Navon (2001) suggests that we interpret our reflection in a mirror as a left-right reversal because it suggests a frontal encounter, yet the image deviates from how we would look if in the same position. I argue that this interpretation involves an arbitrary act of mental rotation, and is overly restrictive. I suggest that the more general reason why we interpret enantiomorphs of objects as the left-right reversal of the original objects has to do with the fact that left and right sides can only be defined once the other canonical axes (top-bottom, back-front) have been established.
2. But there is a problem here. To reach the conclusion that the mirror-image is a left-right reversal, you have had to do some rather elaborate mental work-- that of cloning yourself and moving the clone to the side of the mirror. This also involves mental rotation, since the clone is now facing in the opposite direction to the way you are facing. Unless you carry out this mental transformation, it is not immediately obvious that your reflection is a left-right reversal, and a much more direct interpretation is that the mirror reflection is a back-front reversal. Actually, you don't have to clone yourself-- you could simply imagine the mirror image rotated 90 degrees, and then compare it with the real you, but either way, the 90-degree rotation is necessary to show the reversal as a left-right one. Further, the decision to rotate about the vertical axis is, from a geometric point of view, arbitrary. It would be simpler to leave both you and the mirror-image unrotated and conclude that the mirror-image is a back-front reflection. Which it is.
3. A more revealing exercise, I think, is to imagine your mirror image extracted from the mirror and allowed to roam freely in the world. This is your enantiomorph-- or enantioyou. In this case, it doesn't really matter how you encounter enantioyou; the irresistible conclusion is that enantioyou is your left-right reversal. But in fact, enantioyou can be regarded as a left-right reversal, a front-back reversal, or a top-bottom reversal, or indeed a reversal about any plane at all, depending on how you choose to align it with your own body. Nevertheless there are three canonical ways in which you might align yourself with enantioyou in order to characterize establish difference between you.
4. First, you might align your legs with that of enantioyou, and arrange for your noses to point in the same direction. You would then conclude that enantioyou is a left-right reversal of yourself. This is the alignment implied by Navon's mind game of cloning yourself and placing the clone beside enantioyou. It is also the alignment achieved by standing with one side facing the mirror.
5. Second, you might align the legs, and then align the hands, so that the watch-bearing hand, say, is on the same side in each case. You should then conclude that enantioyou is a back-front reversal. Ironically, this is the alignment provided by the frontal mirror, yet you spurn the direct interpretation and conclude that enantioyou is a left-right reversal. This is clearly not explained by frontal encounter, since the frontal encounter should force the interpretation of a back-front reversal.
6. Third, you might align the noses and the hands. When you have done this, you would find that the legs point in opposite directions, and you should then conclude that enantioyou is a top-bottom reversal of yourself. This situation is achieved by standing on a mirror, but again I think you would be tempted to interpret the upside-down enantioyou as a left-right reversal. The body stretching away from you in the mirror under your feet may well be upside down relative to yourself, but inspection would again show the watch to be on the other hand, the logo on the T-shirt to be left-right reversed, and so on.
7. More generally, I think you would naturally interpret enantioyou as your left-right mirror image regardless of how you and enantioyou are aligned with one another. Enantioyou just seems to use the opposite hand, have a hair parting on the opposite side, wear the watch on the other wrist, and, if you could only see it, have situs inversus of the heart and other internal organs. I don't think this has anything to do with frontal encounters, and you would reach the same conclusion if you never in fact encountered enantioyou frontally, or even imagined yourself doing so.
8. So why is the left-right interpretation so compelling? Bilateral symmetry might have something to do with it, since it is easier to imagine your left and right sides interchanged than to imagine back and front reversed, or top and bottom reversed-- these reversals seem to require more anatomical violence. But symmetry can't be the whole story, since we'd still prefer the left-right reversal if we were grossly asymmetrical, perhaps through the loss of a limb or an eye, or indeed the loss of several lateralized body parts.
9. Labels might have something to do with it, since the left and right sides of the body are essentially labeled the same way, whereas back and front and top and bottom are not. The left and right hands are both hands, and it's not so difficult to imagine them swapped, whereas the top of the body contains a head and the bottom contains feet-- these not only look different (in most of us) but have quite different names. Similarly, the front of the head is a face, and there is no face at the back (although I do have an acquaintance who, when asked to supply a head-and-shoulders photograph for a job application, judiciously had it taken from the back).
10. Underlying these factors, I think, is the general status of the left-right dimension, which cannot be defined until the top-bottom and back-front dimensions are established (see also Corballis, 2000). Top and bottom are typically established by gravity, and back and front by movement. Trees have a top and bottom, but no left and right because they have no back and front. In nature, there is little to demarcate left and right once the top-bottom and back-front axes are defined, but even in the case of asymmetrical creatures, like lobsters, the left and right sides are not defined until the other two axes are established. It follows that we can reverse left and right without disrupting the other two axes. Suppose for example that we had indeed chosen the frontal mirror image to be a back-front reversal-- as in fact we should, because in the most direct sense that is what it is. This would then leave us in a bit of a quandary as to which is the left and right of this creature.
11. Ideally one should say that handedness is unchanged-- it's just that the face has somehow changed places with the back of the head, and so on down the line. But we then have a problem as to which is the left and which the right side. If we regard the back of the head as now the front, then left and right are unchanged. But if we define left and right on the usual assumption that the nose represents the front of the face, then left and right have swapped sides; the wrist that carries the watch is now on the other side. The simpler interpretation is just that left and right have switched. Enantioyou would no doubt be happier with this interpretation, on the grounds that his or her eyes are to be interpreted as looking forward, not backward.
12. In short, I think at least part of the reason for regarding enantioyou as a left-right reversal is just that the top-bottom and back-front dimensions must be specified before the left-right dimension can be specified. One would still be compelled to regard the nose and eyes as defining the front of the face even if front and back were reversed, and the head as defining the top of the body if top and bottom were reversed. It does not really make sense to say that enantioyou is in the distressing habit of walking backwards everywhere, simply because the direction in which a person normally walks is forwards by definition. Inevitably, then, it is the left-right switch that carries the burden.
13. Navon may well complain that I have dealt with the Kantian question of why enantiomorphs in general tend to be interpreted as mirror images, rather than with the question of why we interpret a frontal mirror image as a left-right reversal. He suggests that these two cases have "not much" to do with each other. I disagree. It is true that the mirror imposes constraints that do not apply to other enantiomorphs, such as a pair of shoes, and images in real mirrors usually allow for perceptual comparisons whereas real- world enantiomorphs may not be immediately perceptual. The writing on an ink stamp, for example, is seen as a left-right reversal even though there need be no "normal" version to compare it with. But I think it is unparsimonious to suppose that the principles applying to the ink stamp are different from those applying to enantioyou. Enantioyou, whether viewed frontally as in a mirror, or as having escaped, like Alice, from the mirror's constraints, is subject to the same kind of interpretation as the ink stamp. Writing has an implicit top and bottom and back and front, and once these axes are defined the left and right can then be established, and the ink stamp then seen to have print that is left-right reversed.
14. The behavior of human participants in the classic mental-rotation experiments of Cooper and Shepard (1973) make the point. Asked to decide whether rotated alphanumeric characters were normal or backward, the participants immediately understood which was the top and bottom of each character, and then mentally rotated the character to the upright before deciding whether the character was normal or left-right reversed, illustrating the priority of the top-bottom axes over the left-right one. As far as I know, no participant has ever normalized the characters with respect to left and right, and then made the decision according to whether they are top-bottom reversed. For much the same reasons, we normalize the mirror reflection by aligning tops and bottom and back and front, leaving the left-right axis to carry the burden of the reflection.
Cooper, L. A. & Shepard, R. N. (1973). Chronometric studies of the rotation of mental images. In W. G. Chase (Ed.), Visual information processing (pp. 75-176). New York: Academic Press.
Corballis, M.C. (2000). Much ado about mirrors. Psychonomic Bulletin & Review, 7, 163-169.
Navon, D. (2001). The puzzle of mirror reversal: A view from clockland. Psycoloquy 12 (017). http://www.cogsci.soton.ac.uk/psyc-bin/newpsy?12.017