As for the matter of people’s intuitions concerning whether they would notice a change, those intuitions are certainly worthy of a thorough consideration. I get the sense that, yes, most people would probably report that they feel generally aware of their surroundings and they’re not oblivious to the point that they feel they would mistake one person for another. Saying that you were the kind of person who would miss the substitution of one person for another generally would likely not reflect very positively on one’s character. Precisely how the subjects interpreted such a question, however, might be another matter: did they imagine a very similar looking person in just a slightly different set of clothes (two easy-to-mistake people) or did they imagine two people who looked radically different? Did they assume they were already paying attention to the person they were talking to, as many of the participants in the change blindness research likely weren’t? Depending on how one phrases the question, I would imagine you might see different set of intuitions.

I am aware that further studies had been conducted since, but my main focus is on the information in the paper as presented. On the basis of that presented data, the interpretations that were put forth didn’t seem to be well substantiated. I’m not saying they’re necessarily wrong, just that the conclusions aren’t able to made given that data set. My comments about the cuteness of the study relate to that point, specifically: were some of the shortcomings of the research overlooked because the results were not intuitive to people and/or because the experimental method was interesting? Admittedly, I don’t have an answer to this question.

Don’t get me wrong; I think there is a value in the method that this study used. I feel it would be well adapted for studying questions like the ones I posed in the post (in fact, I assume some people have already done such research, though I am unfamiliar with it). For instance, people could probably tell you that a white male construction worker approximately in their 20s or 30s had asked them for directions without being able to recall the color of their shirt, the sound of their voice, or accurately pick their face out of a lineup. Figuring out which details people attend to, along with what order they tend to be attended to in and why, could lead to many productive avenues of research, and this kind of design would be a good method of assessment.

As an aside, the construction worker manipulation reminded me of this paper: http://www.psych.ucsb.edu/research/cep/papers/eraserace.pdf. Giving someone a badge of group membership – in this case a construction hat – may well trigger mechanisms designed for detecting social coalitions. In fact, giving someone a badge like that may increase the tendency of subjects to be blind to other changes in the visual field, as badges may tend to command attention at the expense of other factors. That is, of course, provided the badges remain constant; were you to swap one badge for another – i.e. replacing the construction worker confederate with one wearing a yamaka – you would predict that people would then be less change blind.

In that context and time, the finding that people could miss a change to a person they were talking to was novel and important because it showed the generality of the phenomenon of change blindness. To this day, most people, when asked to predict what would happen, are convinced that they would notice the change—it is inconsistent with intuitions. Of course, for researchers or academics with the benefit of 15 years of hindsight into research on failures of awareness, it might not seem surprising any more.

The study was admittedly a “demo” study—it was effectively an existence proof of the idea that change blindness would occur during a real interaction. The sample sizes are small by modern standards, and would not be deemed large enough now. That said, the result is easy to replicate and variants of it have been conducted many times, both by me and my collaborators and by undergraduates and graduate students. Follow-up papers have tended to use larger samples, many more conditions, and more systematic manipulations of possible moderator variables. This was the first study of its type.

In the paper, the in-group/out-group hypothesis was one of several possible explanations for the pattern fo results. A full design would have used older experimenters and younger experimenters approaching older and younger subjects, but that proved impractical (it was hard to find older experimenters who had the time to run the task but weren’t already well known to pedestrians in that part of campus). The finding that the same change (actually, in some ways, a bigger one given the clothing differences) was less noticed by the same cohort of subjects shows the power of categorization. In the study, young subjects noticed 100% of the time when the changed people looked like students, but only about 1/3 of the time when the the *same two people* were dressed as construction workers. Might other explanations besides in-group/out-group coding hold? Yes. But again, this was the first demonstration of it’s type, and there aren’t too many manipulations in cognitive psychology that induce a 60+% change in noticing rates.

Would this study be published now? Almost certainly not. It’s cute, but in the context of nearly 15 years of additional work on change blindness, it would no longer be of sufficient interest as a “demo” study. It’s a bit misleading to use this study as an example of one that was published *because* it was cute. Cuteness didn’t hurt, but that wasn’t the reason it was published. It was also empirically and theoretically important *at the time it was published* and in the context of what had come before it in the change blindness literature.

Sometimes a demo study of this sort can be important and influential. For that to occur, the result must be clear and robust (as it is in this case) and there must be strong a-priori reasons (or at least beliefs) that it might not work. Making a paper more engaging for readers (via cuteness, humor, etc) isn’t inherently a bad thing. But, the reason it should be published is not the cuteness but the point it makes. I think the reason this paper has been so heavily cited is because it shows just how well the earlier results generalized to a more real situation and because the result continues to be surprising to those unfamiliar with the limits of attention. It is not a great test of your hypothesis that cuteness alone contributes to notoriety.