David Navon (2002) It is not all in our Mind. Psycoloquy: 13(012) Mirror Reversal (6)

Volume: 13 (next, prev) Issue: 012 (next, prev) Article: 6 (next prev first) Alternate versions: ASCII Summary
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 13(012): It is not all in our Mind

Reply to Corballis on Navon on Mirror-Reversal

David Navon



That people see left-right reversal in various mirror positions suggests to Corballis (2001) that the reversal is not due to any geometric relationship but rather to some cognitive disposition. It is argued here that though the spatial relationship between real-world enantiomorphs greatly vary, a viewer and her mirror counterpart are bound in a specific spatial relationship determined uniquely by mirror position. When that position optically reverses the transversal axis, the perception of left-right reversal has a straightforward, physical account. Thus, the locus of the puzzle is in the case in which that condition is not met, namely in a frontal view. In that view a viewer judges her image in the mirror not by its relationship with her facet which faces the mirror but rather with the putative image of the latter in a frontal encounter. That entails left-right reversal. The cognitive disposition account is argued to be incomplete and doubtfully more parsimonious.


spatial cognition, mirror vision, mirror reversal, left-right reversal, perceptual frame of reference, object perception, frontal encounter, enantiomorphs, handedness
1. Corballis (2001) does not think that left-right reversal has much to do with explicit geometric relationships. He rather argues that the geometric relationship with the enantiomorph produced by the mirror depends on which mental rotation one chooses, yet people by and large see left-right reversal. He concludes that the reversal ensues from a cognitive disposition that generates dominance of other axes over the horizontal one.

I. Travels with my "enantio-me"

2. Corballis first takes issue with the point that my cloning thought experiment (Navon, 2001, paragraphs 8-13) requires quite laborious mental work and that its outcome depends on the particular way the process is implemented.

3. As much as I appreciate thought experiments, I think that their value is not to be judged by the complexity of the mental processes of their imaginary participants. Specifically, the cloning experiment I proposed was primarily meant to suggest that the cognition of left-right reversal does not simply spring from the mirror image itself but rather depends on a comparison between it and some other image. That latter image is actually MENTAL, namely what the viewer expects to see had he managed to frontally encounter himself. The cloning story - illustrating how another SENSORY image would enable the viewer a convenient way to compare between the images - was meant just as a didactic hallway to the main argument.

4. Furthermore, the fact that the outcome of the comparison is decision-dependent was instrumental for what I meant to point to: (a) the reversal perceived depends on a decision, yet (b) that decision is far from being arbitrary. As I explicated in paragraphs 11-13, the viewer decides to put her copy at a posture of a typical frontal encounter with herself, because her objective is to judge how veracious the mirror is in telling her what she would see had she at the moment encountered herself. That decision is analogous to what I suggest happens in actual mirror perception (ibid, paragraphs 14-15): In a frontal view, the perceptual system pits the mirror image against a typical frontal encounter, although it could in principle elect to compare it with any other encounter. It does it because the mirror image matches the typical frontal encounter more than any other encounter.

5. Corballis then invites every "you" of us to some imaginary trips with "enantio-you" been released out of her mirror cage. As much as I liked joining those, I think they are not absolutely necessary. While the distinction between enantiomorphic pair-mates in intrinsic structure is given, the spatial relationship between them can obviously change depending on how they are positioned with respect to each other. The point is, however, that unlike other enantiomorphs (e.g., a pair of gloves), the optically generated enantiomorphic pair, of a viewer and her mirror counterpart, are BOUND in a specific spatial relationship determined uniquely by mirror position.

6. In some positions, left-right reversal is dictated by mirror optics. That is not puzzling at all. The locus of the puzzle is the case in which left-right reversal is not optically determined, namely the highly prevalent frontal view (see Navon, 2001, paragraph 3). In a frontal view, the relationship between the viewer and her mirror counterpart is reversal along her midline axis, being the one that is then perpendicular to the mirror plane. That entails a perfect overlap between the facet facing the mirror, namely her front, and the corresponding facet of the mirror counterpart. Yet evidently, people do not attend to that, let alone feel that that is relevant to how they judge their mirror reflection. They rather choose to judge it by another relationship - its conformity with their real frontal image. Alas, being both the object and the spectator they cannot actually see both images simultaneously. So, they either have to imagine some exotic trip out of their body (or of their "enantio-body" out of its mirror cage) or apply a much simpler operation - retrieve the memory of a frontal image of any human in a prototypical frontal encounter. THAT is what is compared with the mirror image. The output of that comparison is one and the same - left-right reversal (for a terse generalization to mirror reversal of any object, see Navon, 2002, paragraph 17).

7. That actually overflows to other sorts of view in other mirror positions. The reason might be that in order to see the mirror image, the viewer has to slant his head so that his face is almost parallel to the mirror regardless of body posture. For example, if he stood on a flat mirror, he would quite naturally see himself upside-down (Navon, 2001, paragraph 30), yet bending his head to see that would make his face frontally view the mirror. The way the eye views the face seems to matter for left-right perception no less, and perhaps even more, than the way the body is viewed. More basically, the left-right axis can be defined for every FACET of an object regardless of the position which that facet occupies within that OBJECT's object-centered coordinate system (ibid, footnotes 2 and 3). For example, when a viewer focuses on the left profile of another person, he might assign the nose to the left of the left profile, and the nape - to its right, hence ironically entertain two senses of left-right continua concurrently. Accordingly, a viewer standing with the edge of his right shoulder pointed to the mirror might feel, perhaps even without turning his head towards the mirror, that the mirror reverses left and right of his right body side, in the sense that what it reflects is left-right reversed with what another person would see were he looking on it.

II. Is the reversal due to left-right being a neglected axis?

8. Yet, Corballis prefers a different account - that people somehow see left-right reversal in whichever posture they find their "enantio-body" at, probably because the horizontal axis has a lower status in human cognition than the other two orthogonal axes, suggested also by how people tend to interpret real-world enantiomorphs. Such a hierarchy of axes may well exist. If so, it perhaps exists for the ecological reasons Corballis alludes to, and it perhaps helps to explain some phenomena of the sort he mentions. The question is whether or not it predicts phenomena I addressed in the target article. I find the reasoning needed to make such predictions somewhat obscure. True, some gross clues must serve to suggest that the mirror image appears to be a sort of frontal encounter. Those must predominantly be clues that help to differentiate between front and back. However, from then on there is only one path, regardless of whether or not all frontal axes are treated alike - noting that top is opposite to top and left is opposite to left, which is left-right reversed with how it should be in a prototypical frontal encounter.

9. Corballis nonetheless justifies his account by claiming that it is parsimonious, being quite general. General it indeed is, yet perhaps overly general. A phenomenon requires an explanation only when it does not have a straightforward one. We do not wonder, for example, why people judge that one of two bars is longer than the other one, when the former is indeed physically longer. We would start to wonder about such a judgement, if the bars were actually equal in length, as they are in a typical display of the Ponzo illusion. That requires a special explanation, which makes Occam's razor bow out.

10. In the mirror case, the perception of left-right reversal is quite natural when the optics actually reverses position along the transversal axis, as when the viewer points the edge of his right shoulder to the mirror (ibid, paragraphs 2, 30). Yet, a perception of left-right reversal that cannot be due to such actual axis reversal, as in a typical frontal view, does invite an explanation. A general human disposition is a legitimate hypothesis, but it is far from being confirmed by instances of reversal that are readily afforded by the physics of the stimuli. It is, moreover, quite challenged by the shadow example (ibid, paragraph 29) as well as by the atypical cases in which the mirror image is perceived to reverse the vertical axis for good reason (ibid, paragraphs 27, 50).

11. Finally, Corballis cites findings in classic mental rotation experiments (e.g., Cooper & Shepard, 1973) as specially strong evidence that the vertical axis is more dominant than the horizontal one. I am sure he is much better acquainted with findings in that paradigm than I am, but at least whatever I know does not quite seem to warrant such a conclusion. Since left-right inversion is recognized best in the canonical orientation, subjects resort to mental rotation of disoriented characters [1], which requires that they note the deviation in orientation. That in itself does not indicate that they apprehend where the intrinsic top is within the disoriented stimulus more readily than where the intrinsic left is, certainly not in all angular deviations and all stimuli. A 90 deg rotation might be different in that respect from a 180 deg one, and an upside-down G might be different from an upside-down R. Across stimuli, it is yet to be shown that discriminating a left-right inverted stimulus from a canonical one is harder than discriminating a top-bottom inverted one from a canonical one.


[1]. Which brings to mind a question I could not find an empirical answer for: What would happen if the task was presented to the subject as calling for discriminating between normal and top-bottom inverted stimuli (rather than "backward" which must be taken to mean left-right inverted ones)? The two ways of defining the task are formally equivalent, but actually subjects might exhibit faster reactions in stimuli that are presented with the intrinsic left at the subject's egocentric left than in stimuli that have it on the subject's right.


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. (2001). Why mirrors reverse left and right: Commentary on Navon on mirror reversal. Psycoloquy, 12(032). http://www.cogsci.soton.ac.uk/psyc-bin/newpsy?12.032

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

Navon, D. (2002). It takes two for an INVERSE relationship: A reply to Burgess. Psycoloquy, 13(011). http://www.cogsci.soton.ac.uk/psyc-bin/newpsy?13.011

Volume: 13 (next, prev) Issue: 012 (next, prev) Article: 6 (next prev first) Alternate versions: ASCII Summary