Philosophy dabbler; anime-and-manga ace.
I am writing this because I now plan to become active on Zhihu in earnest. Before I do, I need to establish a few starting points to help both my readers and me get our bearings: what do I think of current online and academic research and criticism on form and style in manga and animation? To what extent, and in which respects, do I build on or criticize it? Yet since opening my column in 2018, I have done nothing to develop it. Apart from an essay on VTuber culture that Tata helped post on my behalf, and “Three Dimensions of Slow Animation,” published at Lab on Roof, I have not written anything on Zhihu. I therefore need to write a few essays both to answer these questions and to set the tone for the column’s current series.
By a remarkable coincidence, just a few days after I began thinking about these questions, Shanmao (@山猫嗷嗷嗷) posted an article by Izumi Nobuyuki (hereafter “Izumi”) entitled “The Basics of Gaze Mechanics in Manga.”
Shanmao then set out his own views on manga scholarship:
“Scholarship does not come into being just because you type the word ‘scholarship.’ Scholarship means systematic, specialized learning. Only when there is a system, when manga and its regularities are organized into a discipline, can we speak of manga scholarship. Scholarship involves many people: all those engaged in it explain things together as objectively as possible and check one another’s claims, striving to come as close to objectivity as they can. One person alone cannot constitute scholarship; scholarship is built together. It must be empirically grounded, systematic, and objective.”
Reading Izumi’s article, I could not help frowning, but I also thought it offered an excellent point of entry. Izumi Nobuyuki has by now made something of a name for himself in Japanese manga scholarship, and the errors in his “The Basics of Gaze Mechanics in Manga” are quite representative. This settled the subject of the present essay: by analyzing the various problems in Izumi’s article, I will use this particular case to introduce some of my views on current animation and manga research, and set out in full my position on what “empirical grounding, systematic organization, and objectivity” mean in the study of form and style.
Still, is it appropriate to call this essay “A Critique of Izumi Nobuyuki” when it analyzes and comments on only one article, “The Basics of Gaze Mechanics in Manga”? I can already imagine someone asking, so let me answer in advance. First, “gaze mechanics in manga” is one of Izumi’s principal theories—his “signature work,” you might say—and even now he still trots it out from time to time. Second, I believe that after reading this essay, readers will understand very clearly why this is enough to form a judgment of Izumi himself. In fact, if you really want to learn some manga basics, I suggest spending more time with the relevant books by creators and scholars such as Oran Zhu (Oran猪), Scott McCloud, or Ōtsuka Eiji. You can also read Shanmao’s articles, of course—particularly his highly upvoted answer:
How Can I Learn Manga Panel Sequencing? zhihu.com
I benefited a great deal from this answer myself when I was learning the theory. As for Izumi, my advice is not to spend much time on him.
(The full essay runs to about twenty thousand Chinese characters. If you do not want to read a long and entirely unconstructive catalogue of errors, you can skip straight to the summary and argument in Part II.)
Part I
In “The Basics of Gaze Mechanics in Manga,” Izumi gives a detailed account of a visual-perceptual phenomenon in manga reading, describing it as a perceptual “current” flowing from right to left. Following the order of his article, he first argues that manga reading produces an effect whereby movement from right to left accelerates, while movement from left to right decelerates. Drawing on this property and on specific manga sequences, he then analyzes the current’s important influence on how manga is understood:
“(A character) facing left: the reader’s (protagonist’s) viewpoint; an advantageous position (the upper hand); control of the scene; taking the initiative (active); known and familiar; the reader can empathize directly with the character’s feelings; and so on…
(A character) facing right: an object seen objectively; a disadvantageous position (at a disadvantage); being controlled; responding (passive); unknown and unsettling; the reader imagines the character’s feelings indirectly; and so on…”
A character being on the right (and facing left) or on the left (and facing right) thus becomes a crucial binary running through the article. Readers can find the bulk of the fuller explanation in the section “The Relationship Between ‘Eye Guidance’ and ‘Gaze Mechanics’” and at the beginning of “Positions Divided Between Left and Right”; I will not go into it here. Below, I will refer to these claims of Izumi’s as the “speed theory” (faster from right to left, slower from left to right), the “viewpoint theory” (perceiving subject on the right, observed object on the left), and the “strength theory” (strong on the right, weak on the left).
Selective Omission
Let us begin with some simple problems.
Izumi uses specific sequences from Negima! to explain all three theories. Yet we will find very serious problems of selection in his use of each sequence.
In the fifth section of his article, Izumi uses a fight sequence beginning on page 104 of Negima! volume 8. He tries to show that Akamatsu Ken reverses the two characters’ left–right placement in the image several times to express changes in their relative strength as the scene develops.
In Izumi’s account, the boy initially has the disadvantage in his fight with the man in black, so the man in black is on the right and the boy on the left. In one panel during their exchange, the boy is suddenly placed on the right of the image: their positions are reversed. “After this, the boy is shown recovering his ‘will to fight’; once he returns to normal, (the boy) immediately begins his counterattack.”
The example given in “The Basics of Gaze Mechanics in Manga” undoubtedly fits the “strength theory.” But when we actually open the manga, we find a very different picture:
Between the manga excerpts Izumi reproduces—the two red boxes—he has selectively omitted exactly ten panels. Why leave out ten panels? Because they are unimportant to his explanation? No. They are precisely what matters.
In Izumi’s account, the characters’ left–right positions are reversed in box 1, placing the boy on the right. Then, in box 2 and the action that follows, the boy recovers his will to fight and begins his counterattack. But look at Figures 3 and 4, and you will find that no such continuous process exists in the manga. What is more, three panels after box 1, the man in black kicks the boy flying while the boy is on the right of the image. Until this point, he has just about managed to fend off the attacks and strike back; this time, he finally fails to defend himself against the man in black. This is precisely the first time in the fight sequence that the boy is clearly at a disadvantage—the exact opposite of the “strength theory.” Next, Izumi uses the crop in box 2 to explain that the boy is placed on the right to accompany the recovery of his fighting spirit. Yet in the third and fourth rows of Figure 4, the boy is on the left, while the man in black is placed on the right. Through Izumi’s selective cropping (Figure 2, right), a boy who originally appeared on the left of the image has come to appear on the right. Of course, one could say that Izumi is talking about the direction the characters face, whereas I am talking about the left–right arrangement of several panels in the page layout. But notice that in the third and fourth rows of Figure 4, both the man in black and the boy face left. How is this to be explained by the “strength theory” or the “viewpoint theory”? What is much more intuitively apparent is that, in both rows, the man in black’s panels occupy a far greater share of the image than the boy’s. The boy’s panels are squeezed over to the left. Who controls the scene is self-evident. Izumi’s selective omissions have produced a grossly inaccurate explanation.
Once we actually read the manga, we find that the real function of reversing the characters’ left–right positions does not match Izumi’s explanation at all. Between box 1 and box 2, he says, the boy regains his will to fight and begins his counterattack, with the characters’ placement working in concert. But put these panels back into the original sequence, and we find that the reversal in box 1 serves two purposes:
-
It helps create a change of rhythm. After throwing several quick punches, the man in black sends the boy flying with one heavy punch. The manga changes the viewing angle to work with this shift, while using the composition to bring out the weight of the blow. (Imagine simply carrying the second row’s composition over into the third: it would be difficult to make this heavy punch stand out.)
-
It connects the panels. In the fourth row of Figure 3, the boy leaps up immediately after being knocked to the ground, gathering his strength as he charges at the man in black. In the third row, he is knocked to the right, and the impetus of this movement pushes our eyes directly toward the first panel of the fourth row. The reversal in the third row is precisely what connects this rightward movement with our eyes as we read.
Izumi goes on to say that the boy successfully lands a fine blow during his counterattack. Yet their left–right positions in the image are subsequently reversed again, precisely because “in fact, this attack did not decide the outcome; afterward, the boy again loses control of the fight, and signs of his defeat appear.”
But open the manga, and once again we find that Izumi has selectively excerpted it:
What is missing between the two red boxes in Figure 6 is a series of quick blows from the boy; Figure 7 is missing four panels. Put the contents of the red boxes back into the manga, and once again the expressive function of the left–right reversal is completely different from Izumi’s explanation. The reversal in Figure 6 serves: 1) a change of rhythm—the shadow clones launch a succession of attacks in the second row, culminating in a decisive blow in the third; and 2) the portrayal of the boy getting the better of his opponent by surprise. The shadow clones’ attacks in the second row were all preparation: after successfully confusing the man in black, the boy actually hits a vulnerable spot in the third row. This by no means foretells the failure of his counterattack, as Izumi claims. On the contrary, the blow is unquestionably successful. Readers may notice that the reversal serves a function very similar to the technique we analyzed in Figure 3. This is indeed a technique Akamatsu Ken is using deliberately. But it has absolutely nothing to do with Izumi’s theory.
Through Izumi’s “re-editing,” a manga sequence that did not fit the “strength theory” has been changed beyond recognition, becoming a perfect illustration of his theory.
This is undoubtedly a very serious accusation, so we must also consider whether that is really what has happened.
First, it is plainly difficult to believe these omissions were inadvertent. There is no reason Izumi could omit so much of the manga while just happening to notice and excerpt a small portion of it to explain his theory. If he made a conscious selection, then, does that mean the material he did not select is unimportant? Yet whenever we put his excerpts back into the original panel sequence, we find that their actual functions fail to match his explanations, or even contradict them.
Another possible defense is that I have misunderstood the “strength theory,” or that even where a panel does not fit it, we can still explain the panel within Izumi’s theoretical framework. For example, between the boy preparing to counterattack in Figure 3 and being kicked flying in Figure 4, he initially occupies the stronger position on the right, facing left, yet is struck while still on the right—precisely to catch us readers off guard. Or again, the boy’s thoughts appear in Figure 4 after the left–right reversal, which fits Izumi’s “viewpoint theory” exactly: readers take up the boy’s viewpoint when he is on the right of the image or individual panel. Similarly, we could say that the reversal in Figure 6 also fits the “strength theory,” because, just as Izumi says, the boy subsequently fails to unleash his finishing move and is defeated by the man in black. The reversal already foreshadows this. And the boy’s position on the right in the second row of Figure 7 is simply the artist giving us the false impression that he is about to succeed. He fails to unleash his finishing move: another “feint” by the artist.
My response has several parts. First, if these selectively omitted portions do indeed fit Izumi’s theories, we still need an answer to the question: why did he not excerpt them? Is it, for instance, that the “viewpoint theory” has nothing to do with the “strength theory,” so he chose not to discuss it while explaining the latter? Yet he still analyzes the viewpoint function of the first panel in Figure 6—and gets it catastrophically wrong, as I will explain below. Is it because reproducing everything would take up too much space? Yet we will also see below that Izumi is perfectly willing to reproduce an entire manga page even when it offers no obvious support for his theoretical explanation.
Second, the defense above would greatly reduce the scope for falsifying Izumi’s theories. When a panel does not fit his first theory, we try the second to see whether it fits that one. When the events in a panel do not fit the theory, we say it sets up and foreshadows something an unspecified number of panels later, and therefore does fit. After such maneuvers, it naturally becomes very difficult to say Izumi’s theories are wrong: we have turned direct counterexamples into indirect supporting examples. Whatever happens, he is right—so how could he ever be wrong? Indeed, a close examination of his whole article shows that these are not merely defenses I have made up. “The Basics of Gaze Mechanics in Manga” is riddled with strategies for eliminating falsifiability. This rapidly takes him far away from ordinary scientific propositions as they function in practice, to say nothing of objectivity or empirical grounding. I will criticize this further below.
Third, even if we believe Izumi’s selections are justified and the portions he did not select also fit his theories, problems with the theories themselves remain. For example, if the “viewpoint theory” and the “strength theory” can operate simultaneously, and can also switch back and forth, how is this switching possible? Does it take place over a long or a short period? How do the effects of viewpoint and strength influence one another? When a panel fits theory A but not theory B, why does A operate, and why does B not? When it fits both A and B, which operates—or do both? And why? Looking through the article, we find not only that Izumi has utterly failed to answer these questions, but that the same questions recur and grow more acute later on. He has no theoretical framework integrating his various concepts and claims. His multiple theories do nothing but make his assertions less falsifiable. We will return to this point below and find an alternative hypothesis—at least a provisional one—amid Izumi’s chaos.
In fact, to take a broader view, when we actually read the Negima! series, we find that Izumi’s choice of these particular sequences as evidence, rather than the many others that appear not to fit, is itself a form of selective omission. Go back a step further, and his reasons for choosing Negima! rather than another manga are not sufficiently persuasive either. (I will analyze this in detail in Part II.) To be sure, he offers a little explanation at the end of his article: beyond gaze mechanics, “still more and more complex techniques combine and intersect, making manga expression more complex, more varied, and continually expanding it.” But this does not explain the various errors in his article. It is merely a patch added to fend off readers’ questions.
Let us move on to the next problem.
Single-Cause Attribution, Misattribution, and Misidentification
As Izumi says, manga is complex.
Even in the simplest panel of a manga, different factors always intertwine and work together to influence our cognition. So when we reason backward from effect to cause, we must be particularly careful: precisely which factor or factors have registered in our minds? How do they influence one another? Among all the possibilities—composition, drawing style, text, panel organization, mode of reading, and so on—which play a part, and which dominate? This is a recurring problem in the analysis of form and style. How does Izumi deal with it when he encounters it?
Let us begin with this example from page 155 of Negima! volume 9. Izumi analyzes the first panel as follows:
“In the first panel, an airship is heading right, exactly against the ‘invisible current,’ so we can feel how slowly it is moving. It is like a river fish swimming upstream that appears to remain still in the water.”
Then, in the very next sentence, he says:
“Moving on to panels 3 and 4, the characters walk from right to left with the ‘current’, and the reader’s eyes likewise move gently along, as though ‘flowing’ with it, as the story unfolds.”
When I first read Izumi’s article, this passage left me speechless. I wonder whether readers can immediately appreciate its little subtlety. It is particularly worth noting that the manga page in Figure 8 is the first concrete example in “The Basics of Gaze Mechanics in Manga.” After some abstract deductions about cognitive processes at the beginning of the article—with no rigor whatsoever—this is the first actual example Izumi uses to explain his so-called “gaze mechanics.”
I can only ask: what exactly does this example demonstrate? In Izumi’s account, the airship in the first panel moves against the current produced by visual perception and the reading process, and therefore looks slow. In the third and fourth panels, the characters walk with the current, and they look slow too. What part, then, does the current play?
More ironically still, in an early version of Izumi’s analysis of this example, we find the following:
Even after excerpting portions of the manga panels, Izumi still reminds us in red to view them as a whole, in right-to-left order.
We can establish three things from this passage. First, “rightward-facing airship” and “moving slowly” are emphasized. In other words, Izumi really does believe that the airship looks slow precisely because of his so-called “current,” and he thinks this well worth pointing out. Second, his view did not change between the early blog analysis and the subsequent publication of “The Basics of Gaze Mechanics in Manga”; hence his copying the illustration and analysis, unchanged, into the article for publication. Third, even after cropping the manga into excerpts, he still failed to notice that, quite apart from his so-called gaze mechanics, the airship already looks slow in itself.
We can find all kinds of possible reasons why the airship looks slow: the image has no symbolic cues for speed, such as speed lines; readers already have knowledge about airships; the airship’s central position in the panel creates a balance of forces; even in real-life experience, objects in the sky appear motionless because of their great distance and the lack of reference points; and so on. Izumi completely ignores these highly plausible causes, which are clearly more likely to dominate the effect. Instead, he supplies just one explanation, almost impossible to judge right or wrong, but consistent with his theory: the airship looks slow because it is heading right, “against the current.”
His subsequent explanation of panels 3 and 4 makes it even clearer that going “with the current” or “against the current” either has no influence on (the sense of) speed, or has only an extremely slight one. If an object already looks slow, it will still look slow whether it goes “with” or “against” the current.
We need not say that these two adjacent explanations contradict one another—someone will surely think they are two unrelated applications of technique. Nor need we assume that Izumi is unaware of other factors’ effects on the perception of visual dynamics, since that too would be rather hard to believe. But if so, why on earth did he choose this image as his first example of “gaze mechanics in manga”? As I have said, all we can see is that his so-called “gaze mechanics” has had almost no influence in these panels, or perhaps none at all. At the very least, this is not a good example of so-called gaze mechanics.
Next, the last row of Figure 8. Observing that the movement depicted in each panel is incomplete, but that readers can mentally fill it in, Izumi goes straight to his conclusion: “This ‘sense of movement in a certain direction that can be read even though it is not actually drawn’ is presumably an effect produced by gaze mechanics.”
Although he says, “Let us analyze this more closely,” two paragraphs of description lead only to an extremely crude, rudimentary analysis. A highly skillful panel sequence is reduced to a matter of gaze mechanics in which no creativity at all is discernible.
In fact, this sequence contains many interesting details waiting to be discovered: the way the composition of the second panel in the last row blocks our gaze to the left; the very clever way our gaze is guided between panels; and, most importantly, the way the focal points of the focus lines in the second and third panels deliberately direct our eyes toward the place the moving figure has already left. These are the real reasons the sequence creates “a sense of speed the eye cannot keep up with.” It is difficult to say what role Izumi’s so-called visual mechanics, his “invisible current,” plays here. The third panel, for instance, would clearly still give readers an intuitive sense that the moving figure came from the right even if it were taken out on its own. This is an immediate, unthinking inference backward to the direction of movement. And is our mental filling-in in the second panel caused by the “current,” or by the actual leftward movement of our eyes? I believe the latter is the cause.
In his subsequent analysis of the example in Figure 10, Izumi’s carelessness in attributing causes is laid completely and unabashedly bare. This time, at last, we see him find a second cause for the different impressions of speed in the first and fourth panels: the presence or absence of motion lines. (Then why on earth did he never mention the other factors influencing cognition in the images before?) Yet, sure enough, he shows not the slightest interest in how much the motion lines contribute, or in how much his “current” contributes in comparison. Instead, he leaps straight to a truly remarkable conclusion: the use of motion lines shows precisely that the manga artist, too, is spontaneously making use of this “invisible current.” Faced with a case in which motion lines are very likely to dominate the impression, Izumi needs no argument, no experiment—one could even say no analysis—to turn it into “I carried Kobe to eighty-one points.”1
By now, I imagine readers have a fairly concrete understanding of Izumi’s analytical practice. So I will dispense with further preliminaries and go straight to my conclusion. Although the central purpose of this article of his is to describe a kind of visual perception in manga, what he calls “gaze mechanics,” I believe that he probably lacks the corresponding grounding in visual-perception theory and knowledge of drawing. Or perhaps he knows something about these subjects but cannot identify the crucial structures and apply that knowledge in analysis. This may be the real reason every analysis of his results in single-cause attribution, misattribution, or a simple failure to recognize important elements and techniques in the image.
There is ample evidence for this judgment. For example, Izumi seems completely unaware that, rather than pointing to his supposed acceleration and deceleration, his claims about “gaze mechanics” point more often, and more readily, to visual-perceptual concepts such as balance, openness, closedness, and directed tension—even though he often uses the word “vector.” Again, in his early articles, we find that he makes extensive use of horizontally mirrored images to explain the effects of directionality, a practice that persists even in this article. Yet he does not know that, because visual perception distributes forces differently between left and right, an asymmetrical but balanced composition always becomes extremely unbalanced when mirrored, making this procedure fundamentally invalid. (And I suspect this mistaken starting point is precisely where his thinking about gaze mechanics began.) Or again, let us look once more at this example:
Izumi analyzes this panel as follows: “The right-hand side shares the reader’s viewpoint; this construction gives the reader a sense of ‘fighting alongside the boy’ (this is probably the basic style of action manga).”
There seems to be no reason to think Izumi is deliberately calling black white to force the phenomenon to fit his theory. In his account, this panel already fits the “strength theory”; it has no need for the “viewpoint theory” to add further luster. If he genuinely believes the reader’s viewpoint coincides with the boy’s here, it seems we can only say that he fundamentally lacks the ability to analyze an image’s elements and structure. I find it hard to imagine someone with a basic understanding of the relationship between composition, viewpoint, and identification—or at least someone who knows what an “over-the-shoulder shot” is—taking this composition to unite the reader’s viewpoint with the boy’s. It places the viewing angle behind the man in black, makes the boy the object of observation, and depicts the ferocity of his attack. Surely the reader instead gains partial access to the man in black’s viewpoint and his observation of the boy.
Once this error appears, all the defenses we previously imagined for Izumi’s interpretations of specific manga details effectively collapse. Before we get caught up in whether he has thought correctly about the details, there is a more serious question: is he even capable of identifying those details correctly?
Of course, someone is bound to dispute this view of mine. At the beginning of “Positions Divided Between Left and Right,” for instance, Izumi spends some seven hundred Chinese characters explaining in detail why he believes manga has a “basic model” of “viewpoint theory.” Why could we not say that, in the example above, the reader’s apprehension of this model takes precedence over the apprehension of viewpoint in an individual composition? Frankly, if someone making this objection cannot see the many unreasonable steps and arbitrary leaps in his explanation, there is nothing I can do for them. As elsewhere in the article, his cognitive theory is fundamentally a fantasy built on a mixture of ordinary impressions and fallacies. I believe this really is not so hard to see. But the fundamental problems of Izumi’s theory are precisely what we will discuss next.
Having catalogued Izumi’s various intentional and unintentional errors, we now reach the crucial task of bringing things together. Here we will reconsider the central question of “The Basics of Gaze Mechanics in Manga”: what actually determines the characters’ left–right placement?