Feeding Behavior

Abstract

Feeding behavior is a form of animal behavior—the set of actions by which an animal locates, selects, and consumes food. It spans two phases: an appetitive phase of searching and choosing, and a consummatory phase of ingestion. Behind both lies a control problem: intake must track the body's energy and nutrient needs, yet animals also eat for reward, learn which foods to seek or avoid, and forage economically in a patchy world. This article organizes feeding behavior around four questions—where to forage, how much to eat, why palatable food is overeaten, and how diet is learned—developed through optimal foraging theory, homeostatic regulation, the hedonic control of appetite, and food-related learning. A worked example on optimal patch use and three interactive demonstrations accompany the text.

Keywords: feeding behavior, optimal foraging theory, satiety, hedonic hunger, food reward

An animal that must eat to live faces a sequence of decisions long before food reaches its mouth. Where should it search, and when should it give up on a thinning patch and move on? Once food is found, how much should it eat, and when should it stop? Why does it keep eating a tempting food past the point of need, yet refuse a food that once made it ill? Feeding behavior is the name for this whole chain of search, choice, ingestion, and regulation—and for the biological and psychological machinery that solves it.

This article treats feeding behavior as a control problem with four faces. The first is economic: where and how long to forage, the domain of optimal foraging theory. The second is homeostatic: matching intake to the body's energy and nutrient needs through hunger and satiety. The third is hedonic: the reward value of food, which can drive eating beyond need. The fourth is learning: how an animal acquires, from the consequences of eating, a policy about what is safe and worth seeking. Each face has its own research tradition, and feeding behavior is their intersection.

What Feeding Behavior Is

Feeding behavior is the set of actions by which an animal acquires and consumes food. Ethologists divide it into two phases: an appetitive phase—the flexible, goal-directed search for and selection of food—and a consummatory phase—the relatively stereotyped act of ingestion once food is at hand. The appetitive phase is where most of the interesting decisions live: where to look, which items to take, when to move on. The consummatory phase raises its own question of regulation: how much to eat and when to stop.

Two control systems govern how much is eaten. A homeostatic system matches intake to the body's energy and nutrient needs, generating hunger when reserves fall and satiety when they are replenished (Morton et al., 2006). A hedonic system tracks the reward value of food and can drive eating independent of—or against—homeostatic need (Berridge, 1996). Feeding behavior is the output of these systems acting across both phases, and much of its study is the attempt to separate their contributions.

Types of Feeding Behavior

In the Medical Subject Headings (MeSH) vocabulary, feeding behavior sits under animal behavior and subsumes a set of narrower feeding behaviors, listed in Table 1. The grouping is an indexing classification, not a biological taxonomy: the categories are neither exhaustive nor mutually exclusive, and they cut feeding along several different axes at once. Some describe dietary content (herbivory, carnivory), others a developmental mode of intake (breast feeding, bottle feeding), others a regulatory state or motive (fasting, emotional eating, food preferences). A single bout of eating can fall under more than one—emotional eating of meat is at once carnivory and emotionally driven—so the list indexes the literature rather than partitioning behavior.

The remainder of this article treats the principles that cut across these categories—how food is found, how intake is regulated, why reward can override need, and how diet is learned—rather than each behavior in turn.

Finding Food: Optimal Foraging

The appetitive phase poses an economic question: given that food is distributed unevenly in space and time, how should an animal allocate its effort? Optimal foraging theory answers by treating the forager as maximizing net energy gain per unit time, subject to constraints. Robert MacArthur and Eric Pianka opened this line by asking how an animal should use a patchy environment—which prey types to include in its diet, and which patches to visit—if it is to feed efficiently (MacArthur & Pianka, 1966).

The sharpest result is Eric Charnov's marginal value theorem, which solves the patch-leaving problem. As an animal feeds in a patch, the patch depletes and its intake rate falls. The theorem states that the animal should leave when its current intake rate in the patch drops to the average rate it could achieve over the whole environment, counting the travel time to the next patch. When patches are far apart, the average rate is low and the animal should stay longer in each; when travel is cheap, it should leave sooner (Charnov, 1976). Gordon Pyke, Ronald Pulliam, and Charnov consolidated the theory and its early tests, establishing optimal foraging as a quantitative, falsifiable account of where and how long animals feed (Pyke et al., 1977). The worked example below derives the optimal patch residence time explicitly.

Regulating Intake: Homeostatic Control

Once food is available, how much should the animal eat? The classical answer is homeostatic regulation: intake is matched to energy expenditure so that body energy stores are defended around a set point. Signals of the body's energy state—hormones such as leptin and insulin reporting long-term reserves, gut peptides and stomach distension reporting the current meal—converge on the hypothalamus and brainstem, where they are integrated into the drive to start and stop eating. Gregory Morton and colleagues described this architecture: adiposity signals set the background drive, while meal-related satiety signals terminate individual meals, together holding intake to need over time (Morton et al., 2006).

A comprehensive modern synthesis by Alan Watts and colleagues maps the full set of signals, neurons, and networks that control eating, from the gut and bloodstream to distributed forebrain circuits—showing that the control of intake is not a single hunger center but a network that integrates metabolic, sensory, and learned information (Watts et al., 2022). The negative-feedback logic is the core: a deficit raises hunger, intake repays the deficit, and satiety signals shut the meal down, returning the system toward its set point.

Beyond Need: Hedonic Control

Animals do not eat only to meet energy needs; they eat because food is rewarding, and a palatable food is eaten well past the point of homeostatic sufficiency. Kent Berridge drew the crucial distinction within food reward itself: “liking”, the hedonic impact of a food when it is tasted, and “wanting”, the incentive salience that makes a food's cues attract and motivate—two processes with separable brain substrates, so that an animal can want a food it no longer likes, or like one it is not driven to pursue (Berridge, 1996).

This hedonic system interacts with the homeostatic one rather than running parallel to it. Hans-Rudolf Berthoud framed the relationship as a contest for control of appetite between metabolic need and reward, in which, in an environment of cheap palatable food, the hedonic drive can win (Berthoud, 2011). Berthoud, Heike Munzberg, and Christopher Morrison argued that this is central to obesity: hedonic and homeostatic mechanisms are integrated, and modern food environments exploit the integration to push intake above need (Berthoud et al., 2017). Marcus Rossi and Garret Stuber made the anatomy concrete, showing that homeostatic and hedonic feeding are not served by separate circuits but by overlapping ones in which hypothalamic hunger signals and mesolimbic reward signals converge—so the two “systems” are better seen as one integrated controller with two kinds of input (Rossi & Stuber, 2018).

Learning What to Eat

No animal is born knowing which foods are safe and nourishing in its particular environment; it must learn. The most striking demonstration is conditioned taste aversion. John Garcia and Robert Koelling showed that a rat made ill after tasting a novel flavor will avoid that flavor afterward—but illness will not condition avoidance of a light or a sound, and a flavor will not be avoided when paired with a shock instead of illness. Taste is selectively associated with internal consequences, overturning the behaviorist assumption that any stimulus could be linked to any outcome equally (Garcia & Koelling, 1966). Garcia, Walter Hankins, and Kenneth Rusiniak generalized the point: feeding-related learning is biologically prepared to regulate the internal milieu, tuning intake to physiological consequence (Garcia et al., 1974).

Paul Rozin and James Kalat set this learning in an adaptive frame, arguing that specific hungers—an animal's acquired preference for a food that corrects a nutritional deficiency—and poison avoidance are two sides of one domain-specific learning system, specialized for the problem of selecting a diet (Rozin & Kalat, 1971). A different kind of learning operates within a single meal. Barbara Rolls and colleagues described sensory-specific satiety: the pleasantness of a food that has just been eaten declines more than that of foods not eaten, so appetite is satisfied food-by-food rather than in general. The consequence is that a varied meal is a larger meal—offering several distinct foods raises total intake, because each retains its appeal as the others wane (Rolls et al., 1981). The demonstration above shows this variety effect directly.

Worked Example

The marginal value theorem can be made concrete. Suppose a forager enters a patch and accumulates food according to a diminishing-returns gain function, g(t) = G · t / (t + h), where t is the time spent in the patch, G = 40 is the maximum the patch can yield, and h = 4 min sets how quickly returns diminish. Travelling to the next patch takes τ = 9 min. The animal should choose the residence time t that maximizes its long-run intake rate,

R(t) = g(t) / (τ + t),

because τ + t is the total time—travel plus feeding—that one patch visit costs. The marginal value theorem says the optimum is where the patch's current intake rate g′(t) equals the long-run average rate R(t). For this gain function that condition, h / (t + h) = t / (τ + t), reduces to t\ = √(hτ) = √(4 · 9) = 6 min. The table evaluates R(t*) around that optimum:

The intake rate peaks at t\ = 6 min and 1.600 units per minute; leaving earlier or later both lower the long-run rate. The theorem's signature prediction follows from the formula: because t\ = √(hτ), a longer travel time raises the optimal stay—at τ = 16 min the animal should feed for √64 = 8 min, not 6. When patches are costly to reach, it pays to deplete each one further before moving on. The MarginalValueDemo lets the reader vary patch richness, the diminishing-returns rate, and the travel time, and watch the tangent construction locate the optimum. The numbers are illustrative; the structure—leave when the marginal rate falls to the environmental average—is the content.

Discussion

The four faces of feeding behavior share one logic: eating is a decision made under constraint and uncertainty. The forager cannot see the whole environment, so it uses a local rule—leave when the current rate falls to the average—that approximates the global optimum (Charnov, 1976). The body cannot measure its energy balance directly, so it infers it from hormonal and gut signals and defends a set point by negative feedback (Morton et al., 2006). The animal cannot know a food's value in advance, so it learns it from the consequences of eating (Garcia & Koelling, 1966).

What makes feeding behavior hard to summarize is that these systems do not simply add. The hedonic and homeostatic controllers overlap in the brain, so reward and need are weighed in the same circuits rather than settled separately (Rossi & Stuber, 2018). Sensory-specific satiety means the stopping rule is not a single satiety threshold but a food-by-food accounting, so that variety defeats regulation (Rolls et al., 1981). The behavior that results is neither pure optimization nor pure reflex but the output of several interacting controllers, each a good solution to one face of the problem.

Cognitive Implications

For cognitive science, feeding behavior is a model system for value-based decision-making. The forager choosing when to leave a patch is solving a sequential decision problem under diminishing returns—the same structure that appears in foraging accounts of human memory search, attention, and information seeking, where the marginal value theorem has been imported as a general rule for when to abandon a depleting source. The patch-leaving animal is, in effect, computing an opportunity cost (Charnov, 1976; Pyke et al., 1977).

Food-related learning is likewise a biological instance of reinforcement learning with a long and noisy delay between action and outcome. Conditioned taste aversion solves a hard credit-assignment problem—illness hours after a meal must be attributed to the right food—and the solution is a built-in bias to blame novel tastes, a prior that constrains learning to the biologically plausible (Garcia et al., 1974; Rozin & Kalat, 1971). The wanting–liking dissociation, meanwhile, maps onto the distinction between the motivational and evaluative components of reward that underlies computational models of incentive and, clinically, of addiction (Berridge, 1996).

Current Directions

The most active current work dissolves the old split between homeostatic and hedonic control. Rather than asking which system wins, researchers now trace how metabolic and reward signals are integrated within shared circuitry, so that a state of energy need changes the reward value of food directly (Rossi & Stuber, 2018; Berthoud et al., 2017). The comprehensive network view of eating control—signals from gut and bloodstream shaping distributed forebrain computation—reframes appetite as an inference problem the whole brain participates in, not the output of a dedicated center (Watts et al., 2022).

A second direction returns feeding behavior to its nutritional roots with new rigor. The geometric framework of nutrition, extended by David Raubenheimer and Stephen Simpson from animal foraging to human diet, treats intake as the regulation of multiple nutrients toward a target rather than the pursuit of calories alone—and argues that the mismatch between that regulation and modern processed-food environments helps explain overconsumption (Raubenheimer & Simpson, 2016). The open questions concern how the economic, homeostatic, hedonic, and nutritional accounts fit into one model of why animals, including humans, eat as they do.

Common Misconceptions

Animals eat only when their energy stores run low.

Homeostatic need is one driver, but food reward can initiate and sustain eating independently of energy state; a palatable food is eaten well past sufficiency, and the hedonic and homeostatic systems share overlapping circuits (Berridge, 1996; Rossi & Stuber, 2018).

A forager should stay in a patch until the food runs out.

It should leave when its current intake rate falls to the average rate available in the environment, counting travel time—usually long before the patch is exhausted. Staying to depletion lowers the long-run rate (Charnov, 1976).

Food preferences and aversions are mostly innate.

Much of diet is learned. A single pairing of a novel taste with illness produces a lasting aversion, and animals acquire specific hungers for foods that correct deficiencies (Garcia & Koelling, 1966; Rozin & Kalat, 1971).

A meal ends when a single general satiety signal is reached.

Satiety is partly food-specific: the pleasantness of a food just eaten falls faster than that of foods not eaten, so a varied meal is a larger meal than any single food would produce (Rolls et al., 1981).

Glossary

Appetitive phase.

The flexible, goal-directed phase of feeding in which an animal searches for and selects food, as distinct from the act of ingestion.

Consummatory phase.

The relatively stereotyped phase of feeding in which food at hand is ingested.

Feeding behavior.

The set of actions by which an animal locates, selects, and consumes food; in MeSH, a form of animal behavior.

Hedonic hunger.

The drive to eat for the reward value of food, which can operate independently of—or against—the body's energy needs.

Homeostatic regulation.

The matching of food intake to energy and nutrient need through hunger and satiety signals, defending body energy stores around a set point.

Incentive salience (“wanting”).

The motivational component of reward that makes a food's cues attract and drive pursuit, dissociable from the food's hedonic “liking”.

Learned taste aversion.

A lasting avoidance of a flavor acquired after it is paired, even once and at long delay, with gastrointestinal illness.

Marginal value theorem.

The optimal-foraging rule that a forager should leave a depleting patch when its current intake rate falls to the average rate available in the environment, including travel time.

Optimal foraging theory.

The framework that treats foraging decisions as maximizing net energy gain per unit time subject to constraints such as travel time and handling cost.

Patch.

A localized concentration of food within an environment, depleting as it is exploited and separated from other patches by travel time.

Postingestive feedback.

The metabolic consequences of eating a food, by which an animal updates its preference for or aversion to that food.

Satiety.

The physiological state of fullness that terminates a meal and suppresses further intake, signaled by gut peptides and stomach distension.

Sensory-specific satiety.

The decline in the pleasantness of a food that has just been eaten, greater than for foods not eaten, so that dietary variety increases total intake.

Set point.

The body-energy level that homeostatic regulation defends, raising hunger when stores fall below it and satiety when they rise above.

Key Researchers

Kent C. Berridge

(University of Michigan). Affective neuroscientist whose dissociation of “wanting” (incentive salience) from “liking” (hedonic impact) reframed the psychology and neuroscience of food reward. ORCID

Hans-Rudolf Berthoud

(Pennington Biomedical Research Center). Neurobiologist of nutrition whose work weighs metabolic against hedonic drives in the neural control of appetite and their integration in obesity. ORCID

John Garcia

(1917–2012; University of California, Los Angeles). Psychologist who discovered conditioned taste aversion, showing that feeding-related learning is biologically prepared to associate taste selectively with internal illness. Wikipedia

David Raubenheimer

(University of Sydney). Nutritional ecologist and co-architect of the geometric framework of nutrition, extending the regulation of multiple nutrients from animal foraging to human diet. ORCID

Paul Rozin

(University of Pennsylvania). Psychologist of food choice and disgust whose account of specific hungers and poison avoidance framed diet selection as adaptive, domain-specific learning. Wikipedia

Stephen J. Simpson

(University of Sydney). Entomologist and co-architect of the geometric framework of nutrition, which treats feeding as navigation through nutrient space toward an intake target. ORCID

Frequently Asked Questions

What is feeding behavior?

Feeding behavior is the set of actions by which an animal locates, selects, and consumes food. It includes an appetitive phase of searching and choosing and a consummatory phase of ingestion, and in MeSH it is classified as a form of animal behavior (Morton et al., 2006).

What is the difference between the appetitive and consummatory phases?

The appetitive phase is the flexible, goal-directed search for and selection of food; the consummatory phase is the relatively stereotyped act of ingestion once food is at hand. The two are controlled differently, which is why an animal can be motivated to seek food without eating much, or eat heavily without having searched.

What does optimal foraging theory predict?

It predicts that foragers behave so as to maximize net energy gain per unit time. The marginal value theorem specifies when to leave a depleting patch: when the current intake rate falls to the average rate available in the environment, counting travel time (Charnov, 1976; Pyke et al., 1977).

How does the body know when to eat and when to stop?

Hormonal signals of energy reserves (such as leptin and insulin) and meal-related signals from the gut converge on the hypothalamus and brainstem, which integrate them to generate hunger and satiety and defend body energy around a set point (Morton et al., 2006; Watts et al., 2022).

Why do we overeat palatable food when we are not hungry?

Because food has a reward value—a hedonic drive—that can operate independently of energy need. The “wanting” and “liking” of palatable food can push intake above requirement, and the hedonic and homeostatic systems share overlapping brain circuits (Berridge, 1996; Rossi & Stuber, 2018).

What is conditioned taste aversion?

It is a learned avoidance of a flavor that has been paired with gastrointestinal illness, even after a single pairing and a long delay. Garcia and Koelling showed that taste is selectively associated with illness, while lights and sounds are not—evidence that food learning is biologically prepared (Garcia & Koelling, 1966).

What is sensory-specific satiety?

It is the decline in the pleasantness of a food that has just been eaten, which is greater than the decline for foods not eaten. Because each food satiates largely on its own, a meal offering variety is eaten in larger total quantity than any single food would be (Rolls et al., 1981).

Are food preferences learned or innate?

Both, but learning dominates diet. Animals acquire specific hungers for foods that correct nutritional deficiencies and lasting aversions to foods that made them ill, so the same species eats differently in different environments (Rozin & Kalat, 1971).

References