happens at the wrong scale.
This matters because in many real situations, the payoff you get from an activity depends on how much time you have been spending on it relative to the alternatives. Herrnstein and Prelec’s foundational paper laid out the issue clearly: choices spread out over time, like how much you exercise, how hard you work, how much you save, or how often you gamble, can be “reliably and predictably suboptimal, in terms of the person’s own preferences.”1Journal of Economic Perspectives. Melioration: A Theory of Distributed Choice You end up in a pattern you yourself would not have chosen if you had seen the whole picture upfront.
The trap is subtle. Nobody sits down and decides to underexercise or overspend. Instead, each individual choice feels perfectly reasonable. Skipping the gym today to relax feels better than going. Buying the small treat feels better than saving the money. The meliorating mind does not plan the sequence of choices; it just picks the locally superior option each time. Over weeks and months, those locally superior choices accumulate into a pattern the person dislikes.
Where the Evidence Comes From
Melioration was first studied rigorously in animal behavior labs, where researchers could precisely control what each choice delivered. In classic experiments, pigeons choosing between two reward schedules consistently matched their time allocation to the relative rate of reinforcement they were receiving, rather than maximizing their total food intake. One study using concurrent variable-interval schedules found that pigeons approximately matched their time proportions to reinforcement proportions, and their overall reinforcement rates actually decreased with continued training when conditions allowed for a divergence between meliorating and maximizing strategies. The authors concluded the results provided evidence against reinforcement maximization and supported the melioration hypothesis.2PubMed Central. Melloration and maximization of reinforcement minus costs of behavior
Research with rats told a similar story. In an adjusting-amount procedure designed to study impulsive choice, rats did not pick the option that would have maximized their total reinforcement. Instead, they tracked the value they received from each alternative on a trial-by-trial basis and chose whichever one was currently delivering more. The dynamic settled into an equilibrium where the perceived value of both options matched, exactly what melioration predicts.3Behavioural Processes. Effects of reinforcer magnitude on an animal model of impulsive behavior The animals were not being random or confused. They were following a consistent rule: go where the current return is higher. The rule just happened to lead them astray.
Humans Fall Into the Same Pattern
The natural question is whether people do the same thing when real money is on the line and they can, in principle, think ahead. The answer is yes, often strikingly so. In a series of experiments where volunteers worked for money across six different procedures, the payoff from each option depended on how the participant had recently been dividing their choices. Because of this hidden interaction, always picking the currently more profitable option did not actually maximize total earnings. Yet subjects frequently ignored these dynamics and were guided by the current yields of the two alternatives, meliorating rather than optimizing.4Journal of Behavioral Decision Making. Utility maximization and melioration: Internalities in individual choice
The researchers called these hidden payoff interactions “internalities,” a useful concept. An internality is when your own past behavior changes what your current choice is worth, but in a way you do not naturally track. If eating one cookie makes the next cookie slightly less enjoyable while simultaneously making the broccoli on the counter slightly more beneficial, those are internalities. A perfectly rational optimizer would factor them in. A meliorating decision-maker does not, because melioration only looks at what each option is paying right now.
What stands out about these human experiments is that the subjects were not unintelligent. They were working for real money in a controlled setting. They simply did not detect or respond to the structural relationship between their own allocation history and future payoffs. Melioration appears to be something like a default decision-making mode, one that works well enough in simple environments but creates systematic errors when the consequences of choices feed back on each other over time.
Why This Explains So Much Everyday Frustration
Herrnstein and Prelec pointed out that when people complain about their own behavior, the complaints cluster around exactly the kinds of distributed choices where melioration would predict trouble: working too hard or not hard enough, exercising too little, wasting time, overeating, overspending, and so on.5Journal of Economic Perspectives. Melioration: A Theory of Distributed Choice These are all domains where the returns from an activity change depending on how much of it you do, and where the locally appealing choice diverges from the globally optimal pattern.
Take exercise. The immediate experience of going for a run when you are out of shape is unpleasant, while the immediate experience of sitting on the couch is comfortable. A meliorating chooser compares those two local payoffs and picks the couch. But the more you run, the easier running gets and the more health benefits you accumulate. The returns to exercise increase with regular participation. A person who could somehow commit to the long-run pattern would prefer the one that includes regular running, but the choice-by-choice meliorating process never gets them there.
Savings works the same way. Spending money now feels better than saving it, because the reward from spending is immediate and tangible. But the cumulative benefit of saving grows over time through compound interest and financial security. The meliorating chooser sees the current payoff of spending as higher than the current payoff of saving and drifts toward overspending. The theory predicts a general underinvestment in activities whose average returns increase with the rate at which you do them, and overinvestment in activities that have an addiction-like dynamic where initial pleasure is high but long-run costs mount.
The Connection to Addiction
Melioration’s most provocative application is to addictive behavior. If the theory is correct, addiction does not require a special brain malfunction or a unique psychological disorder. It is what happens when ordinary meliorating choice encounters a substance or activity with a particular payoff structure: the local return from using is high, while the costs accumulate slowly and are distributed across the future.
A commentary in a major behavioral science journal found the case convincing, noting that a melioration choice strategy is sufficient to produce addictive behavior.6Behavioral and Brain Sciences. Is melioration the addiction theory of choice? The implications are significant. Under this view, the difference between someone who develops an addiction and someone who does not may have less to do with willpower or moral character and more to do with the specific payoff structure of the choices available to them. An environment that makes the local returns of substance use very high relative to alternatives, or that makes the delayed costs harder to perceive, would produce more addictive behavior even among people with perfectly normal decision-making processes.
This does not mean melioration is the whole story of addiction. Biology, genetics, social context, and the pharmacology of specific substances all play roles. But the melioration framework offers an explanation for why addictive patterns can emerge from ordinary choice processes without requiring any special pathology. It also explains why addiction often co-occurs with other forms of distributed-choice failure: a person who struggles with substance use might simultaneously underexercise, undersave, and overwork, because all of these domains share the same structural vulnerability to meliorating choice.
How Melioration Differs From Other Accounts of Bad Decisions
You might wonder how melioration relates to other ideas about irrational behavior, like impulsivity, present bias, or hyperbolic discounting. The difference is subtle but real. Impulsivity and present bias describe the tendency to overweight immediate rewards relative to delayed ones. Melioration is a specific mechanism that can produce that pattern, but it is not quite the same thing. A meliorating chooser is not simply impatient; they are comparing the current local rates of return across alternatives and shifting toward the higher one. In simple situations, this looks like impulsivity. In more complex ones, it can produce behavior that impulsivity alone does not predict, such as getting stuck at a particular ratio of time allocation between two activities even when both are available simultaneously.
Melioration also differs from straightforward maximization failure. Classic economic models assume people maximize their utility, and when they do not, economists often invoke lack of information or cognitive limitations. Melioration says the problem is more structural: even with full information, the choice-by-choice process tends to settle at an equilibrium where local rates of return are equalized across options, not where total return is maximized. You can know you should exercise more and still not do it, because every individual instance of the choice tips toward the couch.
Competing models from the behavioral science literature have pushed back on melioration’s generality. One alternative, called ratio invariance, has been shown to be more adaptive and less susceptible to the kinds of “traps” melioration falls into, though researchers have noted that melioration-like processes may still dominate in more complex situations where the decision-maker can distinguish between time periods.7PubMed Central. Quasi-dynamic choice models: Melioration and ratio invariance The debate is ongoing, and the truth likely involves different strategies being active in different environments. But melioration remains the best-supported account of why distributed choices drift toward suboptimal patterns so predictably.
Fighting the Trap With Choice Bundling
If melioration is a built-in tendency, can anything be done about it? One of the most promising strategies researchers have studied is called choice bundling, and it works by changing the scale at which decisions are made. Instead of evaluating each choice in isolation (go to the gym today or not?), you mentally or structurally group choices together (commit to a gym routine for the month). Bundling makes the long-run consequences more visible at the moment of choice, which helps short-circuit the local-comparison process that drives melioration.
A review of research on choice bundling found that it consistently reduced impulsive decision-making. Participants in bundled conditions more often chose the larger, delayed reward over the smaller, immediate one. The effect held in studies with smokers as well, where both suggested and forced bundling led to less impulsive choices compared to sequential, one-at-a-time decision-making. Across human and animal studies combined, reward bundling produced a large effect, suggesting it is an especially promising avenue for helping people overcome distributed-choice traps.8PubMed Central. A brief review of choice bundling: A strategy to reduce delay discounting and bolster self-control
Practical bundling strategies are everywhere once you start looking. Automatic payroll deductions into a savings account bundle your saving decisions into a single commitment. A gym membership with a fixed monthly fee bundles your exercise decisions: you have already paid, so the local calculation shifts. Meal prepping for the week bundles your dietary choices so you do not face a fresh melioration opportunity at every mealtime. Subscription services for healthy food or vitamins do the same thing. The underlying logic is always to replace a sequence of individual choices, each vulnerable to melioration, with a single commitment that locks in the globally preferred pattern.
When Melioration Is Actually Fine
It would be a mistake to conclude that melioration is always a problem. In many environments, the option with the better local return also happens to be the better option overall. If you are choosing between two restaurants and one consistently serves better food, meliorating toward that restaurant leads you exactly where a perfect optimizer would go. Melioration becomes a trap only when the payoff structure has a specific feature: when the act of choosing one option more frequently changes the returns in a way that eventually makes that option less valuable, or makes the neglected option more valuable, than the chooser realizes.
This structural requirement is common but not universal. Most stable, routine decisions in daily life do not involve strong feedback loops between your allocation pattern and the payoffs. Choosing which route to drive to work, which brand of coffee to buy, or which friend to call first does not typically create the kind of internality that makes melioration costly. The theory’s warnings apply specifically to domains where doing more of something changes what it is worth, like exercise, saving, substance use, or effort allocation across tasks with different learning curves.
Recognizing this boundary matters for practical purposes. You do not need to restructure every decision in your life to avoid meliorating. You need to identify the specific domains where your choices interact with their own consequences over time, and apply bundling or commitment strategies there. For everything else, following your local sense of which option is paying off better is a perfectly reasonable way to navigate the world.
Melioration in Group Settings and Social Dilemmas
Melioration’s logic extends beyond individual choices. In social dilemmas, like the classic public goods problem, each person faces a choice between contributing to a shared resource (which benefits everyone a little) and free-riding (which benefits the individual right now). A meliorating player compares their personal local return from contributing versus keeping their resources. Since free-riding usually pays better locally, the group drifts toward undercontribution, even though everyone would be better off if they all contributed more. The structure is identical to the individual case: each locally rational comparison moves the system toward a globally suboptimal outcome.
Workplace dynamics often mirror this pattern. If your performance review rewards visible short-term output, and the long-term health of the team depends on invisible investments like mentoring or documentation, a meliorating employee naturally drifts toward the flashy deliverables. Each individual decision to prioritize the visible task over the invisible one looks sensible, but the team gradually loses the infrastructure that made productivity possible in the first place. Organizations that recognize this dynamic can design incentive structures to bundle long-term contributions into the evaluation process, the same fix that works at the individual level.
Climate action carries the same flavor. The immediate economic return from burning cheap fossil fuel is higher than the immediate return from investing in renewable infrastructure. Each nation, each company, and each household faces a local comparison that favors the status quo. The costs of carbon emissions are delayed and distributed across the future and across the globe, making them exactly the kind of internality that meliorating choosers fail to track. Understanding this does not solve the problem, but it does shift the framing away from blaming individuals for shortsightedness and toward designing structures, like carbon pricing or binding international agreements, that change the local payoff comparison at the point of each decision.