Here I propose an explanation for why we can know that other people are conscious, even if we cannot formally “prove” it.

The argument is simple: if you are conscious, and you know that other people have similar behaviors and biology (especially brains) to you, then surely they must have similar consciousness to you. The alternatives are too implausible to take seriously.

Likewise, we can know that animals have some consciousness: the more similar that animals are to us, biologically and behaviorally, the more-similar their consciousness ought to be as well.

I am not proposing a solution to “the hard problem” of consciousness, meaning I do not offer an objective test of whether any “AI” or alien is conscious, nor do I offer an explicit definition of consciousness. My arguments here do not disprove the possibility of “AI” consciousness. However, they highlight the need to consider mechanisms (biology and/or algorithms) and not just behaviors when attempting to judge consciousness.

The problem

You, the person reading this, are conscious. You have some feelings, experiences, and awarenesses of each moment of your waking life. How do you know you’re not the only one?

Your behaviors relate to your inner life, for example, acting differently based on how you feel, and telling other people about it. And, for your whole life, you’ve observed other people talking about their inner lives, and behaving in ways that seem appropriate. You act one way when mad, and another when relaxed. You can recognize the relationship between others’ behaviors and their stated inner lives, based partly on your own inner experience.

But perhaps it is an illusion, or a lie. Perhaps everyone else around you is basically a machine, performing the behaviors as though they are conscious people like you, but they are not. Perhaps they are all “philosophical zombies.”

Two sources of information

We have two sources of information about other people: their behaviors, and what we know of their biology.

You and I are both human, part of the same species and the same evolution. We share, along with the rest of humanity, the same kinds of brains, hearts, nervous systems, and so on. We can eat the same kinds of food, we need the same kind of exercise and sleep and so on. Of course we mostly focus on the differences between us (say, food allergies, different tastes or preferences, neurodivergence), but it’s not like some of us can hibernate for six months at a time and others can metabolize gravel.

Our brains mostly have the same parts and the same functional elements at different scales. So it makes sense to believe that our mental states are similar too—we share the same broad palette of emotions and mental representations, even though they are expressed differently in different people.

Hence, these facts together give you every reason to believe that other people are conscious:

1. You are conscious
2. Other people are biologically very similar to you
3. Other people talk and behave in ways consistent with your experience of how your behaviors relate to consciousness

Indeed, we naturally take others’ behavior as cues to their mental state, and might question the mental wellbeing of someone who’s behavior seems inconsistent with a sensible mental state.

This, here, is sufficient evidence to live your life without seriously doubting other peoples’ consciousness. Doubting other peoples’ consciousness is purely an intellectual excerise, not something worth taking seriously in how we understand the real world.

The alternatives: paranoid conspiracy theory or irrelevance of consciousness

After all, what is the alternative?

It could be that you alone are conscious, and everyone else is merely performing behaviors of consciousness, despite sharing the same biology. This seems to me a paranoid conspiracy theory, like imagining that we live in a simulation or some other constructed reality. No one can disprove such fantasies, but there’s no genuine reason to believe them either, and, even if true, they’re irrelevant to any decision we’d make in our lives.

It could also be that most people are conscious, but some rare individuals lack consciousness while still functioning in society. There are lots of examples of rare individuals that lack other perceptual or mental attributes, such as colorblindness, aphantasia (lack of mental imagery), stereoblindness (lack of stereo vision), to sociopathy (lacking compassion for other people). Some of these are hard to detect, but they are detectable (even though sociopaths might perform compassion). If there were philosophical zombies amongst us, you might think that we would have already discovered it somehow.

If there were no genuine differences in behavior between people with consciousness, and philosophical zombies (who claim to experience consciousness but do not), then that seems to imply that consciousness is irrelevant for normal life. That is, consciousness has no noticeable impact on a person’s behavior or ability to function; it just randomly appears in some individuals but not others.

Each of these alternatives seem so implausible as to be unworthy of serious consideration.

The only sensible possibility is that everyone who is awake is conscious.

The varieties of consciousness: altered states and animals

How can we know that animals are consciousness?

Note that it’s unlikely that animal consciousness is identical to human consciousness. Perhaps animal consciousness is like human consciousness, but with some differences. Perhaps, day-to-day, my dog experience a vivid alertness that’s a thousand times more than my own peak (like when playing a vigorous sport or video game), but lacking any mental foresight (like when I’m drunk and groggy and barely able to function). There’s reason to believe that octopuses are conscious in very different ways from humans.

Regardless, the same reasoning as above can tell us that mammals are conscious. Consider chimpanzees and dogs, each of which are social animals and each is related to us biologically, including in terms of brain structure. They each exhibit behaviors that seem to indicate consciousness, such as social expressions of inner states (happiness or anger, etc.), and other sorts of behaviors (attention, distraction, etc.). We share similar internal organs, similar brains as monkeys and dogs, and some of the same behaviors. We share evolutionary lineages with them as well; they are evolutionary cousins. Hence, they ought to share some sort of consciousness as well, by the same reasoning as above.

And, the alternatives to animal consciousness seem highly implausible. If no animals are conscious, then consciousness emerged solely in humans. But nearly all of our faculties evolved in earlier creatures; monkeys have many of the same faculties of intelligence as us, as do their ancestors, and their ancestors. Consciousness seems most closely tied to mood, emotion, alertness, awareness, social behaviors—all attributes shared by many animals—and not to the few things that seem to make us biologically unique as humans, such as the ability for complex spoken languages.

It seems likely that more-sophisticated intelligence goes along with more-sophisticated consciousness. Dogs have much less-sophisticated intelligence than humans, and so their consciousness would be much less-sophisticated as well, but still present.

Machine consciousness and alien consciousness

We cannot apply the above arguments for machine intelligence, because machines do not share biology with us. Current “AI” algorithms do not at all operate like human brains. There are ways in which they do operate similarly, e.g., convolutional network models of the visual cortex, dopamine as modeled well by reinforcement learning, but these are small pieces of the overall functioning.

This does not disprove “AI” consciousness, but it shows that establishing “AI” consciousness is far, far, more difficult than human consciousness. Which seems obvious, but apparently is not to some people.

How would we demonstrate or prove machine consciousness? Or, if aliens came down from space, how would we determine if they are conscious? I don’t think anyone has a good answer. However, I do not believe that we can judge consciousness simply by looking at behaviors, like performing Turing Tests. We’ve already seen over and over that people can be fooled by simple rule-following chatbots like ELIZA. We can build convincing philosophical zombies.

How do we know that chatbots like ELIZA are not conscious? We read the paper, or look at the code to see how it works. We can know enough about how LLMs work that to know that they, too, are simple text processing engines that, like ELIZA, literally do nothing more than take text strings as input and generate text strings as output.

It might be that computer algorithms could be described as “conscious” if they have any mental state representation at all, like state variables representing “mood.” But such representations are so different from human consciousness that calling them “consciousness” in public discussion seems misleading and extremely irresponsible (a topic I intend to exapnd on in a future blog post).

All of these arguments indicate that judging consciousness, we have to consider the underlying mechanisms, that is, the algorithms and/or the biology, and not just inputs and outputs. Similarity to our own biology would be a positive clue; algorithms that generate print statements as a functions of inputs would be a negative clue.

Summary

I have argued that you can know that other people and animals are conscious, because you are conscious, and because you share important behaviors, biology, and evolution with other people and with animals. This is not a solution to the “hard problem,” and it is not rock-solid scientific proof. But you can be confident in it enough to assume that it’s true—the alternative involves paranoid conspiracy theories or philosophical zombies well-hidden among us. On the other hand, there’s no good reason to believe that algorithms, which have totally different development, substrate, and behaviors from us share this one hidden attribute.

Specifically, we often describe dog behaviors in terms of knowledge. “She knows that if she sits, she gets a treat.” “He knows he’s not supposed to be on the couch.” “To deal with separation anxiety, she has to know you’ll come back.”

In this blog post, I’ll explain why I believe that dogs do not know things, not in the way that we do. If dogs had knowledge the way we do, they ought to be able to perform some kind of conceptual reasoning or planning. But they do not exhibit any behaviors that require conceptual knowledge. Importantly, dogs do not know about the future, and so they cannot plan for the future either. I’ll talk about what kinds of intelligence dogs do have, and how these can be mistaken for knowledge.

Much of the time, talking about dogs as if they have knowledge or planning ability is simply a convenient shorthand. But misunderstanding dog psychology can lead to bad experiences and poor relationships with them. Conversely, better understanding of dogs can make us more empathetic toward them, strengthening our bonds.

I am not an expert; this is just my interpretation based on my experiences and the books I’ve read on dog training. (For a cognitive psychology take, see this thread responding to this post.)

This post is decorated with photos of dogs that I’ve worked with at the San Francisco SPCA, and my own rescue from the shelter. (The photos on my phone are like 80% dogs.)

Dog intelligence

Here are some kinds of intelligence we see in dogs, and the behaviors that go along with them.

Dogs learn from reinforcement. If you have ever worked on dog training, then you’re familiar with their ability to learn from reinforcement (classical conditioning and operant conditioning). If they do something and get a reward (anything positive that they like), they will tend to do that thing more often. So we give positive reinforcement for desired behaviors.

Suppose a dog has been trained to sit when you say “sit,” and, sometimes you do this, you give them a treat. This is a learned behavior: saying the command “sit” triggers the dog’s “sit” behavior. When the training is solid enough, the dog does the behavior automatically.

Unrewarded behaviors tend to decrease, so we try not to reward unwanted behaviors (this can be difficult, like when your dog jumps on you for attention). Negative reinforcement also reduces unwanted behaviors, but it creates other problems, so it should be avoided.

Dogs form associations. Some associations are innate (my dog loves delicious-smelling things), while other associations are learned (she loves the dog park, so she pulls toward it when we’re near). Some associations can be more specific: my dog goes to the food bowl when she’s hungry. Some spaces are safe: a dog that is scared may try to retreat to a space where they feel safer.

Negative associations can be formed with positive things. When I first started trying to train my dog to relax when I left her home alone, I tried giving her delicious dental chews whenever I left. But instead she learned to associate the chew with her isolation anxiety. Now when I give her a chew and stay home, the poor girl alternates between eagerly running away to chew on the toy, whimpering pathetically, and running back to check that I haven’t left her. (I now believe that she had some abandonment trauma from her previous owner, given what I know.)

Dogs have some spatial awareness; one might call it a specialized spatial knowledge: my dog knows her way around our apartment and many other places. When we visit a friend’s house, she’s visibly excited to be on the doorstep and eager to get inside, seemingly aware (or expecting) that friends are inside.

Dogs form expectations and/or predictions. If you give familiar commands to a dog with a treat in your hand, but repeatedly fail to give the treat, they will sometimes bark in frustration. If I raise my arm to throw a ball, my dog will get ready to chase after it (and occasionally she starts running before I’ve even thrown it). When my doorbell buzzes, my dog runs to the door, excited to meet and/or bark at whomever is out there. She seems to have has learned that doorbell buzzing means that we’re about to see new people in the hall or outside. (The door to my apartment is far enough from the building entrance that I don’t think she can directly sense the people outside.)

Dogs have internal state, related to things like emotions, hunger, sleepiness, and circadian rhythms. My friends’ dog gets a delicious dental chew treat at 8pm each evening, and so, right around 7:55pm, she begins to get very eager, excited, and happy. She does not “know” that it is 8pm, but something in her biological clock triggers certain expectations and behaviors. (When I had this dog over at my house for a few nights, I did not give her the chews, and I never saw her get excited at 7:55pm, so the timing is not the only cue for her; she has to be with her family.)

Dogs make decisions. Dog decision-making often seems to be about being selecting from among desires or avoiding things. Dogs often seem to have a prioritized list of desires and aversions; if my dog is hungry, she’ll go to the food bowl, but if I then pick up her leash and collar, then she’ll get excited to go our for a walk, ignoring her food. A dog that is afraid of noises but wants to go outside might visibly alternate between going out and coming back in (conflicted behavior).

Knowledge: dog, no.

What would it mean for dogs to have real knowledge?

Knowledge, to me, implies some sort of conceptual reasoning ability, planning, or generalization that can’t be explained by any of the above capabilities. We humans can talk about all sorts of conceptual information; we can make decisions based on knowledge of distant objects or abstract concepts; we can plan for the future. Dogs can do none of these things.

Being able to perform an action is not enough: a bird can fly, but it does not know about flight any more than an airplane does (“competence without comprehension”, in the words of philosopher Daniel Dennett). (I think it would be fair to say “a dog knows how to bark,” meaning that it can bark, not that it can understand or explain barking.)

So often, we see dogs behave in ways that betray an absence of planning or reasoning ability.

For example, sometimes I play fetch with a dog that doesn’t seem to quite “understand” the game of fetch. I throw the ball for the dog, and they eagerly run after it, ears and tail high. They grab the ball and bring it back and stare at me eagerly, big smile on their face, ball still firmly in their mouth. It seems like they want me to throw the ball, but they also do not want to give the ball back, and they don’t seem to understand the contradiction. (There are other possible interpretations of their behavior here, but this is what it seems like.) Once you get the ball back, one way or another, you can repeat this this over and over. If dogs could reason or plan, they ought to predict that giving you the ball will be a lot more fun than refusing to give it up, but they will not figure this out on their own. Some dogs will learn to give the ball back with enough repetition, especially if you train them to with rewards.

Training is mostly about training habits, and training a new skill usually takes many, many repetitions; one rule-of-thumb I’ve heard is 100 repetitions. If dogs could reason about training, then perhaps they would be able to learn some skills faster: “hmm, it looks like this is a training session, and the human is trying to teach me a skill—what is the skill I should learn here? Aha, I got rewarded a few times now for touching my paw to his hand after he said ‘shake.’ Let me try a few other variations just to test my hypothesis.” If dogs could reason in this way, then perhaps training would be much faster.

Once learned, dog behaviors often seem automatic. It took a long time to train my dog with the “drop” command, but, now when I say it, the ball seems to fall out of her mouth involuntarily, as if by hypnotic suggestion.

I often remind myself that, not only does not my dog not know what is happening next, she does not expect to know what is happening next. If I take her to a friend’s house, she does not know if we are spending 10 minutes there, or staying for the rest of our lives. And it does not bother her not to know these things, because she does not expect to know, she has never had the experience of knowing a thing. She only experiences what is happening now, in the moment. It’s a pure moment-to-moment experience.

Reframing knowledge as behavior

An exercise that I once set for myself, and that I recommend trying out is: banish “knowledge” and related concepts from your dog vocabulary. When you want to say “she knows” something, or suggesting that she’s reasoning or planning, can you find a way to describe their actions behaviorally?

“He knows he gets a treat if he sits.” Dogs learn skills through conditioning, not through conceptual understanding.

“She knows her name.” When a dog perks up after you call their name, that does not mean they “know their name.” They don’t even know what a name is. You could not use their name in a sentence and have them understand the sentence. They have simply learned to pay attention when they hear that word, and perhaps to form expectations about what comes next.

“He knows how to get out.” During one of my first orientations at the animal shelter where I volunteer, the behavior specialist explained that we shouldn’t let jumpy dogs right out of their kennels. She told us to wait until they’d stopped jumping at the door, and to open it a little. If they start to jump or rush out, then close the door. This way, they learn that they only get to go out if they had all their paws on the ground. She said: “If he knows that he can get out when he jumps and pushes, then he’s going to keep doing it the next time someone comes by.” But the dog doesn’t know this: the dog simply jumps and pushes naturally when you’re at their door, and gets rewarded by escape.

Sometimes dogs learn to open their own kennel doors. These kennels are on the adoption floor and they have regular door handles. These “escape artists” learn to jump on the door handle which opens the door (if you’ve forgotten to latch the other side). One might think that the dog figured it out by some sort of problem-solving: they saw humans push down the door handle to open it and figured they could do the same. But they don’t reason this way. Instead, they naturally jump on the door out of excitement or wanting to get outside. Sometimes, they accidentally jump on the door handle when they do this, opening the door. If they do this often enough, then they learn how to open the door: they’ve stumbled into a new behavior, not found it by problem-solving.

“She knows you’ll come back.” In my struggle to help my dog with her isolation anxiety, I’ve often heard people say my dog needs to “know you’ll come back.” But no dog that you’ve left alone knows that you’ll come back, because they cannot know things. Training dogs to be alone is about training them to feel comfortable and to settle down when they are alone. When I leave my dog with a friend, or a friend leaves her with me, she is sad temporarily about the departure and then seems to forget about it. When I leave her entirely alone, she is unhappy for the entire time she’s alone.

Why it matters

Understanding dog behaviors helps us train them better. Moreover, the more we understand them, the stronger bonds and empathy we will form with them.

When we misunderstand dogs, we often get frustrated and angry. We might feel like they’re deliberately being jerks to us. Better understanding lets us reframe these frustrating behaviors more accurately.

If a dog steals your food off the table, it’s pointless to get angry because “he knows it’s my food” or “he knows he they’re not allowed on the table.” They have no way of knowing either of these things. They’re just doing normal dog things. If you make food available to a dog with poor impulse control, of course they’ll go after it. The primary effect of yelling at a dog is to make yourself more upset.

Wrong training philosophies do get results. People have successfully train their dogs with traditional dominance-based training, based on disproven theories of “alphas” and punishment. But dominance and punishment cause all sorts of negative side effects, including creating aggression and fear in dogs, since the dogs have gotten all sorts of punishment that they did not understand or deserve. These strategies can be especially bad for dogs that start out with behavioral problems.)

Day-to-day, there’s no need to stop saying that your dog “knows” things. It’s often a lot simpler and clearer. However, it’s worth keeping in mind that “dog knowledge” is a shorthand, not literal truth.

One anecdote from trainer Kathy Sdao sticks in my memory. Sdao’s adopted a rescue dog who was sweet and gentle for months, but then he unexpectedly bit guests twice in her house. She asked for advice from several other trainers. One trainer saw Sdao let the dog on her sofa, and insisted that the bites were due to her failure to set a strict dominance hierachy. Another trainer suggested identifying the dog’s triggers: what was in common between the two bite incidents? And Sdao reflected that both bites had occurred when the she had a female guest and they were drinking red wine. This was an “aha” moment: first because it gave some clues about what the triggers were—perhaps the dog had once had an owner who became violent drinking red wine—but also a sign that the dominance-based trainers were operating out of dogma rather than looking at the evidence. By studying the triggers, she was able to manage the dog’s behavior, whereas asserting dominance would have just created more and more barriers between her and her dog, making it a less compassionate and loving relationship.

Relevance to human behavior

Human behavior is animal behavior, but with richer cognition than other animals. So many lessons from animal behavior give lessons on human behavior.

Some of us believe that we are thinking beings who also have emotions and habits, but this is backwards. We are emotional, habitual beings who also have thought. Much of our consciousness focuses on thoughts, creating the illusion that most of our mental process are thoughts.

So many of our behaviors are habitual, unthinking. If you’ve ever left your house and later wondered if you’d locked your door, then you were performing purely habitual behviors without thinking about them. Some strategies for building healthy routines and habits, involve dog-training-like strategies, e.g., reward yourself for successes.

Emotional behaviors are habitual too, like getting angry or pleased at a particular stimulus. Some dogs at the shelter are afraid of all new sights or sounds; others are eager and curious. Some dogs easily feel threatened and bark at every new sound. These are learned habits, and they have direct analogues in human behavior: think about people who are shy and fearful in new situations, including new social situations, people who are quick to anger if they feel their status is at all threatened, and so on.

So often have I been in meetings or witnessed interactions where people behaved badly, it was clear they’d gotten angry about something else. I’ve witnessed myself doing this too.

If someone invites you to an event, and you cannot go, whether you respond in a positive, appreciative way, in a non-responsive way, or in a negative way, will affect the other person’s desire to keep inviting you to things in the future.

I recently heard about a friend’s relationship in which her partner was trying to do a better job at calmly expressing when he is unhappy. I asked how she responds when he does so. She said she typically gets mad. So there’s obviously some negative reinforcement there.

Reading

The best book on behaviorism for animal training (and also humans) that I’ve read is: Don’t Shoot The Dog by Karen Pryor, the trainer that led the way in positive reinforcement training. B. F. Skinner was one of her mentors, and she begins the book with a full-throated defense of Skinner’s behaviorism, which amused me because of all the bad things I’d heard about behaviorism over the years.

Pryor developed her training philosophy with dolphin training at SeaWorld, and dolphins can only be trained with positive reinforcement. She then adapted those ideas to dog training, and found they worked much better than traditional training methods.

“Dognitive science” beyond behaviorism also seems like an interesting topic, but none of the papers I’ve seen in it seemed to get at the things that interest me.

Since then, researchers have extensively studied which aspects of perceptual and other skills seem correlated with improved skill at drawing, but have yet to form clear explanations for why drawing from life is hard.

In a new review paper, Judy Fan and I propose a new way to understand how people draw from life, and why it is hard:

In the paper, we argue that drawing is hard because we humans have very limited vision and memory—much more limited than we think. We just don’t perceive the world as a picture. And so, learning to draw realistic pictures is not about innate talent, it’s about learning skills to bypass these visual limitations that we all share.

One important caveat: this blog post is about just one aspect of drawing: transferring information from what you see to the page, as in copying an existing drawing. We study copying because it isolates some crucial aspects of realistic drawing skill. But drawing pictures isn’t just copying, for lots of reasons. Our paper does survey some other aspects of drawing, and I’ve also written elsewhere about them, including perspective, choice of subject, picking which lines to draw, abstraction, creativity, speed and sketching, taking risks, and the role of different choices along the way.

Our visual limitations

One might think that an artist looks at the world, and then draws their mental picture. But no one can draw accurate pictures this way. We do not see entire scenes this way, and we do not store complete, pixel-level pictures in our heads.

Human vision is severely limited, both in terms of what we see at any moment, and how much of it we remember, even over short periods of time. And, these limitations are not obvious—we are unaware of how much we do not see, until it is pointed out.

As an example, suppose you wanted to draw a picture of this houseplant.

Try looking at the picture, and then looking away from it. How many details do you remember when you look away: how many branches does the plant have? How many leaves does each branch have? Generally speaking, these questions are easy to answer when looking at the picture, but nearly impossible once you’ve looked away. Once you look away, the information is gone from your head. And getting the fine details purely from memory—the curvature of the stems, the precise outlines and shading on the leaves: definitely not possible.

If you were to try to draw the picture from memory, you’d be making a lot of it up, based on what you know about what plants look like.

This demonstrates the first important limitation of human vision: we remember almost no fine visual details. We don’t store pictures in our heads like they were JPEGs. We do remember things about the pictures we’ve seen, but not at the level of detail necessary to accurately draw them.

Here’s the second thing to try: stare at one leaf on the plant, and then, without moving your eye from that leaf, try to answer those questions again (how many branches? How many leaves?). Again, if you fixate your eyes only one spot on the plant, then you get very little detail about the rest of the plant. Again, you couldn’t draw an accurate picture of the plant just staring at one leaf.

This demonstrates this second important limitation of human vision: we get very little detail in peripheral vision, that is, in directions that we’re not looking. As another demonstration, try staring at one word in this paragraph, and, without moving your eyes, see how many other words you can read. The answer will be almost none.

Moreover, not only do we see and remember very little, we’re really unaware of this; we tend to think that we’re seeing everything in front of our eyes, but this is an illusion. I’ve written much more about this illusion and provided surprising demonstrations about at that link. I have also written about how these illusions determine how we use and perceive perspective in pictures.

Once we understand these limitations of human vision, then a whole lot of things about how people draw begin to make sense.

Eye movements in drawing

Here’s Henri Matisse drawing a picture of his son. He doesn’t just look at his son once and then draw from memory. Instead, watch how much he moves his eyes back and forth between the drawing and his son:

What exactly are his eyes doing?

Here’s another video of a person copying a picture, but this time recorded with an eye tracking device. The black rectangle show where the his eyes are looking at any moment.

The artist’s eyes are continually going back and forth between lines in the drawing and the lines in their copy.
In both videos we see that drawing from observation involves near-continual eye movement between the source and the drawing.

Why? Given the limitations of human vision that I demonstrated above make it obvious. Consider just copying a drawing. If the brain can store only very little detail at a time, then you can’t remember a detailed version of the source picture or your copy. You continually moving your eyes back-and-forth to gather information from the source and then see where to put it in the target. It’s a bit like trying to transfer water between two pots using only a leaky teacup.

What happens if you restrict eye movements? One common drawing exercise is called “blind drawing,” in which you are not allowed to look at your drawing, only your subject. Here’s an example of a blind drawing that I made at a workshop, where we were given 30 seconds to draw the person next to us using a single line:

(I took the photo a moment later.) The “errors” are typical of blind drawing: individual shapes are accurate, but their relative proportions and positions are way off. I drew her eyes, nose, and jawline first, then her hairline, but, by the time I got to the hairline, I didn’t know where my pen was relative to the her nose and eyes, and so I put her hairline right over her eyes.

When trying to do life drawings for the first time, beginners tend not to look at the source very much; they spend most of the time looking at their drawing. As a result, they can’t get many details very accurately simply because they never see them. One study found that requiring novices to look at the subject more frequently improved their copying accuracy. Blind drawing is also an exercise often giving to beginner students.

Learning to draw, as they say, is partly about learning to look, and, to start, that can simply mean spending more time looking.

Drawing is a set of skills

Everyone shares these visual limitations. Learning to draw is a matter of learning skills to work within these constraints. Artists develop a range of different techniques, strategies and skills for all parts of the drawing process. Here are just a few examples.

To draw simple lines and curves, the “target locking” technique for drawing an individual line is to position your pencil at the start of the lines, fixate your eyes on the end of the line, and then draw to where you fixate:

Nobody teaches you to do Target Locking, and nobody does it intentionally. In fact, almost no one knows about it, even if they do it all the time. Target Locking is a description of behaviors observed in a few eye-tracking studies. People are not good at being very aware of how their eye movements operate. (Another behavior, from the same study, is “smooth pursuit,” where the eyes follow the pencil.)

When I read about ways that artists use coordinated eye and hand motions simultaneously in copying, I wondered if I drew that way. So I tried drawing a sketch at my desk, and—indeed—I was moving my eyes and hands in parallel just as they described. Before then, I’d had no idea.

A more high-level example. Here’s a simple fact: a person’s eyes are half-way between the top and the bottom of their head. Yet, often our drawings don’t come out that way. Here’s a sketch that I recently drew for a paper figure:

There are a few things to point out here:

1. The cartoon character’s eyes are almost next to the top of his head, which would look totally wrong in real life.
2. You might not notice this anatomical mistake in the cartoon unless it is pointed out.
3. I know this anatomical fact, and yet I still made the mistake.
4. I don’t care about the mistake; I’m perfectly happy with this cartoon. I don’t consider it a mistake; I think the cartoon looks fine without anatomical accuracy.
5. But if this were part of an initial sketch for a more-accurate drawing, then this would be a problem, and one might not notice the problem until much later.

This knowledge can be useful for drawing: one study found that informing novices that the eyes are in the middle of the head improved their drawing accuracy. This conceptual knowledge can help you make conscious judgements to improve your drawing. This is why realistic artists sometimes engage in careful studies of anatomy.

Here’s an explicit technique that you can use to draw faces and avoid this problem. First, draw an oval for the head, then a cross to indicate gaze direction, like in these examples:

People have created all sorts of more-elaborate such rules for other kinds of drawing, including one-point perspective and two-point perspective, and rules for drawing full bodies, which has led to some great parodies.

In the paper we survey many other kinds of strategies and skills that artists learn.

Anyone can learn to draw

In Western culture, we’re taught that artists are some sort of special geniuses with innate talent—you’re either an artist or you’re not—a belief that only dates back to 18th-century Romanticism and misguided educational philosophies.

This mystical nonsense prevents people from learning to draw. But all you have to do is take a chance and try it out. I personally have known so many people who took a drawing class or worked from a book of exercises and went from “I can’t draw” to “Wow, I drew that?”

Our work suggests some reasons why people might mistake the inherent difficulty of drawing with innate lack of talent. We’re unaware of how limited our own vision and visual memory is. So it’s like there’s this invisible mental block—limitations of vision and the illusion of awareness—and we think that it means “I’m not good.” But simple exercises can help you start to work around this mental block. Once you understand the block, it makes more sense.

Drawing is a skill you learn, not an innate talent. I think there are only a few things you need to learn to draw: a little bravery, some exercises (especially from a book or from classes), and lots and lots of practice. I wrote a little more about this here. Just a little time and practice can make drawing an engrossing pursuit that enriches your life and becomes its own reward.

We can divide social relationships into three categories: close ties, like close friends and family; weak ties, like coworkers and casual friends; strangers, like people you might pass on the street. All three kinds of social relationships are important, and this post is about ways that technology has transformed each of them in the past century.

Technology often benefits our social relationships, in ways that are often obvious. The negative effects are less obvious, which is why I focus more on those. I use and enjoy every category of technology discussed in this post: I love listening to recorded music; I use my smartphone and/or laptop all the time; I love connecting with friends and colleagues on social media. I don’t believe any of these technologies are “bad,” but they are not “neutral” either.

Many different trends inspired me to write this post. The past decade has seen a lot of simplistic claims about new technologies being good or bad. Simplistic arguments over technology as good or bad themselves are harmful because they occupy resources and attention, distracting us from the underlying problems. In my work I’ve tried to understand if these arguments make sense, and, the more I’ve dug into them, the more complexity I find, but also common themes. I’ve previously written about the social nature of art and how it impacts the role of computers as art technology, and the effect of recorded music on social interaction; here I consider other technologies’ effect on social relationships. Most directly, this post is a response to simplistic claims like “social media causes mental illness”. Social media has its problems, but the story is a lot more complex than that.

Cars: from villages and cities to highways and suburbs

Automobiles are the Original Sin of 20th-Century American social isolation.

Isolation always existed throughout history. But, by the end of the 19th century, many people many lived in communities where they saw their families, friends and neighbors as part of their daily lives, whether farming villages or larger cities. Even poor immigrants overcrowded into New York City tenements often lacked basic needs but frequently had each other.

 

Henry Ford’s invention of the mass-produced automobile—and the powerful automobile industry that grew around it—radically affected the 20th century development of American cities. Before the rise of the car, city streets were public spaces for everybody: pedestrians, horses, children playing. Automobile manufacturers began a program of remaking cities for cars, through ad campaigns to stigmatize pedestrians in roads as “jaywalkers”, lobbying for laws against jaywalking, and lobbying for more roads. (“Jay” is old-fashioned slang for a country bumpkin; calling someone a jaywalker is like calling them an idiot.) Public transportation and walking became less-and-less practical, leading to a vicious cycle: more and more people drove cars, and then demanded more roads. And, when cars hit pedestrians, pedestrians were blamed.

Automobiles offered freedom and independence: convenient ability to travel far, whether for work or reaction. They signalled social status, wealth, independence, and manhood. They became popular, and then mandatory.

By the 1940s, a generation of urban planners favored cars as the primary mode of transportation. Most famously, New York’s transportation overlord Robert Moses grew up wealthy and never learned to drive, yet rode in cars his whole life. He built a staggering number of roads, highways, and bridges around New York City and Long Island, while blocking decades of public transportation projects.

Throughout the US, a new kind of population center arose: the “suburb.” Suburbs allow people to have big houses outside of the cities where they worked. Zoning separated houses from businesses, and large houses were separated from each other, creating massive urban sprawl, and making driving mandatory for most activities in many suburbs.

In the 1950s, federal funding such as the Federal-Aid Highway Act led to the construction of highways throughout US cities. Many cities drew highways right through urban centers, often destroying vibrant communities (often, poor and/or Black) and creating urban blight. Now, some cities devote an enormous fraction of their space to parking lots—some cities have more parking than housing.

By the 1970s, the terrible consequences of our car-first policies were clear: intense traffic jams on highways; smog that considerably shortened peoples’ lifespans; racial segregation; urban blight; unhealthy lack of exerise when people drive everywhere; dependence on foreign oil, as highlighted by the 1970s energy crisis.

 

Since then, we have undone some of these ills. Emissions regulation mostly eliminated smog; alternative energy modes have reduced dependence on oil; many cities now favor mixed-use urban planning, while investing in bike lanes and public transportation. A few urban highways have been removed, like San Francisco’s Embarcadero Freeway and Seattle’s Alaskan Way Viaduct. But the flywheel of highway demand still drives much of the politics: car drivers, stuck in traffic, continually demand more road in the intuitive but utterly wrong idea that adding roads reduces traffic, and fight against any attempts to share the roads with more-efficient transportation modes.

Car culture and social isolation

But when it comes to social isolation, I think most people still don’t realize just how much our car-first culture has hurt our connections with other humans.

After all, cars seem empowering: if you want to go see friends, go to a restaurant or bar, go to see live theatre or sports… why, you just get in your car and drive there. The car makes it possible! But American urban planning made the car necessary for most of these activities.

Car-first culture reduces meaningful encounters with strangers. As Joe Keohane describes in the surprisingly-deep The Power of Strangers, interacting with strangers and acquaintances creates a sort of societal glue: not only can it improve your own mental well-being, and help you make connections, it ties us together in a community. Even walking down the street, you have passing interactions with strangers; depending on how segregated the society is, you could be passing a cross-section of humanity. It’s much easier to feel some level of empathy for people that you interact with casually, in person. Interactions with strangers has played an important role in human evolution and in many ancient cultures.

When we travel by car, we self-segregate ourselves from all other humans, with virtually no direct interaction beyond fleeting eye contact. Our interactions with others often create stress, and more anger than empathy.

For most of human history, social interactions outside of the home was built into daily life: daily activities naturally involved seeing people. In the suburbs, you can drive to places to interact with people, say, driving to churches, meetups, bars, or public events. You have to make an effort to have interactions with people, as a separate part of your day. Like driving to exercise at the gym.

My father was born in San Francisco, as was his father. After World War II, they moved to the suburbs, like many of that era. My father stayed in the suburbs until retirement, and then moved into a retirement community back in San Francisco. He said ruefully that it was the first time in his life that he felt like he lived in a “neighborhood:” a community where he interacted frequently with his neighbors.

Many seniors choose to stay in their suburban homes throughout their old age, becoming increasingly isolated as their mobility diminshes, leading to increased loneliness, in turn, worsening dementia and other health problems.

Social isolation is particularly bad for children, as I’ll come to in a bit. But first, some parallel developments in entertainment.

Entertainment: From performance to recordings to video games

Throughout most human history, social groups made music together: families played instruments together in their living rooms and porches; city dwellers sang together in taverns; churchgoers sung hymns and gospel together.

In fact, social musicmaking predates human history. Savage et al. survey a range of anthropological evidence that social bonding is the reason why we have music. Music performs important social functions for how we relate to our fellow humans around us.

At the start of the 20th century, Edison invented the first music recording device: the wax cylinder. You no longer had to be in the same room as a performer to listen to music. Recorded music grew in popularity, and, through many formats since (tape reel, vinyl record, cassette tape, DVD, mp3, streaming), remains a huge force in our culture today.

Recordings allow you to develop a sort of relationship with a performer. Sometimes we feel like they’re directly communicating with us. It’s a one-sided, parasocial relationship, but a relationship nonetheless.

With the development of music recording, radio, cinema, and, then, television, communal music-making diminished. Why go through the trouble of learning instruments, getting your amateur friends together, picking a song, and playing it, when you can listen to a recording of some of the best music in the world? Or watch amazing artists perform on television or online?

Many 19th-Century living rooms focused on a piano. By the time I grew up, most living rooms were focused around the television; the living room of one of my high school friend’s centered around an enormous television that we called “The Shrine.” Watching television together can still be a communal activity. Sometimes the show is a reason to get together. But you can also enjoy music and TV alone.

Around the 1980s, home computers and video game consoles appeared. These could also be social—going to a friend’s house to play games together—but one could also spend many hours alone with computers and video games. You no longer needed another person with you for playtime; the computer could be your play buddy. In the suburbs, this was a whole lot easier than actually finding a friend in person. Online gaming and the Internet added social elements, with email, BBSes, Usenet, and then the Web. So you could socialize and play alone, or with friends and strangers online.

With the Sony Walkman, recorded music went into our pockets, we could be out and about, with headphones separating ourselves from the world.

In the early 2000s, computers went into our pockets: with mobile phones, and then smartphones. Now each person carries their own television, cinema, game console, and Internet device around in their pocket. Each person at home or out and about can be in their own separate world of games and passive entertainment.

Children: From free range kids to social media

Let’s come back now to the car-first suburbs, and its effect on children.

As I understand it, children in pre-1970s villages and big cities used to roam freely together for unsupervised play and exploration. Think of the adventures of Huck Finn and Tom Sawyer, or so many other fictional children that explored and played outdoors, unsupervised. The suburbs made this increasingly difficult for many kids.

As a suburban Gen X-er with working parents, I was a latchkey kid who, every afternoon, came home from school alone and stayed at home alone until dinnertime. My closest childhood friend lived over a mile away from me, which, for a kid, was quite a long distance. So we saw each other some days and not others. My other friends lived even further away, and it’s hard to be close with someone you don’t see often. I had a bike, but no interest in using it, which I attribute to a lack of role models, since adults never biked.

My social circle only became really active toward the end of my childhood, once we were old enough to drive, and suddenly groups of us could go do stuff together most evenings and weekends.

A series of “stranger danger” panics made things worse. We were told not to talk to strangers, who might take us away in a white van for unspoken purposes. In essence, we were taught to see every stranger as a mortal threat. But the threat was vastly overstated: stranger abductions were quite rare, much rarer than children dying in car crashes.

Yet, this perceived threat led to seismic shifts in childhood for generations of children. Because of “Stranger danger” fears, millenials were often not allowed out unsupervised alone at all, while helicopter parenting led to a decrease in unsupervised play time entirely. A compelling article by Gray et al. (official link) surveys extensive sociological and psychological evidence that losing unsupervised play time led to an increase in mental illness among millenials.

Children raised in a bubble will struggle outside of the bubble, sometimes for their whole lives.

Since millenial kids had less opportunity to socialize in person, they turned to early social media sites like MySpace and Facebook. Parents responded with another moral panic, fearing the effects of social media on their kids. But, as danah boyd found in her ethnographic studies, these kids were just playing out all the usual teenage social dynamics online. In this case, technology replaced many of the social interactions that these kids were denied in person.

Internet: From social media to “AI”

In the 2000s, social media—Facebook, Twitter, Instagram—became one of the main ways that we interacted with other people in society. More and more, we interacted with our friends, with strangers, even with celebrities online. As newspapers fell to the one-two punch of private equity and Internet competition, we began to get more of our information about the world from social media sites.

Online affordances have long been blamed for a lack of online civility:

When you don’t see someone in person, it’s harder to have empathy for them. It’s hard to communicate in a considerate way. It’s easier to be misinterpreted. And it’s a continuum: a Tweet is worse than a Zoom call, but a Zoom call is still worse than physically meeting in person.

By the 2010s, America’s polarization reached a crisis level of dysfunction, and many people blamed polarization on social media. I blame propaganda machines (Southern strategy, talk radio, Fox news, etc.), but social media really greased the wheels. In these semi-anonymous online services, messages from your neighbors, from people who share your own values, from political operatives, and from Russian state sources could all look the same and carry the same weight.

By the COVID lockdowns in 2020, we had adopted online technologies so thoroughly that it was fairly straightforward to retreat indoors and conduct all of our interactions with other humans electronically. Electronic communication became a life-saver, literally. But we suffered for the isolation, especially children. And, when we did see each other in person during the pandemic, and, then, when we first came back out into world, to our first restaurants, concerts, and parties, we briefly felt just how valuable and important it was to be out with other people.

As someone who has worked in “AI” research for many years, I find the new technologies amazing and wonderful; it’s so cool and fascinating that they work. A few of the times I’ve tried it, ChatGPT has given me correct answers to questions where all other avenues I’d tried had failed. I’m excited about the way new technologies continually revitalize art.

But new “AI” technologies also extend and amplify all of the threats to our social relationships, mental health, and social fabric. In the most basic sense, even casual interactions with customer service is being replaced by chatbots. We hear of professional writing often replaced by text generation (e.g., scientific paper reviews, law briefs); professors get messages from their students that look like personal communications but were obviously written by chatbots. Even more, we’re hearing stories of people forming intense personal relationships with “AI” chatbots—in a sense, fully replacing human interactions with technological interfaces.

More and more, we won’t even know if the words and conversations we see online were written or spoken by a person, further diminishing our trust and sense of connection to the people in our society.

Does modern technology generally lead to isolation?

Did all of these technologies make us less social? Are there any common lessons and themes? Do automobiles and social media have anything in common?

First, it’s worth repeating that technology does not cause isolation. Technology does not cause anything. But it shifts the scales of what is easy and what is possible.

Moreover, it’s worth repeating that these technologies have a complex impact on our social relationships. If you live in a suburb, then having a car will be very important to having in-person social relationship; you can’t get by without it. So, in that immediate sense, having a car enhances your relationships. Likewise, social media lets you connect with friends and family in a way that you might not otherwise, especially when you live far apart from them. But, by virtue of making it easy to connect when we live far apart, they make it easier to live far apart, creating a vicious cycle.

As technologies get better, they automate many interactions that previously required a human.

A 19th century shopper might buy their meats from a butcher, and vegetables from various markets. They would have to travel to different sellers in their neighborhood, a time-consuming process; they might also form casual relationships with each of them. Then, as canned food became more commoditized, they were sold by grocery stores: you’d go to the grocer and tell them what you wanted, and they’d get your food from the shelves. In 1916, grocery stores became self-service: you find what you need and bring it to a human cashier with whom you’d exchange pleasantries. Starting in the 1990s, self-checkout machines meant no human interaction required. Then, delivery apps allowed you to order your groceries online: you could get all your food delivered without even leaving your house. No-contact deliveries in the pandemic meant you might not even see the delivery person. This made grocery ordering fast, easy, and without the least bit of human interaction. Some services are now testing and deploying food delivery robots.

We can repeat that kind of story for many other kinds of activities, that have, over decades, moved from time-consuming, in-person interactions to convenient, human-free apps.

Our online social interactions often have a lot of the same feel and elements of real in-person interaction. Phone calls are a great way to keep in touch, and I still treasure my regular calls with close friends. Watching social media lets you enjoy and participate in a great online conversation; I’ve learned a lot and had many meaningful interactions with colleagues and acquaintances online.

But watching social activity online is a bit like eating junk food. Junk food has the sweetness, umami, and/or saltiness that tells our animal brains that we’re consuming something nourishing. But it is not nourishing. Likewise, social technologies give us some of the elements of social relationships, without the same nourishment. Eating a bag of potato chips, or scrolling influencers on TikTok may be pleasurable, but they are unlikely to leave you feeling nourished.

One could extend the argument backward through history. When I mentioned my thesis above with some colleagues recently, one of them described books as socially isolating, and the others agreed (one described all the time her child spends reading). The industrial revolution was enormously alienating to generations of people who became mineworkers and factoryworkers, toiling in inhuman conditions for most of their waking hours. Or surely, one could say this about war technologies and surveillance—or heck, even the development of agriculture millions of years ago. If we view Pleistocene hunter-gatherer tribes as our ideal state, then every technology development has moved us away from that state, starting with the development of agriculture. But this does not seem like a productive argument.

How to Counter It

We are social creatures, evolved to live in small tribes of hunters-and-gatherers; all of our social behaviors, in some way, can be traced back to our Pleistocene ancestors. In-person interactions with neighbors and with strangers is one of our most basic needs, like food, health, and physical safety. And yet, we live in a modern world, with its complex demands and structures, and we’re not going back to our hunter-gatherer lives.

I still enjoy television, video games, social media; I daily use telephone, email, and instant messaging to keep in touch with close friends, family, and coworkers; I sometimes order food and products with delivery apps; I ride in automobiles when necessary, etc. In our modern world, each of these technologies can benefit our social relationships and other aspects of our lives. And, even if we wanted to give up these technologies, we live in a society in which doing so would be very hard. Their underlying problems are societal problems that we cannot solve alone.

However, I recommend taking stock of which relationships are valuable and/or nourishing. And, what kinds of interactions support those values? How do you feel after an evening with friends, versus an evening scrolling social media? I believe you will find that the in-person interactions are the most valuable, lead to the greatest connection and empathy. These are worth prioritizing. How can you organize your life to best support the things that matter (such as meaningful relationships) and not the things that don’t matter (sitting frustrated in traffic; scrolling social media)?

Moreover, in-person social interactions with strangers, though often fleeting, can really enhance our sense of connection with the larger society around us.

It takes conscious time and effort–and sometimes, resources—to avoid the noise and figure out what really matters.

Some readings I recommend

Derek Thompson. The Anti-Social Century. The Atlantic. 2025.
Thorough and well-researched article that makes many of these points, and more (sent to me after I posted this; thanks, Moshe Vardi).

Joe Keohane. The Power of Strangers: The Benefits of Connecting in a Suspicious World. Penguin Random House. 2022.
A surprisingly deep exploration of the value of talking to strangers, through anthropology, sociology, and psychology, and how talking to strangers is important both personally and societally.

Amy Orben. The Sisyphean Cycle of Technology Panics. Perspectives on Psychological Science. 2020.
How moral panics recur over each new technology, and waste time and resources by misunderstanding the new technologies.

Peter Gray, David F. Lancy, David F. Bjorklund. Decline in Independent Activity as a Cause of Decline in Children’s Mental Well-being: Summary of the Evidence. The Journal of Pediatrics. 2023. (Alternative link.)
A compelling explanation for the rise in mental health issues.

Melvin Kranzberg. Technology and History: “Kranzberg’s Laws”. Technology and Culture. 1986.
Is technology good or bad? No.

Jane Jacobs. The Death and Life of Great American Cities. Random House. 1961.
The influential and insightful account of how mid-century American urban planning fails, and how cities can function successfully.

Robert Caro. The Power Broker: Robert Moses and the Fall of New York. Knopf. 1974.
A gripping account of (among many other things) how car transportation took over New York City, and some of the harms that resulted.

My blog post on status quo bias: peoples’ tendency to assume that new technologies and new kinds of art are bad.

My blog post on recorded music. Some pros and cons of recorded music technology.

This post was inspired, in part, by conversations with Moshe Vardi. Thanks to Steve DiVerdi and Obie Pressman for comments.

They said that “AI” was going to “kill art.” But people said that back when photography was invented too. They said that it’s not art because it’s just “pressing a button.” But people also said that about photography for over a century. They said that “AI” can’t make art because it doesn’t have soul. But people also said that about recorded music. People said that “AI” will take jobs from artists, which also happened with recorded music, computer animation, and many other past art technologies. People said that it is “just copying”… but all art involves some level of copying and inspiration, and especially artforms explicitly based on copying, like photomontage, sampled music including much hip-hop and turntablism), and appropriation art. (The “just copying” claim is factually not true for visual generation, with rare exceptions.) People say that most art with the new technology is bad, but most art made at the dawn of any new art technology is not “good”. And so on, and so on.

Even if all new technologies have a lot in common, they also must have differences. But what are the relevant ones for “AI”?

Certainly the fact that many popular “AI” models are trained from scraped datasets is important in some way, but that alone does not indicate its impact on the arts. Internet search engines are also built and trained from large scraped datasets in ways that can affect the arts.

The value of historical analogies and trends

In my work, I have argued that new artistic technologies tend to appear in consistent ways, good and bad, including the backlash. Teasing apart these historical analogies is important for a few reasons.

First, the historical analogies help us overcome cognitive biases. Our gut reactions to change are often wrong, and historical examples help us see this, and help us better understand the real dangers. For example, a natural response is to try to protect existing artists is to expand copyright, but this requires caution since overbroad copyright has a long history of hurting artists.

Second, if you want to convince people, then you need defensible arguments. For example, Ted Chiang’s argument against “AI” as art seems not to have been informed by any knowledge of the history of photography or conceptual art or computer graphics or the way algorithms make choices or current developments in interactive tools. His piece will not persuade people who know about these things. His piece could have been a good argument that text-to-image is not a good artistic tool, something I agree with, but that’s not the argument he claimed to be making.

One possible difference is that the tools became effective much faster than with previous technologies. Portrait photography took decades to become widespread, and consumer photography took even longer. But it’s hard to judge speed in the moment, amongst all the breathless hype and controversy; some claims of the immediacy of “AI”’s actual impact may be overstated.

The biggest difference

After a recent talk that I gave, a student in the audience pointed out a way to phrase the difference that I think may be important.

Nissim Maruani wrote to me that: “for the first time in history, human and machines are actually creating using the same art medium … what does art become if the spectator cannot distinguish generated and human-created content? More literally, what would be the incentive for a human artist to publish on Spotify if the public cannot distinguish their work from free, on demand, and even maybe personalized music?”

Was it hand-drawn, digitally painted on a tablet, or generated by algorithm? There are some telltale signals (weird physics, extra fingers appearing), and various stylistic clues but it’s often very hard to say for sure. To some extent, telling the difference may be possible with expertise and connoisseurship, although this can often be misled

In contrast, early photography was never mistaken for painting. You would rarely listen to recorded music and think that it’s a live performance (exceptions include lip sync). Early computer animation looked nothing like live action film or hand-drawn animation.

The move from analog to digital techniques might be the closest analogy here, and each of these had some controversies in their time, e.g., digital cinema versus analog photography, digital image editing. Nowadays one cannot tell the difference between computer graphics and live-action cinematography.

Implications

So, unlike a lot of the gut-response objections, I believe that this Indistinguishability one is valid, and worth considering the implications of.

I am not claiming that the idea is new, but that this is a defensible way to describe how these technologies differ. People have been discussing the implications of this fact, even if they might have phrased it differently. And, I think this is something that was missing from my previous discussions of the topic.

Here are some possible takes.

One take is “AI slop” “crushes creativity”, by making all work meaningless when we can’t tell any of it apart.

Culture theorist W. David Marx has argued that cultural value has already stagnated for the past several decades, and that “AI” merely accelerates this trend: “less value is created when all cultural artifacts are procurable with enough money, can be made anywhere by anyone, and offer no useful social distinctions between philistine and aesthete,” fulfilling an “extinction-level destruction of cultural value” predicted by the Postmodern philosophers decades ago. He further predicts “a re-evaluation of real-life experiences”… maybe we’ll get away from our screens more, if online cultural activities offer less status or meaning?

I do believe that history shows that the best art requires human authorship, experience and connection; we care about art because it is made by humans. Improved tools raised the bar for the best art. But there are many areas where we may not care so much, and these are the areas that “AI slop” inhabit.

I also think the analogy to photography remains useful: photographs didn’t look identical to paintings, but they were close enough to radically challenge the existing notions of the value of creating realistic pictures. Once photography grew widespread, making realistic pictures was no longer valued in the way it had been when it required a painter’s skill. The subsequent Modern Art movements reversed the understanding of artistic skill and talent in the contemporary art world; in Van Gogh’s words, recreating reality was “just photography”.

While skill may still be valued in contemporary art, it is nothing without a good origin story and ideas behind it. In response to the idea of connoisseurship, contemporary artist Jason Salavon likes to quote the study showing that wine enjoyment is affected by beliefs about the price of the wine in a way that can be measured neurologically.

Mass culture still values skill in a way that the contemporary art world does not. Perhaps mass culture will go through a similar transition.

Thanks to Kabir Ahuja for comments and pointing out the link to digital cinema.

In a 1906 essay entitled The Menace of Mechanical Music, the popular composer John Philip Sousa warned “I foresee a marked deterioration in American music and musical taste, an interruption in the musical development of the country, and a host of other injuries to music in its artistic manifestations, by virtue—or rather by vice—of the multiplication of the various music-reproducing machines.” In short, he predicted, recorded music will be disastrous for music and our musical taste.

Today, it’s easy to laugh his bonkers predictions, like that mothers will turn their children into soulless machines by playing them soulless recorded music instead of singing their children to sleep. We take music recording for granted, and surely we still have pretty good taste. It’s hard to imagine wishing that records, CDs, and mp3s had never been invented. So it’s very easy to dismiss his doom-saying.

But, the more I’ve thought about it, the more I think that he had some good points.

These days, many people predict artistic doom from our new technologies, and it can be easy to dismiss these predictions, sometimes by pointing back to older predictions of doom. But it’s helpful to look back at the ways that old technological changes did change everything, both in good and bad ways. Doing so can give hints as to how best to navigate new changes.

Was John Philip Sousa right in any way, and what lessons, if any, does his example contain?

Soullessness

Sousa first argues that music must express the performer’s “soul.” “The nightingale’s song is delightful because the nightingale herself gives it forth.” Hearing music from a machine, and “reducing the expression of music to a mathematical system of megaphones, wheels, cogs, disks, cylinders,” robs music of its soul.

As someone who has spent their life listening to recorded music at home—and, often, fallen in love and feel deeply connected to musical recordings—I just think he is wrong about this. It is true that there is really something special about seeing music performed live, but recordings also create deeply meaningful experiences, and a sense of connection with the artist.

Musicians long ago began making records meant to be heard as records, whether we’re talking about music concrète, The Beatles and Pink Floyd composing with tape loops, AOR rockers, densely-composed rap albums like “Fear of a Black Planet” or “Paul’s Boutique,” or EDM musicians like DJ Shadow and Amon Tobin sampling and reconstructing bits of other recordings, and so on. Recording has enabled new kinds of music that we now cherish.

Listening to a recording is more like reading a novel, rather than listening to a storyteller read live. And, like a short story or novel, you can study a recording, and go back to it over and over, in a way that you cannot with a perfromance.

Sousa writes “I could not imagine that a performance by it would ever inspire embryotic Mendelssohns, Beethovens, Mozarts, and Wagners to the acquirement of technical skill, or to the grasp of human possibilities in the art.” In this he is utterly wrong.

So many of our current musicians talk about being inspired by recordings. In the words of one artist: “The first Velvet Underground album only sold 10,000 copies, but everyone who bought it formed a band.”

Whenever someone says that machine-made art is soulless and cannot compare to human art, they should consider whether they are basically making the same argument as Sousa’s, or the complaints that artists once made against photography, or against computer graphics, and so on.

People making music together

But, on the next point, I think Sousa was right on.

Sousa wrote “The time is coming when no one will be ready to submit himself to the ennobling discipline of learning music.” In short, he wrote that people would not learn to make music because they could just listen to recordings instead of making it themselves.

He throws out many instances of camaraderie in people making music together that, individually seemed old-fashioned and absurd when I first read it. People gathering around a campfire won’t tell stories or play music, they’ll just listen to recordings! But, taken together, I think he did have a real point.

Before recorded music, friends, families and communities often made music together. Friends and family would gather in their living rooms; the piano was the center of social life in saloons and taverns.

In contrast, when I grew up, at the center of most living rooms was a television. I never learned an instrument so I’ve never experienced it myself. I’ve only been around friends playing music on a handful of occasion, but, these occasions seemed magical. I’ve heard from several people that singing together in groups and choirs is a powerful bonding experience.

In fact, there’s good evidence that communal bonding is the evolutionary explanation for why we have music at all. Like all arts, recorded music has its own social functions, for example, fans playing music together, and trading and selling, but I’m not sure these make up for not making music ourselves.

For most of us, experiencing music is a passive activity, especially as compared to before recordings. Even for many amateurs who play instruments, social music-making remains rare. He wildly overstated it when he said “it will be simply a question of time when the amateur disappears entirely,” but I think Sousa basically had a good point here.

Copyight

And now we come to the real reason for Sousa’s article: his call for copyright protections for composers. “Composers of the music now produced so widely by the mechanical players of every sort draw no profit from it whatever. … The composer of the most popular waltz or march of the year must see it seized, reproduced at will on wax cylinder, brass disk, or strip of perforated paper, multiplied indefinitely, and sold at large profit all over the country, without a penny of remuneration to himself for the use of this original product of his brain.”

At the time, the US had no copyright law granting composers’ rights, and Sousa’s article was written as part of a campaign to create one. The campaign was successful, leading to the creation of statutory licensing for music. Every time a music recording is played publicly, whether in a dance club or in your local cafe, some amount of money must be paid to ASCAP, which in turn is paid to composers. Licensing is compulsory, meaning that you do not need to obtain permission from the composers.

This system seems to have been a successful and uncontroversial way to compensate and incentivize music composition, so, in that sense, Sousa was “right.”

Is it good or bad? What are the lessons?

Is recorded music good or bad? I, for one, could not imagine giving it up, or going back in time and trying to prevent its creation, if such a thing were possible. Yet, today, we continue to struggle with the way recording has replaced human connection—to oversimplify a bit, we’ve gone from making music together to listening to a few enormously, skilled and talented artists at home alone.

It’s hard to take Sousa’s claims of the purity of performance too seriously, since his own band sold recordings for profits. And, earlier in his careeer, he had used Gilbert and Sullivan’s music without compensating them, to which Arthur Sullivan had said, “It seems to be their opinion that a free and independent American citizen ought not to be robbed of his right of robbing somebody else.”

New music recording and distribution technologies have since led to many, many new copyright battles, including battles over cassette tapes, sampling in hip-hop, file-sharing, pop art, appropriation art, and on and on. New technologies lead to compensation battles as well: video streaming technology led to the 2023 Hollywood writers’ strikes, and compensation for musicians from online streaming remains bad.

Each of these battles are accompanied with absolute claims about art like Sousa’s: not only does the new technology ruin compensation for artists, also it’s soulless, it’s bad for art. A lot of campaigners against new technology don’t just say it’s bad for compensation, they say it’s bad, period. But I think we should separate these issues.

I could not imagine wishing that these technologies themselves did not exist. You may consider yourself lucky if you have never had to go to a neighborhood video store to see what few movies you could bring home to watch, or flipped channels on a TV to see what was on—and ended up watching reruns and commercials. Giblin and Doctorow make a compelling case that the source of current artist precarity is not, primarily, technology or copyright, but near-monopoly market conditions.

With each new technology, the loudest voices make the most extreme arguments, for why the technology has nothing but benefits, or will ruin everything. In the words of historian Melvin Kranzberg, technology is neither good, nor bad, but nor is it neutral.

We, as humans, are so bad at intuitively reasoning about the complexities of these long-term changes. But there are many common trends that occur again and again with new artistic technologies studying these kinds of examples provides a useful counter against simplistic thinking and status quo bias.

Further reading

• The Menace of Mechanical Music by John Philip Sousa. Read it for yourself!

• Background on Sousa and his legal battles by Patrick Warfield

• My blog post on When Technology Changes Art

Thanks to artist Tom White for recommending the Sousa essay to me.

There are lots of other kinds of fiction, writing, TV and movies that changed the way I see the world, but those are much harder lists to curate, so this one is just about books. This is not a complete guide to my worldview, e.g., there are some topics for which fiction and documentary films may have shaped my worldview more than books (for example, I remember being profoundly affected by the novel A Fine Balance by Rohinton Mistry and the San Francisco documentary film “We Were Here.”) However, relevant to this list, the TV series Adam Ruins Everything deserves special mention for covering common misconceptions effectively across a range of topics; it’s the only TV show I’ve watched that included extensive citations.

I almost never reread books so I don’t know if some books would seem hopelessly dated now. A lot is about how the book hits you at the time you read it. Some books wouldn’t have meant anything to mean when I was too young too appreciate them, and others seem to obvious if I read them too late.

History of Science

The Age of Wonder, Richard Holmes (2008)
Rip-roaring adventure, science, and art of the romantic era, and how we came to some of our key modern ideas, including the concept of scientist (as with The Invention of Art which describes where the concept of “artist” came from). Great fun, and really fascinating for understanding where a lot of our ideas came from, from an era before we made any distinctions between “art” “science” and “philosophy.”

The Structure of Scientific Revolutions, Thomas Kuhn (1962)
The classic on paradigm shifts, and how “normal science” operates versus science in crisis.

“The Day The Universe Changed,” James Burke (1985)
The one TV show that I include in this list: a BBC history series that I watched as a child that I remember loving, and I do think it affected my worldview: how scientific and technological changes change how we understand the world and our worldviews. Some of the writing I do about how technology changes art might owe a debt to it. (And, being a simplified story for mass consumption, it has its problems.)

Intelligence

From Bacteria to Bach and Back, Daniel Dennett (2017)
I keep coming back to some of the ideas here, like “competence without comprehension” and the notion of “memes” (I couldn’t get through The Selfish Gene. (Longer review here.)

Are We Smart Enough to Know How Smart Animals Are?, Frans de Waal (2016)
The author describes how scientists have consistently underestimated animal intelligence, and the many forms it takes. One tidbit that spoke to me was his assertion that, while historically “scientific” definitions of intelligence have focused on logic and reasoning, emotional intelligence is far more important to survival, and far richer and more difficult to define or understand. Social animals need emotional and social awareness more than they need language. (Longer review here.)

Don’t Shoot the Dog!, Karen Pryor (1984)
I learned a lot about postive reinforcement dog training while volunteering in an animal shelter, but I didn’t understand the philosophy and evidence behind it. This is the book that put it all together for me, how positive and negative reinforcement can be used to guide behaviors for all sorts of animals, including humans. At the same time I also enjoyed Plenty in Life is Free by Kathy Sdao (2012) a brief memoir and argument against dominance-based training even in positive reinforcement training.

Societies and Cities

The Death and Life of Great American Cities, Jane Jacobs (1961)
The book absolutely transformed the way that I understood city life, so much that it’s hard to even remember what I thought beforehand. I felt like a cipher had been unlocked to help me understand the city in ways I never had before. I wished very much that someone had given it to me to read before I moved to New York City. I only found out about her after I moved to Toronto and went on a city walk celebrating Jane Jacobs’ activist legacy in Toronto. I became a huge Jane Jacobs fan. For more about her history as an activist in New York, along with a bit of Robert Moses’ history, I really enjoyed the short history Wrestling with Moses by Anthony Flint (2009). (Update: I’ve since read The Power Broker, and loved it. The length is intimidating, but once I started it was very easy to keep reading the whole thing.)

The Power of Strangers, Joe Keohane (2021)
A book about talking to strangers doesn’t sound like it would hold my attention for long, but this was a surprisingly deep exploration of the topic, through anthropology, sociology, and psychology, and many real-world experiences and practical tips. The author argues that talking to strangers is easier than we think, and an important glue in the social fabric that is in a poor state, and our country and world would be so much better if we did more of it.

Status and Culture, W. David Marx (2022)
The main thesis it that all culture, fashion, and taste arise from natural human status-seeking, and the book provides a dense analysis of the complex mechanics of both culture and status. Almost every page I was nodding as the book provided broad-strokes theoretical descriptions that perfectly matched my experience or understanding, together with a rich array of cultural examples to make it concrete. (Update: I’m also enjoying his follow-up book: Blank Space.)

Definitions of Art

A few years ago, I started reading a lot of books on definitions of art. Here are some of my longer reviews of those books.

The Art Instinct, Dennis Dutton (2009)
A philosopher argues that art is an evolved behavior. Includes a clear discussion of definitions of art, andwhat it takes to argue that something is an evolutionary behavior.

The Art Question, Nigel Warburton (2002)
A compact survey of 20th-century approaches to definitions of art by a philosopher. If you want to understand definitions of art, and why no simple definition works (“art is about intent”, “art is about communication”, “art is about ideas”) then this is the book to read.

The Invention of Art, Larry Shiner (2001)
How the concept of “art” was invented in the Romantic era; our modern concepts of “art” would be unrecognizable to, say, Leonardo da Vinci or the ancient Greeks. Unlike other books on this list, I found it often a bit tedious, and skimmed many chapters. but it was really worth it. It might be that the paper that inspired it is good enough (haven’t read it yet). (longer review).

Food and Health

In Defense of Food, Michael Pollan (2008)
I loved Michael Pollan’s The Omnivore’s Dilemma, which compared different approaches to food production, although some aspects of the writing bothered me (like the way he anthropomorphized evolution). In In Defense of Food, he distilled his ideas and a lot of research to a basic dietary credo that affects how I eat today.

Ultra-Processed People, Chris Van Tulleken (2023)
A surveyof the theories, evidence, and the broader context for why Ultra-Processed Food is the source of so many of our dietary problems. (And, some of the original research on ultra-processed foods was inspired by Michael Pollan’s book.)

Trick or Treatment?, Simon Singh and Edzard Ernst (2008)
This book surveys the scientific method for medical knowledge, and applies it to alternative medicine. I had had a long list of confusing experiences and interactions around a variety of alternative medicine practices, and this book put it all into context. I still occasionally tell people about how pseudeoephedrine came from distilling the useful parts of ephedra (used in Chinese traditional Medicene) and discarding the toxic parts.

History

Guns, Germs, and Steel, Jared Diamond (1997)
An attempt to explain why Europe and Asia colonized the rest of the world, and not the other way around, through scientific and anthropological theories rather than racist ones.

1491, Charles Mann (2005)
Most accounts of pre-Columbian indigenous life in the Americas have painted it either as either primitive savagery, or, in progressive fantasies, indigenous populations are wise and perfectly-attuned to their environments. This book provides a more accurate account of what we now know about life before 1492: dense populations and civilizations that rose and fell, just like anywhere else. A lot of our modern fantasies of pre-1492 life as being barely-populated hunter-gatherer tribes were shaped by the fact that 95-percent of the population of the Americas died from diseases brought unknowingly by European colonists, long before Europeans visited most of the affected areas. (I found the middle chapters of this book to be a bit of a slog and skimmed them.)

1493, Charles Mann (2011)
The global effects of the contact of the Americas and Europe. A fascinating account through history of post-1492 science, agriculture, and warfare. So many factoids stick in my head from this.

A Peace to End All Peace, David Fromkin (1989)
All the ins and outs of how the British and French colonialists, when carving up the Ottoman Empire at then end of World War I, created the Middle East and set the stage for seemingly-endless conflict and tyrrany.

Collapse, Jared Diamond (2005)
I only read a magazine-article version of this, not the full book, because I find it too depressing to contemplate.

Season of the Witch, David Talbot (2012)
History of San Francisco, from the 1960s to the 1980s. I find it helpful to counterbalance our current dramas with our historical dramas. The book seems a little sensational; after reading it, I asked my father if San Francisco life was as intense in the 1970s as the book portrayed it and he basically shrugged. Another local-history book I loved was The Mayor of Castro Street by Randy Shilts (1982).

Economics and inequality

Evicted, Matthew Desmond (2016)
The crushing effects of poverty, and the cruel systems that keep people poor. I also got a similar message from reading a magazine-length version of Barbara Ehrenreich’s Nickel and Dimed many years earlier, as well as some articles accompanying the Ferguson riots in 2014, but Desmond’s is more far-reaching. I also loved Emi Nietfeld’s riveting memoir Acceptance, about childhood in foster care and homelessness.

The Myth of the Rational Market, Justin Fox (2009)
How motivated economists invented the myth of market rationality (“markets know best”) that supported the degulation disasters of the 1980s and the financial crashes and consolidations that have happened since.

Weapons of Math Destruction, Cathy O’Neal (2016)
This book was my first introduction to the harms of algorithmic decisionmaking as reinforcing inequality and racism, back in a time of optimism about machine learning. I don’t think she gets enough credit for sounding the alarm on AI, but she was writing when it wasn’t called “AI”, it was called “big data.” (Safiyah Noble’s 2018 Algorithms of Oppression discussed this in the context of Google search, again predating today’s AI hype.)

Chokepoint Capitalism, Rebecca Giblin and Cory Doctorow (2022)
How monopolies have created precarious conditions for artists (and also why copyright alone cannot save artists). This book was written before the current AI-art controversies and provides valuable context.

Other

Non-Violent Communication, Marshall Rosenberg (2015 edition)
An invaluable approach to having more-productive and empathetic conversations, especially in the presence of conflict. (Longer blog post)

Hyperbole and a Half, Allie Brosh (2013)
Hilarious cartoon memoir that, while frequently making me laugh out loud, also helped me understand how absolutely debilitating clinical depression can be, for those that suffer it.

Thanks to people who recommended some of the books on this list: Moshe Vardi, Matt Hoffman, Alyosha Efros, Jitendra Malik.

In this blog post, I summarize some of the basic ideas of NVC, and advocate for the ideas in scientific and technical communication. I describe it in a way that makes sense to me, and might be more approachable for some people; the way the NVC book is written really put me off when I first tried to read it. But, if you find this intriguing, I do encourage you read the book itself; this is not meant as a complete introduction to NVC.

Conflict is a natural part of scientific communication, not only because science involves resolving competing ideas, but also because it is a human process. Yet, I don’t think we’re often trained in how to deal with conflict. NVC offers tools for managing conflict, and, more importantly, communicating what is important to us in a way that prevents conflict in the first place.

I also draw connections between NVC and scientific concepts like “umwelt” in this post.

I think that NVC can be useful to everyone; feel free to ignore the science stuff if it’s not interesting to you.

This post was spurred by a very nasty paper reviews I received recently. In the next blog post, I’ll apply some of these ideas to reviewing, including discussing that nasty review, and describe simple, easy-to-follow guidelines that greatly improve paper reviewing. See that post for an extended example of these concepts.

Non-Violent Communication (NVC)

NVC provides communication techniques for managing and preventing interpersonal conflict. NVC can generally improve empathy and connection between people.

A key concept is to separate what you can observe from your judgments and interpretations, the way that a scientist ideally separates the data they measure from their intepretations of the data.

Think about a recent time when you got angry at someone important in your life, or they got angry with you. Maybe someone was late, or talked over you, or broke a promise—and it made you mad. In these cases, it’s tempting to describe your experience as objective facts about the world: “You’re always late” “That was really shitty of you.” “You didn’t even think about other people.” Or, perhaps you know that saying these things won’t help, so you don’t say anything, and continue to feel bad when the same thing keeps happening again.

The problem with those statements that they they describe your own interpretations of events and judgements of people. They’re subjective. The other person may not agree with them, and even feels insulted by your comments, and then things get worse.

NVC offers a recipe for communication in these difficult situations. A key idea is that you can only talk about things you yourself observe. That is, you can talk about what you saw, what you did, and what you felt.

In NVC, statements about things that you couldn’t observe are off-limits. You are not allowed to make judgements about other people, to say what they thought or believed, or to describe who they are. In NVC, these are all called “violent communication:” a violent communication is any statement that imposes an unwanted opinion onto someone else. Telling someone that they’re usually late, or that they’re rude or inconsiderate are violent communications.

Even compliments can be violent communications. For example, in some contexts, someone telling me that I’m very organized might sound to me like they’re saying that I’m uptight or uncreative. Moreover, in my experience, unsolicited advice is usually violent, since it often conveys a lot of wrong assumptions by the advice-giver about the unwitting advisee.

One goal is to find common ground, which means avoiding statements that the other person disagrees with. If you make judgements or interpretations about what happened—especially about them—they may not only disagree, they may get defensive or angry. On the other hand, if you state only things you observe, the other person almost certainly cannot disagree with them, and then hopefully you have at least improved your understanding of each other. Honestly describing your feelings in these situations is so hard, but so powerful.

NVC is about managing conflict, and just describing your feelings or hurt is not enough; you cannot expect the other person to know what you need. For this, once you’ve stated your observations, you can describe the more basic needs you have, and then make a request: something that you want the other person to do. The request needs to be concrete and actionable; not “treat me better” but something very specific that the person can do “when I cook dinner, do the dishes without being asked.”

In summary, the recipe for NVC in a conflict is: 1. say what you saw and observed (observable facts), 2. how it made you feel (internal facts), 3. describe your personal needs, and 4. your precise and specific request for what you want the other person to do. “You said X, it made me feel Y, I need to feel respected, and I request that you to not say that again.” They might say no to the request, but they should ideally have no reason to dispute the facts of what you experienced and what you need.

This is not easy—it’s not easy to dispassionately observe and discuss intense emotions. It takes a lot of practice and iteration to learn the skills of NVC: to filter out the judgements, to try to observe your feelings accurately, and to openly describe them. I still find it quite difficult.

I have not attempted to fully explain NVC, and I’m certainly not an expert. I recommend reading the first few chapters of the book to learn more. As with any skill, reading isn’t enough: you then need to practice difficult conversations in your own words, ideally together with someone else, and iterate. I promise it’s worth it.

Talking about feelings

At the core of NVC is something that sounds antithetical to science: talking about feelings.

Many of us were trained in childhood—perhaps unintentionally—to avoid talking about feelings and peoples’ needs. In technical, scientific, and pragmatic topics, feelings might seem irrelevant and distracting—one works hard to write objective, factual accounts and to make rational decisions. In interpersonal conflict, it’s tempting to focus on what promises were made, who is “in the right” or what people “deserve”.

If so, feelings and needs end up being neglected, when they’re sometimes what need the most attention. Often conflicts that seem to be about logistics or fundamental disagreements are really about miscommunication leading to hurt feelings. Whenever I see someone get angry in a work meeting (including myself), I believe that there’s a real issue of their feelings of being threatened or hurt, or not getting some basic needs met—not whatever practicalities the discussion is superficially about.

In, say, a faculty meeting, you might see a lot of charged discussion, which indicates a lot of difficult feelings. Some junior faculty feel insecure in their status and some senior faculty feel insecure in their continued relevance, and insecurity leads to people attacking colleagues and faculty candidates. Some faculty seem upset about everything, and this often comes from their own insecurity and fears.

At times, just saying how you feel can really help in tense situations. It feels so difficult to recognize your feelings at the core of a difficult situation, and then to acknowledge them out loud. Then when you do, you think “that wasn’t so hard.” (Actually, it is hard, and I am not good at this, especially in the heat of the moment.)

Several years ago, I was frustrated after receiving several rounds of very highly-critical feedback on a manuscript from a mentor. I couldn’t tell if they thought my work was even worthwhile at all. I finally wrote back that “This project is very important to me, … and I found your email to be really discouraging.” While they did not specifically comment on this, I found their subsequent comments to be much more considerate and balanced, even including the occasional compliment. Their comments and support were absolutely crucial to me eventually getting the work completed and published.

One of my favorite comedians, James Acaster, tells stories in his “Hecklers Welcome” tour about confronting conflict and fears. The show culminates in a story of how he disarmed scary bullies on a train by honestly describing how scared he was. It seemed like a textbook use of NVC (although rewatching it now, he doesn’t talk about needs or requests). (At the theater where I saw it in San Francisco, he complained that the hecklers were too nice, which got a laugh.) It’s a very funny show and I highly recommend the comedy special. (I also loved his prior special for different reasons, alternate link).

Even in everyday conversation, we can include discussion of feelings. These are little things: “I’m happy to see you” “I’m disappointed that they’re out of dessert.” But I think a willingness to mention feelings in everyday situations is good practice for more difficult situations; conversely, people who find this difficult might also have similar difficulties in difficult conversations.

Judgments as opinions

NVC says to avoid all sorts of external judgements. But, in everyday conversation, we make judgments all the time, like “The show is funny” or “That’s a good idea.” These judgments are generally statements of opinion, not facts, and different kinds of judgments can function in different ways. When I said that “Hecklers Welcome” is a funny show, what I really mean is that I laughed a lot and enjoyed it, and I think that other people would find it funny. That is, I used an objective-sounding judgment about the show as a shorthand for my own opinion of the show.

I think of most judgments as being shorthands for opinions. Explicitly expressing every opinion as an opinion would be painfully verbose and unfun.

Trouble arises when people treat such judgements as absolute. If you insist that a show is funny—contradicting anyone that says otherwise—then you seem to be saying there’s something wrong with people who didn’t laugh at the show. For this reason, I try avoid such judgments when there’s potential disagreement. For similar reasons, I advocate against evaluative definitions of art (“This is good enough to be art”), and that unnecessary judgements impede creativity and artistic practice.

One alternative is to replace generic judgments with specifics. Instead of “that’s a good movie,” or you comment on aspects of it, e.g., “I liked the acting but not the special effects.”

Interpreting sensations: umwelt and NVC

Now I turn to how NVC relates to scientific theories.

It seems like we experience the world as it is: you see a picture and know what’s in it. A stranger is aloof and unfriendly, and refuses to make eye contact with you. A dog sniffs some boring weeds but misses a spectacular view from a trailhead. Much of the time, these kinds of descriptions work.

But perceptions can mislead us. Visual illusions are not what they appear. Someone who seems aloof and unfriendly might just be shy. A dog experiences a wealth of sensations from smells that we humans cannot experience, whereas dogs are red-green colorblind and overall have worse visual acuity than humans.

This is because we don’t see the world as it is—we interpret our sensations. In biology, this idea is expressed by the concept of umwelt: no organism has access to the true facts of the world, it can only access its senses. Many of us organisms interpret our sensations. Different organisms have different experiences—me and my dog have different senses, and so we have really different experiences when we’re out on a walk.

Scientific research has analogous limitations. A scientist running an experiment cannot know the true causes of what happened in the experiment; they can only take measurements and form theories based on them. A scientist aims to report, as carefully as possible, what they did, and what they measured, and then to cautiously describe their interpretation of the results. Conversely, some cognitive-science theories liken organisms to scientists, measuring the data of their senses, performing experiments, and forming theories in their head, such as “Bayesian brain” theories and Gopnik et al.’s The Scientist in the Crib.

Here’s the complete analogy between umwelt, scientific communication, and non-violent communication. In each, there is an actor (an organism, a scientist, a person), who can observe specific things (senses, experimental data, conversations and feelings). Organisms and scientists form interpretations of their data (the state of the world, scientific theories). In communication, scientists report their data and their interpretations, at some dispassionate remove (separately their feelings from the data), in order to come to a shared understanding of science; people using NVC report their direct experiences dispassionately (including reporting on their feelings) in the hopes of coming to shared understadning.

This is just an analogy, not a literal equivalence; communication would be impossible if we always had to report every sensation on our retinas instead of “the clock said 3pm” or just “it was 3pm.” We can take shortcuts when there is shared understanding, and no real uncertainty about the meaning of our sensations.

Scientific communication typically involves a posture of objectivity, that one can make statements without a point-of-view or positionality. Even if we acknowledge that objectivity is technically impossible, the key idea here is that we always have some shared understanding. We can treat shared understanding as facts (no need to treat “the Earth is round” or “we went to a movie together yesterday” as opinion), and be more careful about uncertain or contested ideas.

Using NVC in scientific and technical communications

Having tried to incorporate NVC into my personal life—it’s difficult, but worth it—I find myself using those habits in some of my writing as well. I don’t advocate employing the full NVC recipe in all scientific and technical communication, but I advocate using some of the ideas. At the very least, one should be very careful with judgments, acknowledge feelings where relevant, and, in debates, avoid violent communication.

I’m currently co-authoring a technical paper on a controversial topic in “AI” art, and a draft used judgemental, emotionally-charged language. I found the language upsetting (I disagreed with some implications of it), and it did not seem possible to concisely summarize the issues—any judgments we made would likely be upsetting to some group. But we also didn’t need to take a position or to make these judgements. I reframed the introduction in factual terms that we could support with citations, so that we could motivate our technical problem without making controversial assertions. This might sound just like careful technical writing, but to me it felt like NVC, since we avoided judgmental language and limited ourselves to factual observations.

Now, when I do discuss controversies directly, I do try to acknowledge the conflicting fears and angers of the different stakeholders, and the sources of those feelings. For example, in a discussion of “AI art,” I discussed the anger that traditional artists feel as a product of their fears for their own livelihoods and identities, many of which are quite understandable. (Although I’ve also discussed it in terms of the value systems of different kinds of artists, which are themselves tied to feelings.) Some artists might find this discussion violent if they do not relate to it—different people have different opinions and feelings—but it seems better than not attempting to acknowledge the feelings.

In my next blog post, I discuss violent communications in paper reviewing (including some really bad reviews I've gotten), and describe a simple guideline for better, non-violent reviews

Many thanks to Rich Radke for comments on this blog post.

This follows my previous blog post, where I reviewed “non-violent communication” (NVC). NVC offers a way to communicate and find common ground by separating observable facts and feelings, while avoiding judgments of other people. For this discussion, a “violent communication” is any statement that imposes an unwanted opinion on someone else, like telling someone that they’re always late or rude. As I discuss in that blog post, even compliments and unsolicited advice can be non-violent in some contexts. Additionally, I discuss the importance of talking about feelings, something many of us are trained to avoid, whether from childhood or scientific and technical training.

Even if you’re not specifically interested in scientific paper reviews, this post may provide an interesting discussion of how thinking about feelings and communication can be useful in a technical context.

Here, I begin by describing two “violent” paper reviews that I’ve gotten in the past few years. The second one was especially nasty, and spurred this blog post. In this discussion, I’ll particularly focus on feelings here—mine, and those of the reviewers. In a few places, I will speculate on the reviewers’ feelings and identities in order to make my points, and this would constitute violent communication, and so would not be conducive to finding common ground in a discussion with these reviewers.

My first toxic review

Since 2019, I’ve been submitting papers to journals in vision science (i.e., the psychology and neuroscience of human vision), far outside of my background in computer science. Overall, my experience has been extremely positive, both in the review process, and with the vision scientists who have given me feedback and guidance on manuscripts. Reviewers have often misunderstood my submissions and written skeptical reviews, sometimes even leading to rejection. This is normal and useful: these misunderstandings give clues about problems in the paper, which I can then fix. But there have also been some toxic reviews.

My very first vision science submission got one violent review, based on misunderstanding the manuscript. Here’s a typical paragraph from it (typos included):

Here, the anonymous reviewer has failed to understand the argument in the paper, and then made wrong statements about my argument. The reviewer writes their wrong judgments with total confidence.

Moreover, the review refers to me by name, and calls my conjecture by my name (I had not given it a name in this version). This make the entire evaluation personal—they make it about me—for no apparent reason.

This reviewer concluded with some advice:

Perhaps the reviewer thought they were being helpful by giving a compliment and advice. But, in so doing, the reviewer implies that I am ignorant of the literature, that I do not know how to choose research problems, or how to write a paper. What sounds like well-meaning advice is offensive and insulting. I want to say back: who are you to tell me what to do?

The paper was rejected. I revised it based on the questions and misunderstandings the reviews, and resubmitted it a month later. After a few more rounds of revision, the paper was published in Perception in 2020. I did not follow any of that reviewer’s bad advice. The fact that the paper was published feels vindicating: the reviewer was wrong about the core of the paper, and wrong about me.

But the review should not have been about me.

It’s normal for a reviewer to not understand a manuscript’s argument; seeing how readers misunderstand a paper is invaluable information for authors. I work very hard to understand reviewer comments and I heavily revise papers accordingly.

But it is not okay that the reviewer judges me, personally. This is violent communication. Not only were their comments inaccurate, they are toxic: they hurt, emotionally. They are insulting, they are painful to read. They make me want to respond with my own violent communcation—to say that the reviewer is an asshole—and to discount everything they say.

Violent communication is counter-productive: it’s hard to learn much from someone when they are hurting you. (This is also one reason that dog trainers have moved to training based on positive reinforcement instead of negative reinforcement.)

Toxic communication harms the community. When editors allow these kinds of comments, they send the message that this kind of reviewing is acceptable, even encouraged. Toxic communication poisons open communication and can even drive people away, especially students. I feel insecure about my ability to participate in vision science research, and feel like an outsider.

Authors need to feel safe to take the risk to submit papers. If, when I first submitted vision science papers, all the reviews had been like this, then I surely would have given up writing vision science papers entirely.

My second toxic review

My coauthors and I recently received a far worse review from Journal of Vision:

That is, they called us ignorant and described our beliefs. Later in the review they continue to describe our mental states and beliefs. They later commented directly on our training:

The reviewer concluded with:

Here the reviewer again describes us as ignorant and makes baseless about how the paper was written. Moreover, the reviewer says we should feel bad.

This review made us very angry and upset, and we are still mad and hurt. Even writing this blog post has made me upset.

I shared the review with some vision scientists colleagues. They sympathized, and said they’ve received reviews like this too. They speculated that these kinds of reviews come from scientists who did a lot of psychophysics in the 70s and 80s and now feel left behind by the field.

We have a ton of work ahead of us, to revise and overhaul the work, making use of the information in the reviews. We will certainly try to prevent these misunderstandings in our revisions. But it’s so much more difficult to study this review when it makes us angry and upset. As I said before, it’s hard to learn much from someone when they’re hurting you.

The reviewer made many useful criticisms; our submission had many problems. But that doesn’t explain the intense emotions that the reviewer reports. The review gives the impression of someone suppressing rage.

Here’s my theory: the reviewer misunderstood one of our claims about the literature—they thought we were saying that no one had worked on these problems. They took it as an insult to their work. Instead of recognizing their hurt (or questioning their interpretation, or even noticing that we actually did cite papers and books from this literature), they lashed out at at us. The genuine flaws in our manuscript just added fuel to the fire. But that’s just one possibility.

The harsh reviewer might think that they’re justified in being harsh—as if having a paper rejected, and seeing that it needs a major overhaul, isn’t negative feedback enough. I could imagine legitimate arguments that some papers should not be submitted. But, if that’s what they believed, then the reviewer should have made those arguments, instead of just giving emotional abuse.

I have speculated about the reviewer’s emotions, state, and background in order to make a point. In an actual conversation, this would all be violent conversation, and thus not constructive to finding common ground.

A reviewer feels threatened or insulted by a paper, so they write an angry review attacking the authors. The reviewer has bad emotions and tells the authors that they should feel bad. The reviewer discussed the manuscript in terms of the authors’ thought processes, thereby making the reviews unnecessarily personal, and sometimes insulting (especially when they were wrong). The authors in turn feel angry and upset by the review and then struggle to incorporate the feedback and figure out the next steps. This does not seem healthy for anyone.

Reviewing guidelines

I believe that scientific reviewing should include guidelines to prevent these kinds of reviews. The most simple principle is:

The reasoning is simple: reviews are judging the paper for acceptance. Nothing about the authors is relevant to this decision. Moreover, directly discussing the authors makes the process personal, and can often be insulting to authors. Even compliments can come off as personal judgments.

Most places that I have published do have explicit rules about this. For example, the SIGGRAPH policy states:

(I might have contributed to this, I don’t remember). The CVPR policy uses almost the same wording. These policies are enforced by the papers committee: if a review violates the policy, the reviewer will be required to fix it before it is sent to the authors.

This seems like it’s obviously the right thing to do. The primary purpose of paper reviewing is to decide whether to publish the paper, and the secondary purpose is to give useful feedback on the paper manuscript to the authors. Discussing the authors does not further either goal. Moreover, the reviewer does not have enough information to say anything useful and meaningful about the authors. Such statements would just end up being violent communication.

Moreover, I do think that this rule leads to better reviews because it reminds reviewers to focus on the paper and its contents, and not get distracted by the authors’ identities and judging their competence. (Blind review is even better for this.)

The whole point of peer review is to judge a paper on its merits, not the authors’ supposed merits.

For as long as I can remember, I have always followed a stricter rule for my own paper reviewing:

I don’t ever remember seeing violent reviews in computer science, and I don’t remember ever having to enforce the rule myself as an Area Chair/Editor. I think that simply having the guideline creates a more civil review culture.

One corollary I follow is to not address the authors directly (“you”). I often see reviews written in second person (“you did X, why?”). This, too, feels unnecessarily personal, mismatched to the primary purpose of the review: to inform the editor/committee/reviewers about whether to accept the paper.

Obviously, making review processes double-blind would also improve these reviews (as they are in CS), non-blind reviewing has been shown to create biased reviews in many studies, e.g., 1, 2). It seems bizarre to me that there remain some non-blind reviewing processes.

Exceptions can be made for certain cases, e.g., opinion and perspective pieces often reflect an author’s viewpoint and their identity may be relevant for these.

Reviewing style

As a reviewer, how can you best implement these guidelines? As a reviewer, I find it very easy: instead of “the authors say X”, I write “the paper says X.” I use passive voice when needed (e.g., “in the first experiment, participants were asked to …”) And, of course, I avoid any temptation to mention the authors’ own motivation, knowledge, background, and so on.

For example, I might restate the initial quote from my first toxic review (above) as:

There’s also some added uncertainty here—maybe I misunderstood the paper? It’s a lot more likely that a manuscript is unclear than that the authors genuinely mean something nonsensical.

If you really feel compelled the discuss the authors in reviews, then it’s worth seriously examining why, and whether doing so is actually beneficial.

This advice does contradict other advice I would give, like avoiding passive voice. In this case, I think avoiding violent communication is more important. Also, when writing about a paper that is published, I might mention the authors as saying/writing whatever they claim; it feels different when it’s a published paper versus an unpublished manuscript. My only rationalization for this is that a manuscript is just a draft, and the power relationship is different between an anonymous reviewer and an author of a submitted manuscript, versus the author of a published paper and someone commenting on it afterward.

One other factor is that I sometimes write theoretical papers with some first person in order to avoid passive voice (including that Perception paper), which might encourage reviewers to discuss the author, but I still argue that the review should be able the arguments.

Reviewing and Non-Violent Communication (NVC)

The core argument of this blog post could be expressed in terms of the four steps of NVC:

1. What I observed in the paper reviews I received
2. How I felt after receiving the reviews
3. My needs. I need to feel “safe” from attack when submitting work for publication, and I believe other authors do too.
4. My request: create and enforce reviewing guidelines not to mention authors in reviews.

I also claim that the rule leads to better reviews and better reviewing culture overall.

I am not advocating the full use of NVC in reviews. In this post I have discussed NVC as a way of understanding (a) how one’s feelings influence the process, and (b) what not to include in a paper review, just because it becomes counter-productive.

I think similar ideas can be helpful in other kinds of scientific and technical communication as well.

Thanks to Rich Radke and Maneesh Agrawala for comments on this blog post, and to the vision scientists who have given me advice and commisseration on submitting to vision science journals.

During my recent sabbatical, I painted a picture that, finally, can illustrate some of the decision-making that went into it, because I remember enough of the decision-making that went into it, and there were enough changes of direction to be interesting. This isn’t my favorite painting, but hopefully sufficient to demonstrate the circuitous nature of painting.

The main point here is: I only started with a vague idea of what the painting would look like, and I changed my mind about it several times during the process. Moreover, the final style was unplanned, but influenced by paintings I’ve seen in the past.

I did it while looking at this view out of a hotel window:

Here’s the full timelapse of the finished painting that I want to discuss:

On the previous day, I’d drawn this sketch:

As a subject for the picture, the building naturally drew my eye. On the second day, I had the same impulse to center the building. But I realized that I didn’t want to do that. The building is not that interesting, it’s the open space that’s interesting. So I made a conscious decision to focus on the open space.

So I started by drawing the horizon and the building:

But even after this I decided to move the building further to the side. So I moved it and then started painting the beach:

I noticed that I was drawing simple, relatively clean strokes, with the horizon being a set of stripes. I got interested in the idea of doing the whole painting like that, similar to, say, a Georges Schwizgebel animation or Diebenkorn: solid geometric regions, rather than the more sketchy stuff I normally do. Maybe I’d do the whole painting with this simple, geometric look.

The then I started filling in more of the mid-ground region. I composed three main elements in the center, selecting from the real scene and distributing them horizontally on the canvas:

Then I added some texture, enough to get a sense of what it would look like:

It would have been easy to just spend all my time filling in those details. But at some point I realized it was enough for now—time to move to the foreground to finish the initial composition.

The real scene had a lot of texture and detail in the foreground. Of course I couldn’t draw all of it. I drew rough versions of some of the shapes in the real scene:

Then I thought I’d draw lots of detailed grass texture, lots of little blades of grass:

This is a really different style from the simple geometric idea I’d had a few minutes earlier.

But after awhile, painting all these blades of grass seemed too time-consuming and not worth it. So I pressed “undo” a bunch of times:

Then I started to use a much larger brush to draw the grass texture:

As I started to draw these textures, I became conscious of how different it was from what I was seeing. But it also reminded me of some David Hockney iPad drawings I’d seen. This gave me a sort of “permission” to continue: the feeling that maybe this was something to expore rather than avoid.

Then, the rest of the time was just filling things in. Here’s the final painting:

Here’s the portion of the view that contains the elements of the painting:

In the painting, you can see that elements of the view are arranged differently in the canvas than in the photo. The painting contains a few different styles from different phases of the process: the relatively abstract beach and sky, the moderately-texture mid-ground, and the most-textured foreground.

The final painting has more distinct shapes, more like distinct objects, like caterpillars or single-celled organisms or paisley, or more like abstract art shapes. None of this was planned. This style was definitely not intentional or planned, it’s something I stumbled upon, and I wouldn’t have had I not spent time looking at David Hockney paintings sometime in the past.

None of these choices, or the styles within, were planned.