Surprise! We like smiling avatars!

Maybe we like smiling avatars after all?
Maybe we like smiling avatars after all?

Now isn’t this interesting! As my readers might know, I love experiments – even when they are tight and tiny samples of the type we experience in daily life. Here’s a fun one that just occurred to me, post hoc, that leads me to think we might like pictures of avatars smiling more than avatars making other facial expressions.

Many of the comments associated with this meme from bloggers and photographers suggested that we tend not to like the look of smiles on our avatars – the word “creepy” gets used a lot. Yet, very few people seem to question why – unlike in the real world – we don’t like to see each other smile in the virtual world.

Universally, smiles are associated with pleasure, sociability, happiness, and amusement – all good things. There is evidence to suggest that smiles make one more likable and more approachable. In some cases, a smile can be an advertisement of sexual interest. Again, all good things! 🙂

Perhaps what puts people off is the lack of lines that tend to surround the corners of our eyes (a Duchenne smile) when we are genuinely smiling, which is a dead give away for the kind of pasted-on smile we might share when we’re not really smiling, or have overindulged in botox. Still, most people would agree that the first of these following emoticons is overwhelmingly more popular than the others:

🙂 :-/ 😦 😕

Maybe it’s the falseness of the teeth, in all their ultra-bright and rectangular glory? It’s true, Second Life avatar teeth would do with a more realistic representation – something I can’t imagine would be impossible to make more lifelike. On the other hand, maybe it’s the “uncanny valley”, a hypothesis in the field of aesthetics which holds that when features look and move almost, but not exactly, like natural beings, it causes a response of revulsion among some observers.

Whatever the reasons, I’ve long shared the hypothesis that people tend to like neutral (emotionless) or pouty facial expressions on our avatars, which supports the ubiquitousness of the seemingly bored, gloomy, and emotionless facial expression on much of my Flickr Feed.

Two days ago I uploaded 6 photos to Flickr (5 rejects and one final) from my high school yearbook photoshoot. I thought it’d be interesting to see which ones would get the most favourites per views.

The 6
The 6 faces of Becky

At the time of this measurement the results are as follows:

Name Expression Favourites Views Favourable Proportion
Reject 1 Static 7 126 5.56%
Reject 2 Cross-eyed 4 115 3.48%
Reject 3 Angry 2 111 1.80%
Reject 4 Sad 3 95 3.16%
Reject 5 Worried 2 88 2.27%
Final Smiling 26 179 14.53%

To my surprise, the photo of me smiling was the most favourited – both in volume and in proportion of views. Could we like pictures of smiling avatars more than we say we do?

I considered any artefacts that may have crept into my analysis. There are some angle differences in the pictures, but generally there isn’t much change from picture to picture (except for the frame around and name under my final shot). I’m measuring the proportion of favourites, as opposed to total favourites, so this measurement wouldn’t be influenced by picture differences, online longevity or time of upload.

Another difference was that I submitted my final smiling photo to Berry’s Meme groups – perhaps people there are more favourable – still, it’s the proportions that matter, not the actual favourites. Also, I’m not seeing a significant difference in proportion of favourites I get from this photo, when compared to other photos on my Flickr feed that were not submitted to these groups.

One other issue that might be screwing with my results is that Flickr presents multiple uploads with one picture as the lead, and the rest inset – which might cause the first of the five rejects to be more favourited, because it’s more highly exposed. Again, though, analysing proportions would outweigh these issue, because it’s based on the number of raw views.

Still another issue might be that my smiling avatar was “on target” with the meme’s instructions, and therefore qualified as “correct”, meriting more favourites per view. There might be a similar effect due to my naming some pictures as “rejects”, with the most favourited one not a reject. Despite these possible artefacts, it’s certainly clear that people didn’t like my angry face, preferring my static face as the favourite among the rejects. Whilst not massively significant, it’s interesting to see that my cross-eyed face won over my sad face (marginally), and that my angry face won over my worried face (which was the least liked).

I’m keen to try more Flickr experiments like this – and next time, I’ll aim to wash away the nasty artefacts the might unduly influence my results. What do you think of these results? Do you agree or disagree that we might actually like smiley avatars more than not? What implications might this have for animators and mesh head makers? What does this say about High Fidelity’s plans to dynamically match facial expressions to our own for their virtual world? What other experiments would you like to see?

4 thoughts on “Surprise! We like smiling avatars!

  1. I think the “creepy” comments you noted are in line with “Uncanny Valley.” I think they were making the same point.

    IRL smiles are nice, and generally preferred in pix. Although the smiling photo is the ultimate trope. It’s funny smiles “feel” like the authenticity we crave, yet they are almost certainly entirely performed images. You might catch an occasional “candid” smile, but the “smiling portrait” is, by definition, a manufactured image. Still, even less-than-perfect smiles seem to be more appealing than non-smiles. I suspect there’s lots of psych research on this: babies and smiling faces, etc…

    Even though smiles are “preferred,” I wonder if the most impactful images aren’t smile-free: rocker with attitude, politician weighing delicate international incident, etc

    As for SL, I do think it’s uncanny valley. The smiles are “fake” or “forced” and they are uncomfortable to look at.

    If you look at my latest selfie, in the 1st photo of this post:

    I’m not smiling, and I’ve even obscured my eyes with sunglasses, yet I feel like the image conveys presence and attitude. For sure though, I do agree that a smile and eye contact are the most accessible way to present yourself.

    I’m not really excited about High Fidelity. In the great Augmentation vs Immersion pendulum, I feel like High Fidelity is all about Augmentation and SL2 is about Immersion. Who knows, maybe Rosedale will finally realize his lifelong dream of attracting huge enterprise customers and cash-in with High Fidelity. But for the immersive experiences many of us have come to value in SL, I’m more optimistic about SL2. And maybe by 2016 they’ll even figure smiles out!

    Liked by 1 person

    1. It’s hard to judge what the most “impactful” images might be without defining what one means by impact. Also, concepts such a “presence” and “attitude” are so grey, you’d have to define what you mean by that as well. I’m not talking about art here, I’m discussing likability, and using a crude tool like favouriting on Flickr to correlate with liking. I would assume that favouriting on Flickr (like most social media liking) has little to do with the artistic value of an image, versus how the image makes one feel (if the photos are by the same artist).

      The evidence that smiling make one more likable and more approachable is conclusive and long standing:

      I’m going to hold on off on deciding whether our facial expressions in virtual worlds will appeal or not. If the adoption of voice is anything to go by, I doubt it will appeal, but I’m curious to give it a try.

      On the augmentative and immersive scale, I’m beginning to look at these concepts in very different ways. I’ll likely share this new understanding in a few blog posts from now, but I’m seeing a much more comprehensive view of this relationship, that might suggest we’re looking at these words all wrong.

      Thanks for your comment!


  2. The data seems misleading to me because of the use of mesh heads and mouths. The range of possible expressions on SL avatars (system) depends the face so oddly, I think our brains don’t process the result as being associated with a known emotion. Mesh heads and mouths can more closely approximate that aspect, but we haven’t got any of the components naturally that make expression genuine (laugh lines, sparkling eyes, eyebrow movement) that signify visual cues to which we are accustomed.

    Liked by 1 person

    1. It’s fair to say that system faces might reveal different results to what mesh faces would – it’d be interesting to try this with a system face using facial animators.

      I’m not sure what you mean when you say: “The range of possible expressions on SL avatars (system) depends the face so oddly, I think our brains don’t process the result as being associated with a known emotion.”

      I disagree that our brains wouldn’t process facial expressions shown only on faces because the data refutes that entirely. If you what you mean is that facial expressions are somehow incomplete, then with that I can agree, as I mentioned in the post.

      Evidence for the smile you refer to (the Duchenne smile) was demonstrated as early as the mid-19th century, when a physiologist named Duchenne identified two distinct types of smiles: one which involved the contraction of the facial and ocular muscles, and one that didn’t engage these muscles (a “fake smile”) that many generally reserve for politeness.

      I would hope that any animators that are working on next generation avatars that reflect facial expressions would take these aspects into account, as the human eye is so trained on making interpretations from facial expressions, it’d be pretty hard to fool us.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s