How Animals Learn: Learning Theory, Training, and Behaviour Modification

Learning theory explains how animals change their behaviour through experience. These principles apply broadly across species, though each animal’s senses, motivation, emotional state, and normal behavioural needs shape how learning appears in daily life. For owners, learning theory is useful because it explains why behaviour develops, why some training works, why some methods fail, and why behaviour modification must address both behaviour and emotion.

Animals learn constantly. A dog may learn that barking at the front window makes the mail carrier walk away. A cat may learn that hiding behind the washing machine prevents being put in a carrier. These behaviours are not moral choices. They are patterns shaped by consequences, associations, context, and prior experience.

Classical conditioning is learning by association. One event begins to predict another event, especially one that matters emotionally. For example, if a cat only sees the carrier five minutes before a car trip to the clinic, the carrier may begin to predict restraint, travel, unfamiliar smells, and handling. The cat may hide before anyone reaches for it. If a dog hears the doorbell and then strangers enter the home, the doorbell may begin to trigger barking before the door opens. In 1980, Pearce and Hall described Pavlovian learning as a process in which cues become more or less important depending on how well they predict meaningful outcomes.

Classical conditioning matters because it changes emotion, not only visible behaviour. A fearful animal does not simply need to “stop reacting.” The animal needs the trigger to become safer, more predictable, and less threatening. A dog that barks and lunges at dogs across the street may need to start at a distance where it can still eat, orient to the owner, and recover quickly. At that distance, the appearance of another dog can be paired with food, space, and calm movement away. Over time, the trigger can begin to predict a safe routine rather than panic or conflict.

Operant conditioning is learning by consequence. The animal performs a behaviour, and what happens next changes how likely that behaviour is to happen again. If a dog sits in front of the fridge and receives a piece of cheese, they may sit by the fridge more often. If a cat scratches the sofa and the owner rushes over, the cat may scratch more because it reliably elicits attention. If a parrot bites when a hand enters the cage and the hand withdraws, biting may increase because it creates distance. Operant learning is not about whether the animal is “good” or “bad.” It is about whether the behaviour works.

Reinforcement means that a consequence increases behaviour. Positive reinforcement means something desirable is added after the behaviour. This may include food, play, access to a favourite person, access to a perch, permission to sniff, or distance from a trigger. Negative reinforcement means something unpleasant is removed after the behaviour. For example, leash pressure may stop when a dog turns back toward the owner, or halter pressure may stop when a horse steps forward. Both forms of reinforcement increase behaviour. The terms positive and negative mean “added” and “removed,” not “good” or “bad,” not “kind” or “cruel.”

Punishment is a consequence that decreases behaviour. Positive punishment means something unpleasant is added after the behaviour, such as yelling, leash corrections, spray, shock, or physical intimidation. Negative punishment means something desirable is removed, such as ending play when a puppy bites too hard. Punishment can suppress behaviour, but suppression is not the same as emotional change. If a fearful dog is punished for growling at a child, the growling may stop while the fear remains. The warning signal may disappear, but the underlying risk may increase.

Training methods matter. In 2008, Blackwell et al. reported that dogs trained using punishment-based methods were more likely to exhibit behaviour problems than those trained using reward-based methods. A practical example is a dog that is corrected for barking at visitors. The barking may decrease temporarily, but the dog may still feel unsafe when visitors enter. A better plan teaches the dog to go to a mat, reinforces quiet behaviour at a manageable distance, and changes the prediction that visitors create conflict.

More welfare-focused studies point in the same direction. In 2014, Deldalle and Gaunet found that dogs trained with negative reinforcement showed more stress-related behaviours and less attention to their owners than dogs trained with positive reinforcement. In 2020, Vieira de Castro et al. reported that aversive-based training methods were associated with poorer welfare indicators in companion dogs, both during training and outside the training context. For owners, the practical message is direct: methods that rely on fear, discomfort, or intimidation may stop behaviour in the moment, but they can worsen stress and reduce trust.

Observational learning also matters. Animals may learn by watching people or other animals. A young dog may follow an older dog through a dog door. A cat may approach a food puzzle after watching another cat use it safely. In 2016, Fugazza et al. showed that dogs could recall and reproduce human-demonstrated actions after incidental encoding, supporting the view that animals can learn some behaviours through observation. Observation can help, but it does not replace direct reinforcement, safety, and structured practice.

Food aversion is another important learning process. Animals can learn to avoid a food after nausea, vomiting, illness, or distress, even when the food did not actually cause the illness. This is called conditioned taste aversion. A cat that feels nauseated after eating a new kidney diet may refuse that diet later, even if the nausea came from disease rather than the food. A dog that vomits after receiving medication hidden in cream cheese may begin refusing cream cheese, tablets, or both. In 1966, Garcia and Koelling showed that animals form selective associations between taste cues and illness, meaning that some associations are biologically easier to learn than others.

Habituation and sensitization are also central to behaviour modification. Habituation occurs when an animal gradually stops responding to a repeated, harmless stimulus. A puppy may stop reacting to the dishwasher if the sound occurs daily and nothing frightening follows. Sensitization is the opposite: the animal becomes more reactive after repeated or intense exposure. A cat repeatedly chased by a dog may become increasingly reactive to footsteps, tags jingling, or movement near a doorway. Repetition alone is not therapy. Exposure helps only when the animal remains below the point at which fear, panic, or defensive behaviour takes over.

Extinction occurs when a behaviour no longer produces the expected outcome. A dog that paws at the table for food may paw harder, bark, or jump when food stops appearing. This temporary escalation is called an extinction burst. The animal is testing whether the old strategy still works. Extinction is rarely enough on its own because it can create frustration. It is more effective and fairer to reinforce a replacement behaviour, such as lying on a mat during meals, while preventing access to the table.

Training works best when it combines management, reinforcement, and emotional change. Management prevents repeated practice of the unwanted behaviour. Reinforcement teaches the replacement behaviour. Emotional work changes the animal’s response to the trigger. For example, a dog that jumps on guests can be managed with a gate before the door opens. The dog can be reinforced for standing with four paws on the floor, then for going to a mat, then for remaining on the mat while the guest enters. At the same time, guest arrival can predict calm food delivery rather than chaotic social contact.

The same principles apply to carrier training. The carrier should not appear only before stressful travel. It should remain available, contain comfortable bedding, and predict food, rest, and voluntary entry. Training can begin by placing treats near the carrier, then inside the entrance, then at the back. The door should close only briefly after the animal is relaxed, then enter and stay inside. Movement of the carrier, placement in the car, and short trips should each be trained separately. Each step should be easy enough that the animal can succeed.

Good training also requires timing and clarity. The reinforcer must occur close enough to the behaviour for the animal to connect the two. A marker signal, such as a tongue click or marker word, can identify the exact behaviour that earned reinforcement. In 2016, Chiandetti et al. compared food alone, a spoken word, and a mechanical marker when shaping novel behaviours in dogs, providing a direct study of secondary signals in reinforcement-based training. The marker is not inherently meaningful. It becomes meaningful because it is repeatedly followed by something the animal values.

The best training question is not “How do I stop this?” The better question is “What do I want the animal to do instead, and what does the animal need to feel safe enough to learn it?” If the behaviour is normal but inconvenient, such as scratching, chewing, digging, vocalizing, or seeking attention, the plan should redirect it to an acceptable outlet. If the behaviour is driven by fear, anxiety, or frustration, the plan must reduce the emotional load before expecting reliable responses. If the behaviour is reinforced by escape, attention, food, access, or distance, the plan must change the pattern of consequences.

Learning theory makes training fairer and more effective. It prevents owners from blaming animals for behaviours that have been accidentally reinforced or emotionally conditioned. It also gives owners practical tools: change the prediction, reinforce the behaviour you want, prevent rehearsal of the behaviour you do not want, and avoid methods that create fear or conflict. The goal is not obedience at any cost. The goal is a safer, more predictable, and more emotionally stable animal.

References

Blackwell, E.-J., Twells, C., Seawright, A. M., and Casey, R. A. 2008. The relationship between training methods and the occurrence of behaviour problems, as reported by owners, in a population of domestic dogs. Journal of Veterinary Behavior, 3(5), 207-217. https://doi.org/10.1016/j.jveb.2007.10.008

Chiandetti, C., Avella, S., Fongaro, E., and Cerri, F. 2016. Can clicker training facilitate conditioning in dogs? Applied Animal Behaviour Science, 184, 109-116. https://doi.org/10.1016/j.applanim.2016.08.006

Deldalle, S., and Gaunet, F. 2014. Effects of two training methods on stress-related behaviours of the dog and on the dog-owner relationship. Journal of Veterinary Behavior, 9(2), 58-65. https://doi.org/10.1016/j.jveb.2013.11.004

Fugazza, C., Pogány, Á., and Miklósi, Á. 2016. Recall of others’ actions after incidental encoding reveals episodic-like memory in dogs. Current Biology, 26(23), 3209-3213. https://doi.org/10.1016/j.cub.2016.09.057

Garcia, J., and Koelling, R. A. 1966. Relation of cue to consequence in avoidance learning. Psychonomic Science, 4, 123-124. https://doi.org/10.3758/BF03342209

Pearce, J. M., and Hall, G. 1980. A model for Pavlovian learning: variations in the effectiveness of conditioned but not of unconditioned stimuli. Psychological Review, 87(6), 532-552. https://doi.org/10.1037/0033-295X.87.6.532

Vieira de Castro, A. C., Fuchs, D., Morello, G. M., Pastur, S., de Sousa, L., and Olsson, I. A. S. 2020. Does training method matter? Evidence for the negative impact of aversive-based methods on companion dog welfare. PLOS ONE, 15(12), e0225023. https://doi.org/10.1371/journal.pone.0225023