People talk to their pets every day: offering praise when they’re good, reassurance when they’re confused and affection when they’re cuddling. We also speak to animals when they misbehave. “Why did you do that?” someone might ask their dog. Or we might scold the cat – “Don’t touch that!” – as we move a family heirloom across the room.

But is it ever appropriate to punish or rebuke an animal?

When people talk about “punishment,” this implies more than a loss of privileges. The term suggests someone is being asked to learn a lesson after breaking a rule they can understand. But an animal’s understanding is different from a human’s, which raises questions about what lessons they can learn and what, if any, rebukes of animals are ethical.

These issues involve what researchers know about different animals’ cognition. But they also go beyond this by raising questions about what kind of moral standing animals have and how people who interact with animals should train them.

As an ethical theorist, I’ve explored these and related questions, including with some of my colleagues in psychology and anthropology. I would argue it is important to distinguish three types of learning: conditioning, instruction and education.

Conditioning

One type of learning, called “classical conditioning,” was popularized by the psychologist Ivan Pavlov just after the turn of the 20th century. By repeatedly ringing a bell while presenting food, Pavlov famously induced dogs to salivate from the bell ring alone. Such learning proceeds merely from associating two types of stimuli: a sound and a snack, in this case.

When scientists talk about punishment, they normally mean “operant conditioning,” which was popularized by the psychologists Edward Thorndike and B. F. Skinner shortly thereafter. In operant conditioning, positive or pleasurable stimuli are used to reinforce desired behavior, and negative or painful stimuli are used to deter undesired behavior. We may give a dog a treat, for example, to reward it for following a command to sit.

The kind of learning that operant conditioning aims to achieve, however, lacks a crucial ingredient of human punishment: responsibility. When people punish, it is not just to discourage an undesired behavior. They are trying to drive home that someone has transgressed – that the individual’s behavior merits punishment.

But can nonhuman animals transgress? Do they ever deserve rebuke? I would argue they do – but with key differences from human wrongdoing.

Instruction

Training for many animals, such as horses and dogs, goes beyond conditioning. It involves a more sophisticated kind of learning: instruction.

One important way instruction differs from conditioning is that an instructor addresses their trainee. Pet owners and animal trainers speak to cats and dogs, and though these animals have no knowledge of grammar, they can understand what many human words refer to. Caretakers also often listen to their animals’ vocalizations in an attempt to understand their meaning.

To be sure, people condition cats and dogs – consider spraying a cat with water when it nibbles on a houseplant. The goal is for the cat to associate an off-limits snack with an unpleasant experience, and so to leave the plant alone.

But training pets can go beyond changing their behavior. It can aim to improve animals’ ability to reason about what to do: a trainer teaches a dog how to navigate an agility course, for example, or how to get through a new pet door. Instruction involves understanding, whereas learning based on mere conditioning does not.

An animal’s ability to be instructed stems from the nature of their mental life. Scientists do not know exactly which animals’ cognition involves understanding, genuine problem-solving and the ability to reason or infer.

But research on perception – on how humans and other animals convert sensory information into mental representations of physical objects – has helped philosophers and psychologists distinguish thought from more basic mental capacities such as vision and hearing.

It is extremely likely that some nonhuman animals – including dolphins, apes and elephants – do think, as philosopher Gary Varner argued in the 2012 book “Personhood, Ethics, and Animal Cognition.” My research suggests the distinction between thinking and nonthinking animals tracks well with the distinction between animals that can be instructed and those that can, at most, be conditioned.

This difference is crucial to how different pets should be treated. An owner should have concern for their pet frog, of course, and care for its needs. But they do not need to recognize the frog the same way they should recognize a dog: by addressing it, listening to it and comforting it.

Though an owner may rebuke the dog to hold it responsible for its actions, they must also hold themselves responsible to the animal, including by considering how the pet has interpreted events.

Education

Some nonhuman animals have demonstrated impressive cognitive abilities in experimental settings, such as recognizing their bodies in mirrors and recalling past experiences. Some birds, for example, display sensitivity to details about food they have cached, such as its perishability and how long ago it was stored.

Still, scientists do not possess strong evidence that animals have critical thinking abilities or a concept of self, the key requirements for genuine education. Unlike conditioning and instruction, education aims to enable a learner to explain the world, to evaluate and debate rationales for decisions. It also prepares people to ask – and to try to answer – ethical questions like, “How should I live” and “Was that action justified?”

A cat or dog cannot pose these questions. Much of the time, human beings do not concern themselves with these questions, either – but they can. In fact, caretakers pay great attention to these matters during child-rearing, as when they ask children, “How would you like it if someone did that to you” or “Do you really think it’s OK to act that way?”

Assuming that animals do not reflect and criticize, and therefore are not capable of education, I would say that they have no moral obligations. It is fair to say a pet has transgressed, since animals such as dogs and cats can come to understand how to act better. But morally speaking, an animal cannot commit wrongdoing, for it lacks a conscience: It may understand some of its behavior, but not its own mind.

In my view, addressing an animal and acting with an understanding of how it interprets events is central to the ethical training of pets. But if someone treats an animal as though it were responsible for justifying itself to us, as though it could offer excuses and apologies, they anthropomorphize the animal and ask too much of it. Pet owners often do this in a mock way, saying things like, “Now you know you shouldn’t have done that” – the same phrases they might use with a child.

Unlike a child, however, the animal’s transgression is not a failure to fulfill a moral obligation. In human relationships we aspire to relations of mutual justification, where reasons are exchanged and excuses and apologies evaluated. But that’s not the nature of our relationships with our pets – however tempted we may be to think otherwise.

Jon Garthoff does not work for, consult, own shares in or receive funding from any company or organization that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.