Valerie V. Peterson
petersov@gvsu.edu
Introduction
Vision opens the door to human image-making and symbol-use. We share the sense of sight with other animals, but it is, for us, more than a survival-enhancing function of the eye and brain. The claim offered here is that humans are essentially ‘image-making’ and ‘image-beholding’ animals—an accomplishment enabled by vision,[i] and a designation bringing with it special freedoms and responsibilities. A phenomenological approach to humans and to images supports this claim.
This philosophical approach to vision as it relates to human freedom is indebted to Hans Jonas’s The Phenomenon of Life: Toward A Philosophical Biology (1966). In this painstakingly detailed scholarly treatise, Jonas shows how the mind is prefigured throughout organic existence, from metabolism to perception to free action—and ultimately to capacities for thought and morally responsible behavior. In the middle of the book, he theorizes the birth of human’s special agency: symbolic action. He moves from explaining the essence of human vision and how it differs from the other senses, to showing how visual representation (image-making and beholding) is the best means of differentiating humans from other animals. This article is an effort to translate and expand upon Jonas’s ideas, illustrate them with contemporary examples, and share their implications with a broader audience.
Motivation for this inquiry comes from concerns about human behavior in non-human entities, particularly AI. One approach to discovering such behavior has been the Turing Test. The Turing Test started out as a thought-experiment in which human respondents are pitted against a machine—to test the machine’s ability to display “human” responses and intelligence. This test is still being used to measure the “human-ness” of AI, but it’s not the only way, nor is it the best way, to approach the matter.[ii] Many people have been temporarily fooled by a robo-call or catfished by a chat-bot, and most of us know of someone cheating their way into a grade or role that they didn’t deserve. But we understand that simply “fooling” a judge or a measure is not the same thing as qualifying for that designation or that measure. To know if a thing ‘qualifies’ as something, we need to know its essence.
Phenomenology is practiced by people interested in understanding phenomena (things, senses, feelings, perceptions, desires, emotions, and other ‘entities’) which may or may not be physically present. A phenomenological approach to a question is experiential rather than experimental, requiring close attention and description. Phenomenologists start with the experiences and perspectives of people themselves—the person or people experiencing phenomena—rather than with scientific instruments or methodologies. One of the main goals of phenomenology is to discover what is invariant (not varying/unchanging/essential) about particular phenomena under investigation. In this inquiry, the phenomena of interest are first, ‘the human’ itself, and second, ‘the image.’ Consideration is given to what the ‘essence’ of a human might be. A phenomenological exploration of images and a few observations follow.
Human Essence: An Inquiry
We begin by pretending we are explorers from another planet in search of a human. This way of approaching the task of ‘invariance-finding’ is appropriately demanding and avoids relying on ‘the usual’ shortcuts of recognition. We are looking for the best indication of a human: the most basic, ‘essential’ difference between humans and other living things on the planet—and we must be careful not to jump to conclusions.
Most of the time, humans identify other humans by sight. We see something with a head, two arms, a torso, and two legs, and we assume it’s a human. However, mannequins, statues, and wax sculptures look like humans (have morphological similarity), but are not alive. Pictures and video recordings also look (and sound) like humans, but are not alive.[iii] Non-human primates look something like humans, and can stand for a time on two legs (as can prairie dogs).
At the other end of the spectrum, some humans are ‘unusual looking’ and might not pass a ‘looks like a human’ test. Some people, for example, can’t stand up (babies, the paraplegic). Some people are missing limbs and/or are significantly deformed (e.g., John Merrick—aka the Elephant Man). More generally, some cultures regularly and inappropriately disqualify some bodies from humanity (or from full humanity) based on arbitrary physical differences (sex, skin color, age, etc.). These kinds of distinctions are unethical and especially undesirable.
Then there is the possibility of creatures or inventions ‘crossing the boundary’ into humanity.[iv] How would we know if a lab rat grafted with human brain tissue started ‘acting like a human’? How would we know if a device that uses machine learning, vast data sets, multiple sensors, and complex networking is ‘acting like a human’? At what point do androids or biohybrid robots deserve human rights? Without answering these rather difficult questions, we can at least say that while ‘shape’ is a useful shortcut in most practical instances of recognizing what is or isn’t human, the essence of humanity is not shape.
Other characteristics unique to human animals include evolutionary accomplishments of the brain and voice. Human brains, for example, have proportionately larger cerebral cortexes than brains of other animals. The cerebral cortex is largely responsible for ‘high-level’ mental functions such as reason, thought, memory, and consciousness. However, measuring cerebral cortexes requires expensive instruments, surgery, or autopsy; and claims about brain matter, and the size, relative proportions, and convolutions of that matter are debatable. The ‘voice box’ is also a uniquely human feature, but it’s not the only noisemaker in the animal kingdom. Many non-human animals are capable of generating a wide range of sounds. To a visitor from outer space, chattering humans might sound much like chattering squirrels, and figuring out the difference in sound-making anatomy would again require instruments or surgery. We can agree that the cerebral cortex and voice box are evolutionary ‘milestones’ along the human path contributing to human uniqueness, but, by themselves, they fail the ‘easy to identify’ and ‘essential’ qualifications sought.
The most frequently resorted-to ‘markers’ of humanity are not just features of human bodies, but common practices. They fall into two categories: religious rituals—including prayer, the building of sacred spaces, burial rites, etc.[v]; and tool-use—including the building of homes and hearths, the use of fire, and the cooking of food. There is also the invention and use of speech. Each of these practices will be taken in turn.
Religious rituals seem entirely human, but there is some evidence of them in the animal kingdom. Elephants, for example, behave in ways that suggest mourning when one of their herd dies. Extended family members and other elephants come from long distances over a number of days to visit and touch, with their feet and trunks, the carcass of a dead member of their species. Chimpanzees, giraffes, domestic pets and other animals also exhibit behavior associated with mourning and some understanding of death.
Tool-using animals are more common (Bentley-Condit, V., & Smith, 2010). Crows use sticks to pry grubs from wood, dolphins use sponges to stir up prey, and sea otters use stones to crack open abalone shells. This tool use is rudimentary compared to what humans have created and use, but the difference is one of degree, not of kind. Regarding the building of homes, many animals make nests, dens, webs, hives, hills, and other living ‘complexes.’ As for food, some animals, such as horses and birds, ‘prepare’ their food by dipping it in water (to ease swallowing and maybe even to flavor it) (Equus, 2019; Goldman, J. G., 2017). And finally, while no other animal has ‘invented’ the use of fire on its own, chimpanzees fed cooked food by human scientists not only preferred the cooked food to uncooked food, but also learned (from those scientists) how to use technologies to cook food. The chimpanzees displayed an understanding of what cooking ‘is’ by delaying gratification—saving up raw vegetables for cooking, and waiting until the food was cooked before eating it (Warneken, F., & Rosati, A. G., 2015). In sum, while humans may be more sophisticated at religious and tool-using practices than other animals, these practices exist along a continuum in which other animals appear. For our visitor from outer space, a clear ‘hard-break’ between humans and other animals is not apparent.
And finally, there is speech. Speech (spoken language) should not be confused with communication, which is a broader category of activity common to all animals.[vi] Animals communicate by chirping, wagging their tails, emitting hormones or other chemicals, marking territory, growling, assuming aggressive or submissive postures, etc. But it’s quite another thing to communicate via complex combinations of signs and symbols. Speech involves the use of vocabulary (words, word-like sounds), tenses (past, present, future), conditionals (when, after), logical shifters (some, many, if this, then that; either this, or that, this but not that), and more. By means of speech, humans name and affect the world.
Speech has long been considered ‘essential’ to humanity. It seems to offer a clear line of demarcation between humans and other animals. But this simplicity is deceiving. Domestication and experimentation have brought dogs, chimpanzees, apes, and other primates to the fringes of human language.
It would be easy to fixate on debates over language-use in non-human animals, so let us accept the complexity of the territory and consider the following: a dog that ‘knows’ the word ‘treat’ would not be as good at assessing the message “I will give you a treat tomorrow if you stay calm when my friend comes to visit.” It is one thing to respond to a word as a ‘signal’ in the moment, another to ruminate about the past, or plan for the future. On the other hand, dogs have been shown to practice “functionally referential” communication (intentional communicative behavior) with humans, specifically by using expressions to ‘show’ humans the location of a desired object (Miklosi, et. al., 2000). Along the same lines, primates have been taught lexicons of words which they sometimes (situationally but not grammatically) combine (West, 2019). The ‘humane’ treatment of dogs, primates (and to this list we may add dolphins, orca, octopi, ‘sentience-exhibiting’ AI, etc.) is of increased concern to humans when creatures are ‘brought into language’ by humans or when entities exhibit aspects of human language.
Even if we allow for the speech-related achievements of a few ‘advanced’ animals, other problems with using ‘speech’ as the best ‘dividing-line’ between humans and other animals exist. First, speech can be hard to identify when languages are foreign or new. What looks or sounds like words to a native speaker can seem like noise to the outsider, and languages may go entirely unnoticed if ‘spoken’ in unfamiliar or ‘hidden’ codes. Second, everyday uses of the word ‘language’ confuse people about what ‘counts’ as language. For instance, we use the expression ‘body language,’ but meanings of non-verbal gestures, postures, and comportments are often ambiguous; non-verbal activities can communicate emotional states, but lack the symbolism and precision characteristic of spoken language. Along different lines, a popular book outlining five ‘love languages’ (Chapman, 2015) complicates the meaning of ‘language,’ by taking each love language’s ‘vocabulary’ outside linguistic territories and into acts of service, receiving gifts, and spending quality time, and other kinds of activities and cultural practices. And finally, language is not the best dividing line between humans and other animals because languages are complex. Like algebra or calculus, languages are an accomplishment. If you see an animal doing algebra or calculus, you can be fairly certain the animal is a human (so calculus may be considered ‘essential’ to humanity). But not every human can do algebra or calculus. Likewise, speech is a skill that takes time to learn, and languages vary in complexity. There may be some more basic, ‘essential’ faculty related to language that we have not yet quite discerned.
In sum, many ‘essences’ of humans are suggested by things humans are, have, and do, but each popular option considered above has weaknesses. ‘How something looks’ is not a good way to decide what is human, and neither is measuring an animal’s cerebral cortex or finding its ‘voice box.’ Some animals behave in ways akin to religious rituals, and others use tools, build shelters, and have been taught how to use fire to cook food. Of all the possibilities considered, speech is clearly ‘essential’ to humanity, but it is controversial, difficult to identify, and complex. Still, speech gets us closest to our goal. What we seek appears to be related to speech, but is a ‘simpler’ (more basic) accomplishment.
Image-Making and Beholding
Imagine our explorers from another planet observing a small group of humans through the window of their living room. A mother sits and watches her two very young children, who are drawing with crayons on pieces of paper. The younger child drags his crayons across his paper making random lines of various colors. He takes pleasure in using his hands to move the crayons around, and in making marks on the paper (and occasionally the floor). He appears to be doing the same thing as his sister, but his sister does something different—something the mother has not seen her do before. This child picks out a crayon, scribbles a blob on the page, and puts a smaller blob on top. Then she picks out another crayon to make a few intersecting lines across the top blob. The mother sees this configuration of lines and shapes and gets excited. “Oh, it’s Ozzie! You drew Ozzie! . . . there are his whiskers! . . . That’s so great! I will put it up on the refrigerator.”[vii]
Why get so excited about the daughter’s scribbling? Why is this considered an accomplishment worthy of celebration and display?[viii] Here, our explorers find what they have been seeking.
The drawing of Ozzie performs no structural function, was not accidental, and suggests the likeness of a form—in this case a living form of a cat encountered by the child. Like the Lascaux cave paintings, aboriginal petroglyphs, totem poles, Renaissance figure paintings, architectural drawings, social media photos, and myriad other images, the daughter’s drawing provides clear and simple evidence of a human. Her image did not need to be aesthetically pleasing, perfect, or even ‘good,’ it just needed to be recognizable with one thing (the image) ‘standing for’ another thing (the cat). In this case, the shape and whiskers were enough to achieve ‘representation.’ Such a product confirms the special quality of its maker—her potential as a speaker, a thinker, and an inventor, and her status as a ‘symbolical’ being.
‘Image-making’ recommends itself as the ‘essence’ of humanity for many reasons. First, it is more basic than speech; we are more likely to agree on what a picture is than on what a word is, and we can share images without having to share a language. Second, no other animal creates biologically useless representations. Other animals do things and make things in order to satisfy immediate practical needs (nourishment, reproduction, shelter etc.), but humans animals make images (which can’t satisfy hunger, pass on genes, or protect their creators from the elements). Humans can care about and try to shape ’images’ of ourselves or in the eyes of others; we can build images in our minds about other people; we can become wrapped up in an image of the future by means of ideology; we can imagine identities of nation, gender, ethnicity and race; we can accept, make, prohibit, or deny images of a deity or deities, and much more. Other animals don’t complicate their lives in this way. In sum, image-making animals are those animals that participate in the making of biologically useless objects, or have goals beyond immediate practicalities, or can serve their goals in different kinds of non-practical (symbolic) ways. Finally, understanding what images are and how they function can bring us closer to understanding other more complex human practices such as speech, art, and reason.
The Essence of Images: Eight Properties
If the essence of humanity is image-making, then we need to understand images—what images are and how the qualities of images shape their human makers and users. Because images are essential to the ‘symbolical’ animal, exploring the ‘essence’ of images is key to not only understanding what images are and what they entail, but also to what humans are and who we might become.
To be clear at the outset, “image,” as referred to in this discussion, are physical representations of some object or imagined object or ‘sight.’[ix] Other things considered ‘images’ may be related, but do not fit this definition. For example, a reflection of a face in a pool of water looks like that face, but it is not a physical representation of the face; it is a mirror-image of the face relying on the presence of the face to exist. Films (movies/cinema) and videos are image-filled media, but also involve motion and time (Jurgenson, 2019). They have their own special properties (rhythm, pacing, narration, etc.), so the list of qualities offered here only deals with part of their complexity. Likewise, a person’s ‘image,’—their reputation, persona or ‘brand,’ is related to but goes beyond the scope of the subject at hand. A person’s ‘image’ includes not just still images of themselves posted on social media sites, but shared moving images, memories, personality traits that people might notice and remember independent of visual qualities, etc. That said, much can be learned about broader and more complex image-related territories by starting with the basics.
The ‘essence’ of images can be spelled out in eight properties. Each property is essential to any image, and any image has these eight properties.
1. The first property of an image is “likeness.” An image must look enough like the thing it’s an image of so as to be recognizable. The image of Ozzie the family cat looked like the cat to the mother (even if just barely); the image was recognizable as the cat.
2. The second quality of an image is “intent.” An image must be an intentional likeness, not an accidental one. A cloud that looks like a bunny is not, as we define it here, an image, because there was no human intention behind the creation of the cloud. If, however, a sky-writer airplane made an outline of a bunny in the sky out of its exhaust—that would be an image. If the child in our example had accidentally created a likeness of Ozzie the cat, her drawing would not have qualified as an image. There needs to be ‘intent to make an image’ for an image to be an image and not simply a serendipitous or chance likeness.
3. The third quality of an image is “incompleteness.” A complete replica of some ‘thing’ in the world would be a duplicate of that thing, not an image. Because of this, a clone is not an image. Neither is a twin an image (one twin is not an image of the other twin). Something must be missing from some original ‘imaged’ object(s) so that the image is “standing for” but not “exactly replicating” the original thing(s). An exact replica of a table is not an image of that first table—it’s another table.
Taxidermy is a borderline case because these ‘stuffed’ animals are partly made out of the animal itself (the skin/fur), but are not fully the animal (there is an armature, stuffing, glass eyes, only one pose offered, no internal organs, etc.). Because a taxidermied animal differs from the ‘original’ live animal, and because it is presented to viewers as an image (and not intended to fool them into thinking it is a ‘live’ animal), we could say a taxidermied animal is an image of the once-live animal.[x]
Viewers also need to be able to tell that the image is not real, otherwise it’s acting as a fake rather than an image of a thing. A wax statue of a person may look very much like that person, but it’s still an image. If a wax statue of a person is put somewhere to ‘fake us out’ and we are fooled into believing a real person is there, the statue is not an image, it’s a fake. An image is by definition partial and even superficial in some way—it ‘means’ to be so and we and it are ‘ok’ with that. Knowing that an image is not the ‘thing’ that is being ‘imaged’ is not a problem—it is what we expect of an image.
4. The fourth quality of an image is closely related to the third. Incompleteness assumes “degrees of freedom.” This means there is some choice about what to include and what not to include when making an image of some sight or object. Many images turn three-dimensional things into two-dimensional images. This leads to choices about perspective and framing. Should we draw it from the front or from the side? Should we take the picture from slightly below or slightly above? What parts should go in the frame? Three-dimensional images require choices as well. Sculptures, architectural models, relief maps, dolls, and other volume-having images, for instance, omit aspects of ‘things imaged’ such as size and detail, and poses can be chosen from many options.
Whether images are created by hand, or with the use of technologies, whatever is left out of them says something about whatever is included. Often, ‘representative’ (relevant, significant) elements are featured. Elvis as a wax dummy in a rhinestone studded leather jumpsuit in full lunge with microphone to the lips—yes. Elvis as a wax dummy with wrinkles and messy hair in polyester slacks sitting on a couch—no.
The less complete the image, the more the focus on featured elements, and thus the more an essential ‘likeness.’ On face emojis we get eyes, noses and mouths, but seldom cheekbones, eyelashes, or wrinkles. This ‘idealization’ is not so much about beauty, as about economy, and makes it even clearer that what we are dealing with is an image and not ‘the real thing.’
In the drawing of Ozzie, the whisker-lines and ‘blobs’ sufficed as a likeness to the child (and mother)—even without eyes, ears, or a tail. The whiskers and stature of the cat ‘mattered’ to the child and the ‘proper’ combination of these elements was recognizable as the family cat to the mother. Choices about what to include or leave out of images reflect priorities. In photographs, for example, we see what matters to people when they have freckles erased from a close-up senior portrait, use a full-body shot of their child—in his football uniform—as a screen saver, or crop an ‘ex’ out of an image posted to social media.
5. The fifth quality of an image “positive difference” is also linked to the prior two qualities. “Positive difference,” is a ‘degree of freedom’ that goes beyond just deciding what to include or leave out from some imaged sight or object; an image can differ in positive (additive) ways from an imaged thing. Colors can be changed, parts expanded or shrunk, features added, accentuated, or embellished, and more. These positive changes can happen by accident, they can be made because an image-maker lacks skill, or they can be made on purpose.
Positive difference opens the door to invention. For example, we can add wings and fiery ‘plumes’ of breath to a drawing of a lizard and get a dragon; we can decorate a three-dimensional representation of a room with imaginary furniture; we can commission a giant-headed bug-eyed (but still oddly ‘accurate’) caricature of ourselves at a local fair; and we can fashion a taxidermy mount of a jackalope by combining a rabbit’s head with a deer’s antlers.
6. The sixth quality of an image is “visual shape.” This seems obvious, but deserves mention. When we make an image of some object, ‘shape’ (form) is essential. We may not exactly replicate colors or the size or the lighting of the thing (due to skill or accident), or we may intentionally vary them, but some resemblance in shape is needed—even in cases of caricature and pastiche. Color, size, lighting, and other aspects of images are not essential to representation. If they were, we wouldn’t recognize black and white pencil drawings of places derived from colorful scenes, or people displayed in inches-high digital images a phone. If an image, say of a person, is to be recognized, what we need is some accuracy in the shape and proportions of the face, body parts or body. When trying to get a portrait ‘just right,’ facial features need to be the right shape, and the right distance from each other, or the image doesn’t look like the model. A few millimeters off, and it’s time to start over. But there is no problem if the image is rendered all in purple, or made out of dots, or super-huge. It’s the form that matters.
7. The seventh quality of an image is “inactivity.” An image is a frozen moment in time; it is not, itself, dynamic. An image may signify motion by using symbols—as in comic books with lines or ‘clouds’ of dust, or by appealing to the logic or expectations of viewers, but images, themselves, are at rest. Subjects or objects of an image may go on in time, but images do not. In film, images do not move, although they appear to do so on the screen when projected. With video, pixels of light are projected and give the impression of movement, but that’s the result of a combination of light, color, and time. Images projected, painted, drawn, or otherwise wrought don’t show how long they took to get made or what actions creators went through to create them. When we look at an image ‘as an image,’ we look at it as a finished, static object; we don’t think about what it took to get the image to become itself, we think about what the image represents.
Neither is an image a ‘trace.’ A trace, such as a footprint or a fossil, is a cause-related ‘record’ of something that happened. A foot causes a footprint; a skeleton’s impression causes a fossil. But an image is not a ‘record’ of an event. An image is not ‘about’ the cause of its own becoming; it’s a timeless thing intentionally made to represent. We may choose to attend to how an image ‘comes to be,’ but images themselves are not concerned with those histories.
A borderline example of a case between ‘image’ and ‘trace’ is a person dipping their feet in paint and decorating a curtain with footprints. If the intent is to make ‘foot images’ for others to recognize, the technique might qualify as image-making (along the same lines as taxidermy, which is also a hybrid of image and actual body). However, a clearer example of image-making would be if the curtain-maker used a foot-shaped stamp, or painted foot-images on the curtain by hand with a paintbrush. With such techniques, the ‘how’ does not supersede the ‘what.’
8. The eighth quality of an image is its participation in a three-part (triadic) relation between object, image, and eidos. More specifically, the three-part relation is between any physical object or scene being imaged (a cat, a landscape), the image of that object or scene made of some medium/media (on paper, a screen, a wall, a canvas; out of pixels, charcoal, paint, etc.) and the ‘eidos’—the image independent of any material (the image in the mind/minds).
Using the ‘Ozzie the cat’ image as an example, the cat being drawn (the animal itself) is the object. The drawing of the cat the child made with crayons on paper—the one destined for the refrigerator—is the image. If the mother took a picture of that image and sent it around to her friends on social media, that also would be ‘the image’ (this time manifest in pixels on various electronic platforms). If the drawing on the refrigerator were to be destroyed or removed, but the mother, child, or anyone else retained a memory of that image in their mind, that memory would be the eidos. Or, if anyone were to refer to (think of, talk about) the image of Ozzy (the one on the refrigerator) independent of its material form, that would be the eidos. The individual or shared memory of an image is the eidos of that image. Not surprisingly, the word ‘eidos’ is related to the words ‘idea,’ ‘ideal,’ ‘and ideology.’
This three-part relationship brings along with it three implications: there can be many images of the same object, there can be many copies of the same image, and an image can stand for many objects. There can be many images of the same object because images of objects can be depicted from a variety of angles, be all sorts of colors and sizes, include (and exclude) all kinds of details, and be altered and embellished in myriad ways. For example, an image of Ozzie the cat drawn from directly above the cat would be another image of Ozzie.
There can be many copies of the same image because even if the ‘physical carrier’ of an image differs, we still recognize form. For example, a painting of sunflowers hanging in an art gallery is the same image as a picture of that painting in an art history book—or a poster on a college student’s living room wall—even though one is made out of paint on canvas and the others are ink on paper. The image of the sunflowers has to do with form, not with the particular media by which it appears. In other words, the image of the sunflowers is not just one painting, it’s an eidos capable of appearing as ‘the same image’ in many different places, via different media, in different sizes, colors, etc. Any interest in ‘the original’ painting (as compared to copies of it) highlights the image’s means of creation (techniques, tools, and materials; purchase history; value as a collectable or investment, etc.), rather than its status ‘as image.’
And finally, an image can stand for many objects. This is because we can use an image to refer to an ‘essence,’ or a ‘class’ or ‘category’ of things. One common example is the image of a printer on a computer screen; it ‘stands for’ all printers. Another example is the image of a cow on the See ‘N Say children’s toy. This cow ‘stands for’ all cows. The same is true for the ‘dog’ image in the ‘don’t let your dog do its business on my lawn’ sign. That dog image is supposed to ‘stand for’ all dogs regardless of size, color, or breed.
Image-Beholding and Image-Making: Implications
Image-beholders are potentially image-makers, and vice versa. The ability to perceive likeness involves being able to separate form from matter (to separate the eidos from the ‘hard copy’). This requires abstraction. Dogs may direct human attention, and primates signal via borrowed lexicons; but humans are a different order of being operating on a different plane of action. Language, cognition, memory, motor skills, and imagination enable the uniquely human accomplishment of image-creation.
A few comments about AI are worth making here. AI are not image-makers (we would not call cameras image-makers either). So far, the intention to make images with AI is human, and only those images that ‘look like something’ to a human, qualify as ‘likenesses.’ To put it another way, prompting an AI to generate an image of a ‘strange new animal,’ results in an image of a ‘strange new animal’ only if human viewers acknowledge the visual phenomena generated as a ‘strange new animal.’ Moreover, adjustment of prompt parameters can result in the generation of “eye gari” (Anton, 2023). Eye-gari are unfamiliar but often stimulating visual outputs—a loose visual equivalent to what in text-generation has been called ‘hallucination.’ It/They are more like bizarre wallpaper or visual ‘junk’ than images.
When we call AI-generated visuals ‘images,’ we are using the term somewhat metaphorically. Complexities in ‘likeness’ and ‘intent,’ mean computer-generated visuals do not always ‘count’ as images—as we have defined them here. A better way to refer to what we now call ‘AI images’ (those that are clearly likenesses, as well as the ‘strange new animal’ type, and the ‘eye-gari’ type) might be ‘collaborative visuals’ or ‘visual collaborations.’’[xi] Such phraseology acknowledges that AI visuals draw from a plethora of data, and that machines are not intelligent in the same way humans are.[xii] If, however, we were to find an AI burning images of the view of the room opposite the processor’s video camera onto the floor, or projecting laser light-show self-portraits of some of its inside mechanisms onto the wall, or improvising caricatures of lab personnel as impromptu screen-savers, then we really should take a second look at what’s going on.
Image-creation has a rudimentary truth-aspect to it: an image can be more or less ‘true’ to the original object. It is because we can represent that we can lie. This kind of lying is not simply lack of skill. Nor is it deceptive camouflage or behaviors selected for by evolution and evident in other animals. This kind of lying is intentionally altering how we re-present what we see and what we know.
The truth-aspect of images is illustrated in a popular but debunked myth dating back at least 300 years concerning the word ‘sincere.’ In one version, Greek (or Spanish) sculptors of marble would sometimes accidentally chip off too much stone from their work. Rather than start over, the sculptors would fill in the chipped part of the stone with some wax. In Latin, ‘sine’ means without, and ‘cera’ means wax, so ‘sincere’ was (incorrectly) taken to be derived from a combination of the two: ‘without wax.’ To be sincere, was to display ‘honest’ or ‘true’ (wax-free) craftsmanship and not to visually ‘lie’ about the quality of the carving. Apparently, false etymology is no match for a good myth about images, lies, and truth.
And finally, image-making is similar to, and related to, ‘naming.’ Just as an image can stand for a thing, so too can a sound or set of sounds stand for a thing. This sign-process is the basis of language. In a well-known Biblical story (Genesis 12:19-12:20), one of the first things the ‘first human’ did was name all the animals. Naming is an act of dominion over what is named. It sets humans apart from ‘the named.’[xiii]
Because humans are symbolic beings, we also may have been destined to invent writing. It is impressive that a symbol or letter in an alphabet can ‘stand for’ that letter in any font (each marking has an eidos). Thus, literate people can read a range of handwriting styles and fonts. It is also an accomplishment to create a shape such as “a,” and have that “a-shape” represent a sound ‘aaaayy’ or ‘aaaahhh’ when seen in the context of other shapes near it (like this text is getting you to do by reading the words ‘pay’ and ‘park’). When a letter can stand for a fraction of a sound, when sets of letters can represent the sounds of spoken words, and when a set of letters written on a page can ‘stand for’ an eidos, idea, category, ideal, or class of things, ever more complex representations follow.
Humans can design calendars and plan their future. We can represent relative amounts of products using wedges in a pie chart. We can make statements and use logic to debate claims. We can use irony and sarcasm, expecting hearers to understand more than ‘just what the words say.’ We can create computing machines with 0’s and 1’s to represent values. And we can delegate the representation of value by algorithms to algorithms generated by the machines we have made. Using letters, words, sentences, signs, symbols, and technologies, humans can tell the truth, lie, or delegate tasks for a whole host of non-survival-related reasons: to assuage loneliness, to bolster our own egos, to avoid responsibility, to take pleasure in others’ suffering, to build someone’s self-esteem, to get a laugh, and more.
Humans often ‘act like animals,’ but other animals do not, as far as we know, make and break laws of their own design, practice religions, kill for ideas, or record and refer to their individual and collective pasts. Nor do other animals make public promises of fidelity, tell utopian or dystopian stories, fantasize about future love-interests, or spontaneously create, appreciate, and critique their own artwork. Because of image-making, humans accomplish the symbolic making-over-again of the world, via perception, imagination, memory, and discourse. Image-making opened the door.
References
Anton, C. (2023). A.EYE CANDY: A museum of imaginary robots and other digital delights. Institute of General Semantics.
Bentley-Condit, V., & Smith. (2010). Animal tool use: current definitions and an updated comprehensive catalog. Behaviour. https://psycnet.apa.org/record/2011-15274-000
Berger, J. (2008). Ways of seeing. Penguin Classics.
Bédard, P. (2014). Genesis 2:19-20 – Did Adam Have Enough Time to Name All the Animals? Reformed Perspective, 4. https://www.christianstudylibrary.org/article/genesis-219-20-did-adam-have-enough-time-name-all-animals
Chapman, G. D. (2015). The 5 love languages: The secret to love that lasts. Chicago, IL: Northfield Publishing
Brahambhatt, R. (2023, December 15). If AI is making the Turing test obsolete, what might be better? Ars Technica. https://arstechnica.com/ai/2023/12/do-ai-improvements-call-for-something-better-than-the-turing-test/
Equus Magazine. (2019, July 25). Does Your Horse Soak His Supper? When horses soak their hay or grain, there might be an underlying medical reason. Find out why. https://equusmagazine.com/horse-care/eqsoak480-10242/
Goldman, J. G. (2017, September 29). Australian Bird Dips Its Dinner. Scientific American. https://www.scientificamerican.com/podcast/episode/australian-bird-dips-its-dinner/
Jonas, H. (1966). The Phenomenon of Life: Toward A Philosophical Biology, New York: Harper & Row Press.
Jurgenson, N. (2019). The Social Photo: On Photography and Social Media. Verso Books.
Lanier, J. (2023, April 20). There Is No AI The New Yorker.
Marsh, J. (Director). (2011). Project Nim [Documentary]. Red Box Films.
Miklosi, A., Polgardi, R., Topal, J. & Csanyi, V. (2000). Intentional behaviour in dog-human communication: an experimental analysis of “showing” behaviour in the dog. Animal Cognition, 3, 159-166.
Thompson, R. H., Cotnoir-Bichelman, N. M., McKerchar, P. M., Tate, T. L., & Dancho, K. A. (2007). Enhancing early communication through infant sign training. Journal of Applied Behavior Analysis, 40(1), 15–23. https://doi.org/10.1901/jaba.2007.23-06
Warneken, F., & Rosati, A. G. (2015). Cognitive capacities for cooking in chimpanzees. Proceedings of the Royal Society B: Biological Sciences, 282(1809), 20150229. https://doi.org/10.1098/rspb.2015.0229
West, E. (2019, December 6). Can primates learn signs and acquire language? British Deaf News. https://www.britishdeafnews.co.uk/primates-sign-language/
[i] Images also may be made in the mind, and in various media, by people who lack eyesight. Touch (particularly the use of hands), movement, and memory make this possible. A blind person may, for example, ‘feel out’ an object, create a mental image of that object, and accurately describe or re-create that object (via sculptural media or tactile-guided marking implements). Blind people also learn concepts such as perspective and shading despite their lack of experience with those phenomena. This is because vision is not simply an ocular matter; it is mental and intellectual. As such, humans can have ‘vision’ even when they lack the faculty of sight.
[ii] Brahambhatt (2023) argues the Turing Test is not a meaningful measure of human-ness (for different reasons than those presented here). Brahmbhatt faults the Turing Test for its opacity regarding ‘what the systems themselves’ understand or can do. Does it have ‘inner-monologues’? Can it conduct long-term planning over abstract horizons? This is a good criticism of the test, but focuses on rather sophisticated ‘indicators’ of humanity rather than the ‘essence’ of humanity.
[iii] For a humorous puppet-sketch of Sesame Street ‘aliens’ using human characteristics to mistakenly identify a clock as a human, see: https://www.youtube.com/watch?v=dhKhGyUnc2E
[iv] The documentary “Project Nim” (Marsh, 2011) interrogates such boundary issues. It tells the story of an experiment where a chimp was raised as ‘one of the family’ by various human surrogates, and taught sign language. After becoming aggressive, the chimp’s education was ended, and he was dropped off at a facility. The regrets of participants in ‘the experiment’ acknowledge the cruelty of bringing a creature into human relationship and language, and then leaving it to live alone, without speech.
[v] This inquiry is a ‘philosophical anthropology’ rather than a religious inquiry. In other words, the ‘essence’ of humanity is sought out in humans themselves, individually and collectively. Thus, ‘religious practice’ may be a marker of humanity, but a particular religion may not. Sub-groups of humans whose religion says they differ from other animals because they ‘have a soul,’ ‘have fallen from grace,’ ‘are the chosen people,’ etc., locate humans in larger human-divine stories. Such stories complicate or obviate the discovery of a human ‘essence’ in humans themselves. Because particular religious understandings of human ‘specialness’ are a matter of belief rather than reason, they are not applicable to the more generalized intellectual inquiry conducted here.
[vi] It is tempting to think that animals understand us when we talk (especially our pets), but animals don’t speak—they communicate. Animals can sense people’s emotion, act in response to the ways people are acting, communicate states of mind (fear, contentment), and they may even get some rudimentary ideas across, but animals and pets don’t ‘talk’ to people or with people.
[vii] The mother may include the younger child’s drawing on the refrigerator as well—so that he does not feel left out, but the reason for posting the older child’s drawing is qualitatively different.
[viii] A related achievement includes when a baby learns how to ‘sign’ a word (Thompson et al, 2007). At around eight months of age, babies can be taught to sign basic words like ‘milk,’ ‘done,’ and ‘up.’ Such learning is impressive and gratifying to parents, and the ‘head start’ in representational practices gives these children a slight developmental advantage over their peers. If a baby is not taught sign language, excitement is typically expressed when it speaks its first word, which is also a symbolic accomplishment.
[ix] For a similar definition of ‘image,’ and a sophisticated discussion of ‘ways of seeing,’ and the impact of photography, sexism, and capitalism on art, see Berger, J. (2008).
[x] Embalming presents a similar case.
[xi] This verbiage borrows from Jaron Lanier (2023) who points out how the label “Artificial Intelligence” is itself a misnomer. Because data used by large language models is derived from humans, he offers ‘social collaboration’ as an alternative label for their activity and output. Such terminology recognizes the power of the technology while keeping that technology ‘in its proper place’ relative to humans.
[xii] For example, an A.I. asked to find the highest prime number will continue to work on that problem ad infinitum, but an intelligent person working on that same math problem will, at some point, give up the project, or at least stop and have dinner.
[xiii] For a fascinating literalist explanation of how Adam had enough time to accomplish naming all the animals in a 24-hour period, see Bédard, P. (2014).
