Visual Illusions: Nothing to Lose but Your Chains—A Reply to Oliver Braddick

Oliver Braddick’s (2018) strident editorial, written in response to the recent publication of The Oxford Compendium of Visual Illusions (Shapiro and Todorović, 2017), resurrects opinions that he presented in a 1972 review of J. O. Robinson’s The Psychology of Visual Illusion. Braddick’s polemical tone, along with his attempt to rebrand illusions with a disparaging label, ¹ is unfortunate. Although some of Braddick’s cautionary words about the study of visual illusions may be worth attending to, his argument neglects many of the useful aspects of visual illusions for our field. Braddick’s editorial is a missed opportunity to give a more balanced (and more informed) update regarding the psychology of visual illusions, such as that given by Rose (2018) in the same issue of this journal.

Braddick is not so much against a particular illusion or even against illusions per se; rather, he claims that the act of collecting illusions in a single volume leads to an artificial category that will prevent students from thinking correctly about fundamental issues of perception. So, as he says in his 1972 review: “[I]f we suggest that the illusion effects are the essential problems, we blunt our efforts to reorient the student so that [they] can look at [their] everyday psychological processes afresh.” And Braddick in 2018: Illusions as a category are mainly stuff for the “sweetshop window”; “Illusions appear in children’s science books … because they have an immediate impact that doesn’t require any thought.” Before dismissing illusions as nonserious because their appreciation “doesn’t require any thought,” we should not forget that many fields of science have benefited greatly by examining illusions, games, and other such wonders. It is tempting to simply create a list of examples where exposure to phenomena have led to fundamental discoveries; I will refrain from doing so, but not without calling attention to the fact that Newton’s Optics involves observations with a prism (a toy Newton picked up at a fair).

At a trivial level, I too am impressed with the immediate sense of wonder that illusions can create in both experienced and inexperienced observers. Indeed, Dejan Todorović and I chose the word compendium for the title of the compilation because a compendium (among other definitions) describes a set of games and puzzles. A 5-in-1 games compendium, for instance, typically includes chess, checkers, backgammon, dominoes, and cribbage. However, like visual illusions, the games in such collections have a history that extends over generations and cultures, have been a means for recreation as well as objects of intense study, and have inspired or have served as analogies for models about how we think and how we perceive the world. Ours would be a less informed world if we were to take the stance that these games are merely useful diversions, and that we should not try to generalize from one game to another.

What does Braddick suggest we do? In both 1972 and 2018, Braddick recommends that we challenge students to think about mechanisms behind what many would consider more ordinary aspects of perception: as he stated in 1972, “it is easy to find the Müller-Lyer effect mysterious; it is more difficult to realise that the ability to compare the lengths of two unadorned straight lines is equally mysterious.” The goal is to “reorient the students” to get them to think about the more mechanistic questions, “What kind of process might do that? How could we try to find out?”

At some level, I applaud the approach that Braddick outlines (I even once had a blog with the subtitle “Why are we surprised by only some of the things that we see?” which was intended to convey much the same idea about many “ordinary” perceptions). But—and it is an important but—I grow apprehensive when someone is prescriptive about the correct pedagogical approach, especially when Braddick seems simply to have substituted one phenomenon for another: Straight lines are essential and salubrious; illusory bent lines are froth and will stunt students’ thought processes. Why should one type of line be better than the other for getting students to think about perceptual processes? Braddick comments on this directly in 1972:

[The question] “What mechanisms enable us to perceive?” (which is what the study of perception is all about) is raised both by veridical perception and by illusion. But in the latter case it tends to be preempted by the more naive question.

As if students are only capable of asking one question or only capable of having one reaction! I would imagine that students would be more interested in asking and answering questions about visual mechanisms that can produce straight lines if they consider why straight lines occasionally don’t appear straight. Illusions, like Socrates, rock music, and video games, can change one’s perspective of the world; accusations that they corrupt, limit, or stunt the minds of the young should be treated with caution and perhaps a bit of suspicion.

Braddick also claims that while illusions may be helpful in elucidating particular conceptual boxes (spatial perception or color appearance, for instance), not much is to be gained by thinking about relationships between illusions: “If we had a good mechanistic theory of the Müller-Lyer illusion … would it be any help in deciding how colour contrast works?” (2018). Since Dejan Todorović addresses this issue in detail, I will point the reader to Todorović’s response. Here, I will note that illusions from different areas of the field may have more in common than one may expect, and that these commonalities may be informative about the processes underlying perception. For instance, contrast phenomena can occur in nearly every visual dimension (the Compendium has examples of contrast with regard to color, size, blur, faces, motion, shape, etc.). So, when Braddick (2018) writes, “If we want to understand how we assign surface colours, by all means let’s explore colour contrast,” he confines color contrast to a problem related to surface color perception and ignores other relationships that have potentially deeper connections and would be worth understanding.

Braddick suggests that Müller-Lyer and color contrast are not related to each other, and he may be correct. However, I would not be surprised if both effects will eventually be better understood with the application of more sophisticated models of spatial processing, and I certainly wouldn’t want researchers to be constrained from trying to think more broadly about processes that cut across current conceptual divisions. There are reasons why many influential general theories of perception often test their mettle by accounting for wide swaths of seemingly disparate visual phenomena. If we were to confine illusions to a narrow conceptual space, we would limit our ability to seek answers to the question, “What mechanisms enable us to perceive?”

Despite my objections to Braddick’s main points (and certainly to his tone), I do want to emphasize the importance of what he is getting at when he objects to phenomenon-centered research. Illusions have always been problematic for our field, and for good reason. In the Introduction to the Compendium, Dejan Todorović and I recount an instructive story from Ebbinghaus in 1913, who has a new illusion that, he claims, has “an appropriate and immediately obvious explanation.” Indeed, says Ebbinghaus, “[the phenomenon] can only be accounted for in such and such a manner, otherwise one would deny one’s capacity for theoretical thinking.” Yet when Ebbinghaus shows the illusion to a colleague, the colleague “maliciously rejects” Ebbinghaus’ explanation and insists on a different solution “with the same self-assuredness and same claim of obviousness.” And so it continues to the third and fourth person, “everyone pursuing the matter in accord with [their] own ideas.” Ebbinghaus’s conclusion is that “we currently have a great number of theories which differ in their basic principles and also in the application of those principles.” I think it is fair to say that Ebbinghaus’ conclusion sounds familiar to most contemporary vision scientists. More than that, the number of theoretical explanations does not seem to be decreasing over time. While it may be tempting to blame this fault on the study of illusions themselves, and I would certainly agree that studying phenomena directly is at least part of the problem, failure to eliminate longstanding alternative hypotheses might have less to do with illusions and more to do with the diversity of approaches in the field.

To me, the most disappointing aspect of Braddick’s current editorial is that he doesn’t mention a single illusion that wasn’t around when he wrote his 1972 review or, for that matter, wasn’t around a half-century earlier, when Lenin wrote the speech from which Braddick takes his title. The Oxford Compendium has 115 chapters of brilliantly presented visual phenomena, written by authors who created the phenomena or have published research articles about them. The authors have different theoretical perspectives and methods of creation. Braddick might be surprised that many of the illusions are outgrowths of particular theories (i.e., the illusions are research-generated phenomena). The illusions in the Compendium are worth considering on their own merits and in conjunction with others, but Braddick’s 1972 and 2018 statements make it seem as if Braddick decided early on (and in the infancy of this journal) that such phenomena should be dismissed (after all, as he says, illusions are appealing because they “[don’t] require any thought”).

As scientific topics go, visual illusions have an extraordinary longevity; many of the older ones have been the topic of research, discussion, and, yes, amusement for generations, and I suspect that many of the illusions in the Compendium will share the same happy fate. The reason for this longevity is not simply because the illusions are eye-candy; rather, they serve to illustrate the complex nature of how the brain constructs reality. One can certainly snipe about an illusion and its popular uses, but please don’t lose sight of the fact that visual phenomena have provided key insights in neuroscience and in physics, have given scientific inspiration to people in nearly every field, and have created a sense of wonder in both children and adults. We are fortunate to be part of a field that intertwines fascinating visual phenomena with topics that are of theoretical significance. My guess is that the Compendium will enhance, not stunt, both the vision scientist’s and undergraduate’s intellectual growth and perspective. I will close with a rejoinder to Braddick’s Lenin allusion in his title: Readers of Perception, look at the Compendium, and think about the topic for yourself! We have nothing to lose but our chains, and the perceptual world to win!