When neuroscientist Gregory Berns and his wife set out to build a small, regenerative farm in rural Georgia, they didn’t expect cows to become central to his scientific curiosity. But as Berns began caring for a small herd of miniature zebus, he found himself drawn into a deeper exploration of their emotional lives, social intelligence, and communication. 

In Cowpuppy, Berns weaves together personal storytelling, careful observation, and accessible neuroscience to reveal a surprisingly rich inner world behind those familiar brown eyes. In this excerpt, he takes us inside one deceptively simple question: What does a cow actually mean when she moos? 

You can also explore more of Berns’ work and perspective in our recent conversation, where we dive deeper into animal cognition, interspecies connection, and what cows — and dogs — can teach us about ourselves.

Everyone knows what a cow sounds like. The cow says Moooo! A classic example of onomatopoeia, a word that sounds like its meaning.

Or is it? In English, the cow may say moo, but in French, they meuh. In Italian, Spanish, and Portuguese it’s mu. Welsh cows say the unpronounceable mw. Danish cows go muh, but in Dutch they boe. While most of the world seems to agree on the monosyllabic /m/ sound, to my ear it wasn’t really an /m/ sound, or even one syllable. If we’re going for acoustic verisimilitude, a few languages get pretty close to cow-talk. In Finnish, it’s ammuu. In Bengali, it’s hamba. But I think the closest to what cows really sound like, especially with emphasis on the second syllable, is heard in Tagalog: ungaa; or Korean: eum mae. Of course, my cows were of the Asian variety, so that may have had something to do with the similarity of their vocalizations to languages of southeast Asia—a cattle dialect of sorts.

There is, in fact, tremendous variation in the sounds cows make. Each cow has a repertoire of vocalizations that presumably mean different things, and no cow sounds exactly like another. As in the human voice system, cow vocalizations originate with vibration of the vocal folds of the larynx. The emerging sounds are then modulated by the configuration of the tongue, mouth, and lips. Sometimes a cow will vocalize with her lips closed, in which case it really does sound like mooo. Other times she will keep her mouth open, resulting in a nasal version, like Steve Urkel might do if he were a cow. And then there’s lowing, a deeper, more resonant sound that can be heard at great distance.

It is a generally accepted tenet of biology that the reason animals vocalize is to communicate with one another. But for this to work, there has to be both a sender and a receiver. Although humans talk to themselves, other animals don’t. Even the solitary wolf, seemingly howling at the moon, is communicating with other wolves, who may be miles away. He might be calling for a mate or warning others to stay away from his territory. Vocalization is intentional and meant to convey information to a recipient.

To understand what cows are saying, we have to examine mooing from both the sender’s and receiver’s perspective. What is the sender trying to convey? And what do they expect the receiver to do? If we understood the receiving end, this would open the door to verbal communication, like people have with dogs.

All of my cows mooed, some more than others, and the females were generally more vocal than the males. It wasn’t until the third season that I could say this with some confidence because it wasn’t until then that I had enough males and females to observe. The boys were more physical, but the girls were more vocal. A bit like humans, I guess.

Ethel had always been the most vocal of the herd. She mooed nearly every evening when it was time for the grain. She mooed incessantly at each of her calves until they were a month old, to the point that it appeared they learned to ignore her entreaties. Sometimes I would hear her mooing in the pasture for reasons only she knew. Unless it was repetitive or sounded particularly urgent, I ignored her, just like the rest of the herd did. (Okay, I didn’t really know if the herd ignored her, but there were no obvious reactions from the other cows when Ethel was doing her regular mooing.)

Although Ethel was the most vocal of the herd, the others mooed from time to time, and for a few weeks after Xena’s calf was born, she mooed even more than Ethel. As a first-time mom, Xena did not possess the knowledge and security of a seasoned cow, like her momma, Lucy. Newborn calves don’t do much other than drink milk and sleep. The momma, though, has an increased caloric requirement due to lactation and needs to eat roughly 50 percent more than normal. Poor Xena was torn between filling her belly with grass and checking on her calf. She would graze for a bit with Luna sleeping nearby, but as soon as Xena moved on to fresh forage she would bellow at Luna until she got up and followed her to the new location. Ethel had done the same with all of her calves, but Lucy radiated confidence and rarely needed to moo. Her calves just followed her.

Except when a calf managed to slip through the fence. It was inevitable. Every one of the calves, at some point, found their way to the wrong side of, the fence. They had a tendency to nap beneath the rails while their mommas were grazing, and I think when they woke up, they just rolled the wrong way. Other times, they were curious and squeezed underneath. However they got out, though, the momma went into a tizzy, running back and forth along the fence line, mooing desperately at the wayward calf to get back in, which, of course, they never seemed to figure out. One time, when Tex was a calf, we found him curled up under a tree in the forest. When Lucy mooed, I knew something was up. The only clue to his approximate location was the direction Lucy was aiming her bellows.

The boys’ vocalizations were entirely different than the girls’. They rarely mooed, and when they did, it was a harsh sound, almost like a donkey braying. The only times Ricky Bobby mooed were when he was separated from the herd. This didn’t happen often, just when he was castrated and when he was having his hooves trimmed. I think his braying was more from the pain of separation than the physical pain. BB’s moo sounded more like a grunt, but he, too, did it only when he inadvertently got separated from the herd, like when he lingered in the barn too long and the herd decided to leave. Poor BB would come out and, realizing he was alone, call out for his friends.

These descriptions would seem to imply that cows moo mainly when they’re in distress. Temple Grandin used cow vocalizations as a type of negative metric of how well they were being handled at slaughter. Vocalization is also increased when calves are separated from their mothers (more so with the mothers than the calves). The same is true for dairy cattle. Fo these reasons much of the cattle industry has gravitated toward mooing as a marker of stress. A quiet cow is a happy cow. However, the context of any vocalization must be taken into account. Slaughterhouses and dairies are inherently stressful situations, so any vocalization in those contexts will necessarily indicate physical or emotional stress. What about other, nonstressful situations?

Although my cows vocalized most prominently when they were unhappy about something, they also mooed when they appeared perfectly fine. Like Ethel, they occasionally mooed in excitement when I was about to give them treats or when I was moving them to a fresh pasture. When the mommas mooed at their calves, sometimes they were just saying, Hey! Let’s go! Other times, I think they were teaching the calves what their momma sounded like. Cows identify each other by their signature calls. One study found that three-to five-week-old calves can differentiate recordings of their mothers from other cows. And it’s not just the cows identifying one another. I could do it, too, just as I could identify each of my dogs by how they sounded.

In an effort to understand cattle vocalizations, researchers have put considerable effort into decoding moos. The classic approach was through acoustic analysis, which relied on a spectrogram. A sort of vocal fingerprint, a spectrogram plots how different audio frequencies change over the course of a vocalization. They are especially common in the analysis of birdsong. When applied to speech, acoustic analysis is termed phonetics. The fundamental unit of phonetics is a formant, which describes the frequency of a sound. Speech, though, is composed of several frequencies, each changing in amplitude during the course of a vocalization. The formant with the lowest frequency is called F1, the second F2, and so on. When this type of analysis was done on cow vocalizations, the average F1 was 790 Hz and F2 was 1942 Hz. This combination of F1 and F2 sounds similar to the formants produced when an adult human male makes the long a sound.

While formants describe vowel sounds, speech involves a lot more than that. How a formant is initiated and how it ends determines the basic unit of human speech, called a phoneme. Even though there are five vowels in written English, there are twenty to twenty-four phonemes that describe how to pronounce vowel sounds. For example, in grade school, kids are taught the difference between the short a sound in cat (abbreviated /a/) and the long a of cake (/./). There are another twenty or so phonemes for the consonants. Cattle vocabulary, if there is one, would have to use a much more restricted set of phonemes because cows don’t have nearly the degree of vocal control humans do.

A single cow vocalization lasts one to one and a half seconds, but there are several ways a cow can alter how she moos. Presumably, these alterations are done intentionally and, therefore, contain information that is being transmitted to someone—another cow, a human, or maybe another farm animal. One study found that moos could be divided into open-and closed-mouth types. With mouth closed, a cow produces a lower frequency formant, and these types of moos are more frequent when a momma is near her calf, especially during the first month after birth. With mouth open, the sound projects farther and is a higher pitch. These types of moos are used when momma and calf are separated and the momma is calling for her calf. Further analysis has revealed that individual cows could be distinguished by differences in the formant frequencies—a vocal signature—but only in the closed-mouth moos. The authors speculated that vocal identification is more important when cows and calves are near each other, with open-mouth moos serving as a more general call. Taking this type of analysis one step further, Australian researchers determined that vocal signatures are maintained across different emotional states. Cows were assumed to be in a positive emotional state when the farmer was getting their feed ready and a negative state when they were isolated or denied access to food, but the cows were recognizable by their moos in both positive and negative contexts.

No study has yet been able to differentiate positive and negative emotional states just from mooing. It may be that cows, like most animals, have no need of mooing when they’re happy, instead reserving vocalization for distress and warning calls. But that wouldn’t explain Ethel’s moos during the evening grain. No, vocal decoding is a hard problem. Machine-learning algorithms are just beginning to get a handle on decoding human emotion from human speech.

The difficulty is compounded in domesticated species. Because humans have altered their natural evolution, the vocalizations of domesticated animals may be directed at either their compatriots or humans. Wild animals, in contrast, have no need to communicate with humans, so their vocalizations are directed at one another or occasionally to frighten off potential threats.

In terms of domesticated animals, the most progress in understanding vocalizations is with dogs. Dogs are evolved from wolves, which are famously vocal, and this may have something to do with dogs’ ability to understand human speech, as well as our ability to decipher their vocalizations. Research has shown that people are generally quite good at recognizing some dog emotions from their vocalizations, such as aggressive barks at strangers versus barks when they’re playing, although this ability is largely dependent on a given person’s experience with and thus exposure to dogs. In general, low-pitched barks are understood to be aggressive in nature, whereas high-pitched barks convey a more positive emotional state, like playfulness.

In fact, this high-low vocal rule has been observed in many animals. Harsh, low pitches signal hostile intent, and purer, high pitches mean friendliness. There is a simple explanation for this relationship. Physics dictates that large animals will have lower-pitched vocalizations. To the extent that an individual can vary the pitch of its sounds, lowering the pitch will make them sound bigger and more threatening. Conversely, raising the pitch will have the opposite effect, signaling that they are small—even juvenile—and don’t want any conflict. Although this rule has not been studied with cattle, it likely holds true.

If humans can broadly recognize whether an animal vocalization is hostile or friendly, it is straightforward to program a computer to do the same. It’s really a matter of having a lot of data. An early decoder of dog barks, for example, could tell the difference between barks produced in six situations: playing, fighting, walking, being left alone, being approached by a stranger, and being shown a ball. However, the algorithm required six thousand barks and was about 50 percent accurate (substantially better than guessing, which would be correct 17 percent of the time). To make a moo-alyzer, you would need a comparable number of audio samples, about one thousand per type of moo. Doable, but a person would still need to hand code what each moo represented before training a neural net on the data. One promising approach to decoding cow vocalizations recognizes that moos often occur in sequence, almost like a sentence. Instead of focusing on a single moo, the transitions between open-and closed-mouth mooing suggest a simple way to understand what a cow is saying, especially in the context of mother-calf interactions. In a typical sequence, the momma uses an open-mouth

moo to call her calf and then transitions to closed-mouth when they are reunited.

Mooing isn’t even the whole story of cow communication. Cows make other types of vocalizations more frequently than moos. They grunt and snort. Ethel, as the most vocal of the herd, was a prolific snorter too. She would only do this when her calves were doing something she didn’t like, usually by wandering off too much for her comfort. What was interesting was that Ethel used a graded level of snorting. Mild irritation was signaled by a soft, brief snort. As she became more unhappy, this would increase in volume and stridor to the point of sounding like a grunt. The best way I can describe it is how Marge Simpson sounds when she grunts in disapproval. Although scientists have studied cow mooing, there has been no research on these types of subvocalizations in cattle. However, grunts have been categorized in other ungulates. Deer have been said to emit four types of grunts/snorts distinguishable by their acoustic parameters: the grunt, the alert-snort, the snort-wheeze, and the aggressive snort. All but the alert-snort are produced when deer are near each other, some being affiliative and others aggressive. The alert-snort is a loud propulsion of air by a lone deer warning others within earshot of a potential threat. I heard alert-snorts every day when I walked through the woods around the farm. It seems likely that cows have a similar vocabulary. I would often hear the cows emit brief, high-pitched grunts when they played with each other. This was especially true of the calves, but everyone did it when they got going, like when they played in the sand pile.

There was another type of cow vocalization that I could find no reference to. I call it cooing. This was something all the calves did. When I crouched down and called them over, they would often nuzzle my neck, licking it while making soft snuffling sounds, which sounded like the coos of an infant. It was so quiet, you could only hear it if you were on the receiving end. I assumed the calves did it with their mommas too. Sometimes even the adults cooed. Xena tended to do it a lot when she wanted neck scratches. The coo of human babies is a potent releaser of oxytocin in mothers, and scientists think this promotes and maintains bonds with the parents. I think cows do the same thing, which would explain why I became so attached to them, especially the calves.

Taken from Cowpuppy by Gregory Berns. Copyright © 2024 Gregory Berns. Used by permission of Harper Horizon, an imprint of HarperCollins Focus, LLC. www.harpercollinsfocus.com/