Three sheep walk into a scientific experiment. Looking up at the wall, the first, a ram, says “Hey, that’s Barack Obama!” Upon which the dispenser gives him a food pellet.
The second, a ewe, says “No, silly, that’s a picture of Barack Obama!” Upon which the dispenser gives her two food pellets.
The third, a lamb, pipes up, “Either way, he has ears just like mine!” Upon which they all start laughing, and the dispenser gives the lamb three food pellets.
•
Forgive me. After hours of reading about the visual capacities of ruminants and ungulates, I couldn’t resist.
Emma, Fiona, Jake, and Barack (as the sheep encountered them)
I didn’t find it surprising that sheep can tell one human face from another. Clearly they can distinguish individuals in their flock, and the specifics of their herd’s sheepdogs and shepherds — otherwise they’d panic each time their protectors showed up, taking them for strangers. But, given my long-term interest in vision and images, I wanted to learn exactly how they discriminated between one image of a human and another.
•
That led me to the original experiment itself, as described in a paper published on November 8, 2017 at the Royal Society Open Science website: “Sheep recognize familiar and unfamiliar human faces from two-dimensional images,” by Franziska Knolle, Rita P. Goncalves, and A. Jennifer Morton, all from the Department of Physiology, Development and Neuroscience, University of Cambridge, England. You can read it online there or download it as a pdf file. Some excerpts:
“During the training stages, the sheep were taught to discriminate repeatedly presented photographs of four unfamiliar people (learned- familiar faces) from novel unfamiliar people. We used photographs of four celebrities: Emma Watson, former US President Barak [sic] Obama, newsreader Fiona Bruce and actor Jake Gyllenhaal. These people were chosen because of the availability of high-quality images in different perspectives via the Internet using the Google-search function.
“Having shown that sheep can learn to identify faces of unfamiliar individuals in different perspectives, our final question was whether sheep would also be able to recognize a ‘very familiar’ person from a two-dimensional photographic image. To test this, three copies of a photographic image of one of their regular handlers were randomly interspersed among the stimuli used in the test-probe (four celebrity faces in different perspectives). The handler was one of the two principal handlers who trained the sheep daily. She would routinely spend at least 2 h a day with the sheep. The sheep followed the handler voluntarily and fed from her hand. … Our results show that the photograph of the handler was chosen over the picture of an unknown person 71.8 ± 2.3% of the time … This selection performance was significantly above chance … and reaction times were similar to those of the learned-familiar faces. …
“Our findings extend the understanding of face-recognition abilities of sheep and suggest that sheep possess holistic face-processing abilities.”
•
I find it heartening to note that the experimenters distinguished carefully between photographs of people and the people themselves. It appears, then, that sheep can recognize photographs of their handlers. Regrettably (due perhaps to the difficulties inherent in celebrity scheduling) they did not have the opportunity to round out their well-conceived experiment by seeing if it worked the other way round: Can sheep recognize real people from their photographs? In this case, that would have required Barack, Emma, Fiona, or Jake to have shown up in the flesh, perhaps repeatedly.
Maybe next time. Of course, it wouldn’t have to be celebrities. They could use convenient locals whom the sheep hadn’t ever seen, showing them the photos first and then introducing the people.
Fig. 4 from study of the ability of sheep to recognize faces, Univ. of Cambridge (UK)
More important, from my perspective, the experimenters do not define what they mean by a “human face” from a sheep’s-eye point of view. This may seem self-evident, but I think otherwise, because what comprises a “face” depends to a considerable extent on how we see one, both in an image and in the flesh.
The fovea present in the eyes of birds, fish, reptiles, and simian primates (including humans), those “rods and cones,” enable depth perception and awareness of minute detail in things close to the eyes. Such eyes normally get placed side by side on the front of the head. This enables humans to “read” faces, registering not just the features but nuances of expression — narrowing of the eyes, clenching of the jaw, and other cues.
To compensate for their lack of fovea, the eyes of ruminants and ungulates (sheep fall into both categories) give them an almost panoramic field of sight. As a handout on sheep, “Understanding Sheep Behaviour,” from the Saskatchewan Ministry of Agriculture points out,
“Humans have binocular vision, focusing both eyes simultaneously to achieve good depth perception and clarity for objects directly in front of them. However humans [sic] peripheral vision is very limited. Sheep see the world through a different set of eyes than ours. Sheep have their eyes set on the side of the head. They have a narrow field of binocular vision in front of their head and wide peripheral fields of monocular vision.”
“The goat and sheep’s eye is similar to a human eye, with a lens, cornea, iris and retina. A crucial difference though, is that the retina is shaped like a rectangle. This offers these ungulates massive peripheral vision, a panoramic field of 320-340 degrees! However, this trait tends to reduce the animal’s ability to perceive depth as well as its binocular field to that of only 20-60 degrees. This is often a sacrifice made by prey animals who gain advantage by superior area vision. … Interestingly Sheep and Goats are farsighted with a slight astigmatism, another common trait of prey animals. …
“Due to their hyperopia (farsightedness) with astigmatism sheep likely perceive a well-focused image of objects in middle and long distance ranges. This is very helpful in a prey species. Sheep have the ability to recollect up to 50 unique human or sheep faces for as long as 2 years.”
In short, sheep have a hard time focusing on and registering detail in things directly ahead of and close to them. When confronting a digitally printed head shot of a person in this experiment’s environment, therefore, what do they actually see and notice that enables them to distinguish between one photo and another? The specific features and their distribution across the visage? The 2-D outline of the face? The paper by Knolle et al does not address this, alas.
•
All of this matters, by the way, because sheep have much larger brains than lab mice and rats, enabling their use in studies of “neurodegenerative diseases … in which cognitive flexibility is impaired,” among them “Huntington’s disease … and Parkinson’s disease, as well as psychiatric disorders such as autism spectrum disorder and schizophrenia.” So any improved understanding of the perceptual systems of sheep has (forgive the pun) ramifications for the medical treatment of humans.
And sheep have value beyond such recent service to science and medicine, not to mention their millennia of providing wool, milk, and meat:
•
Special offer: If you want me to either continue pursuing a particular subject or give you a break and (for one post) write on a topic — my choice — other than the current main story, make a donation of $50 via the PayPal widget below, indicating your preference in a note accompanying your donation. I’ll credit you as that new post’s sponsor, and link to a website of your choosing. Include a note with your snail-mail address (or email it to me separately) for a free signed copy of my 1995 book Critical Focus!
But wait! There’s more! Donate now and I’ll include a copy of The Silent Strength of Liu Xia, the catalog of the 2012-13 touring exhibition of photos by the dissident Chinese photographer, artist, and poet, currently in her sixth year of extralegal house arrest in Beijing. The only publication of her photographic work, it includes all 26 images in the exhibition, plus another 14 from the same series, along with essays by Guy Sorman, Andrew Nathan, and Cui Weiping, professor at the Beijing Film Academy.
SPJ Research Award 2014
Ceçi N’est Pas Un Obama
Stop me if you’ve heard this one:
Three sheep walk into a scientific experiment. Looking up at the wall, the first, a ram, says “Hey, that’s Barack Obama!” Upon which the dispenser gives him a food pellet.
The second, a ewe, says “No, silly, that’s a picture of Barack Obama!” Upon which the dispenser gives her two food pellets.
The third, a lamb, pipes up, “Either way, he has ears just like mine!” Upon which they all start laughing, and the dispenser gives the lamb three food pellets.
•
Forgive me. After hours of reading about the visual capacities of ruminants and ungulates, I couldn’t resist.
Emma, Fiona, Jake, and Barack (as the sheep encountered them)
You might well ask: Just how did this topic cross my radar screen? As a result of several news stories reporting on a recent British experiment, I answer, among them “Sheep learned to recognize photos of Obama and other celebrities, neuroscientists say,” by Ben Guarino, in the November 7, 2017 Washington Post, and “Sheep ‘can recognise human faces'” by Paul Rincon, Science editor at BBC News, from November 8, 2017.
I didn’t find it surprising that sheep can tell one human face from another. Clearly they can distinguish individuals in their flock, and the specifics of their herd’s sheepdogs and shepherds — otherwise they’d panic each time their protectors showed up, taking them for strangers. But, given my long-term interest in vision and images, I wanted to learn exactly how they discriminated between one image of a human and another.
•
That led me to the original experiment itself, as described in a paper published on November 8, 2017 at the Royal Society Open Science website: “Sheep recognize familiar and unfamiliar human faces from two-dimensional images,” by Franziska Knolle, Rita P. Goncalves, and A. Jennifer Morton, all from the Department of Physiology, Development and Neuroscience, University of Cambridge, England. You can read it online there or download it as a pdf file. Some excerpts:
“During the training stages, the sheep were taught to discriminate repeatedly presented photographs of four unfamiliar people (learned- familiar faces) from novel unfamiliar people. We used photographs of four celebrities: Emma Watson, former US President Barak [sic] Obama, newsreader Fiona Bruce and actor Jake Gyllenhaal. These people were chosen because of the availability of high-quality images in different perspectives via the Internet using the Google-search function.
“Having shown that sheep can learn to identify faces of unfamiliar individuals in different perspectives, our final question was whether sheep would also be able to recognize a ‘very familiar’ person from a two-dimensional photographic image. To test this, three copies of a photographic image of one of their regular handlers were randomly interspersed among the stimuli used in the test-probe (four celebrity faces in different perspectives). The handler was one of the two principal handlers who trained the sheep daily. She would routinely spend at least 2 h a day with the sheep. The sheep followed the handler voluntarily and fed from her hand. … Our results show that the photograph of the handler was chosen over the picture of an unknown person 71.8 ± 2.3% of the time … This selection performance was significantly above chance … and reaction times were similar to those of the learned-familiar faces. …
“Our findings extend the understanding of face-recognition abilities of sheep and suggest that sheep possess holistic face-processing abilities.”
•
I find it heartening to note that the experimenters distinguished carefully between photographs of people and the people themselves. It appears, then, that sheep can recognize photographs of their handlers. Regrettably (due perhaps to the difficulties inherent in celebrity scheduling) they did not have the opportunity to round out their well-conceived experiment by seeing if it worked the other way round: Can sheep recognize real people from their photographs? In this case, that would have required Barack, Emma, Fiona, or Jake to have shown up in the flesh, perhaps repeatedly.
Maybe next time. Of course, it wouldn’t have to be celebrities. They could use convenient locals whom the sheep hadn’t ever seen, showing them the photos first and then introducing the people.
Fig. 4 from study of the ability of sheep to recognize faces, Univ. of Cambridge (UK)
More important, from my perspective, the experimenters do not define what they mean by a “human face” from a sheep’s-eye point of view. This may seem self-evident, but I think otherwise, because what comprises a “face” depends to a considerable extent on how we see one, both in an image and in the flesh.
The fovea present in the eyes of birds, fish, reptiles, and simian primates (including humans), those “rods and cones,” enable depth perception and awareness of minute detail in things close to the eyes. Such eyes normally get placed side by side on the front of the head. This enables humans to “read” faces, registering not just the features but nuances of expression — narrowing of the eyes, clenching of the jaw, and other cues.
To compensate for their lack of fovea, the eyes of ruminants and ungulates (sheep fall into both categories) give them an almost panoramic field of sight. As a handout on sheep, “Understanding Sheep Behaviour,” from the Saskatchewan Ministry of Agriculture points out,
“Humans have binocular vision, focusing both eyes simultaneously to achieve good depth perception and clarity for objects directly in front of them. However humans [sic] peripheral vision is very limited. Sheep see the world through a different set of eyes than ours. Sheep have their eyes set on the side of the head. They have a narrow field of binocular vision in front of their head and wide peripheral fields of monocular vision.”
To which add this, from the archives of Evergreen State College in Olympia, WA:
“The goat and sheep’s eye is similar to a human eye, with a lens, cornea, iris and retina. A crucial difference though, is that the retina is shaped like a rectangle. This offers these ungulates massive peripheral vision, a panoramic field of 320-340 degrees! However, this trait tends to reduce the animal’s ability to perceive depth as well as its binocular field to that of only 20-60 degrees. This is often a sacrifice made by prey animals who gain advantage by superior area vision. … Interestingly Sheep and Goats are farsighted with a slight astigmatism, another common trait of prey animals. …
“Due to their hyperopia (farsightedness) with astigmatism sheep likely perceive a well-focused image of objects in middle and long distance ranges. This is very helpful in a prey species. Sheep have the ability to recollect up to 50 unique human or sheep faces for as long as 2 years.”
In short, sheep have a hard time focusing on and registering detail in things directly ahead of and close to them. When confronting a digitally printed head shot of a person in this experiment’s environment, therefore, what do they actually see and notice that enables them to distinguish between one photo and another? The specific features and their distribution across the visage? The 2-D outline of the face? The paper by Knolle et al does not address this, alas.
•
All of this matters, by the way, because sheep have much larger brains than lab mice and rats, enabling their use in studies of “neurodegenerative diseases … in which cognitive flexibility is impaired,” among them “Huntington’s disease … and Parkinson’s disease, as well as psychiatric disorders such as autism spectrum disorder and schizophrenia.” So any improved understanding of the perceptual systems of sheep has (forgive the pun) ramifications for the medical treatment of humans.
And sheep have value beyond such recent service to science and medicine, not to mention their millennia of providing wool, milk, and meat:
•