A little bit of background, first. Rochat (2003) argued that we possess five different levels of self-awareness;
- Self-consciousness or "meta" self-awareness
He also proposed a "zeroth" level of self-awareness, confusion (or no self-awareness), but made the claim that even newborns have a sense of differentiation - immediately after birth, children have a sense of their own body as a distinct entity. A further claim is that the first two stages are merely indicative of implicit self-awareness, and that it is only in the final three stages that a developing child gains of sense of "Me".
Children in the first and second stage of self-awareness might not be able to recognise that their reflection in a mirror is them, per se, but they are apparently aware that it is different from other images - children in these states, for instance, typically choose to observe an image of novel face as opposed to an image of their own, and will pay less attention to their own reflection than to an experimenter "shadowing" their every move.
In the new paper, Legrain, Cleeremans and Destrebecqz (2010) propose methods of measuring Rochat's three explicit levels of self-awareness. The first was the classic mirror task. The second involved taking a picture of the child several days before the experiment proper, presenting the child with their own photo amongst three pictures of other children, and then asking them to "Show [the experimenter] where [they were.]" The third method was a more complicated variant of the second; both the child subject and three peers were instructed to wear animal masks several days prior to the true experiment, and a photo of each child was taken whilst he or she was wearing the mask. They were then presented with the photos, and asked to point themselves out.
Legrain et al. found that the percentage of children successfully completing each task decreased as the complexity of the experiment increased; 93% of their participants passed the mirror task, and 84% passed the basic picture self-recognition test, but only 62% passed the mask test. These results were taken as supporting evidence for Rochat's developmental theory, but the authors did acknowledge that alternative interpretations were possible; it may, perhaps, be the case that the differing level of success was not due to development of self-awareness, but rather due to differences in broader cognitive abilities (most notably attention and memory).
Legrain et al.'s study is certainly good evidence towards the theory that self-awareness is not a singular property, and that it is instead composed of multiple levels (or at the very least can progress towards ever-increasing complexity).
As a side-note, Scientific American reported today on some similar research by Tanya Broesch, which investigated cultural distinctions between success in the mirror self-recognition task. Such research suggests that we should not only take into account differing developmental levels of self-awareness, but also variations in behaviour within each level.
Broesch, T., Callaghan, T. C., Henrich, J., Murphy, C., & Rochat, P. (in press). Cultural variations in children's mirror self-recognition. Journal of Cross-Cultural Psychology.
Legrain, L., Cleeremans, A., & Destrebecqz, A. (2010), Distinguishing three levels in explicit self-awareness, Consciousness and Cognition, doi:10.1016/j.concog.2010.10.010
Rochat, P. (2003), Five levels of self-awareness as they unfold early in life, Consciousness and Cognition, 12, 717-731