Our Metaverse Avatars — Avoiding Uncanny Valley

Why our metaverse avatars don’t look like us (and why they never should) and how they can help in areas like online therapy and business

Catriona Campbell
4 min readJan 28, 2022

In my last blog, I wrote about digital twins, a technique used by all sorts of organisations, such as supermarket chains, to gain a competitive advantage through simulation. If you happened to read the post, you’ll recall that I made it very clear I wasn’t referring to what people may consider to be another type of digital twins: our metaverse avatars.

For those who need a quick update, the metaverse is an immersive, digital world that humans access using virtual reality goggles, widely considered to be the next iteration of the internet.

However, avatars aren’t really digital twins. Imagine you’re strolling through, say, the computer-generated aisles of Walmart’s proposed virtual supermarket. You pop jars of pixelled gherkins, medjool datas, and all sorts of other tasty bytes into your digital trolley (I really hope you see what I did there). Then your avatar happens to catch a glimpse of itself in one of the mirrors bookmarking the refrigerated aisles. Do you think you’d see a true digital likeness of yourself? No, you absolutely wouldn’t.

In truth, you’d find a virtual self that looks nothing like you, but one that instead appears a lot cartoonier. It would resemble the avatars you or your teenagers can create on video games consoles — or even on any number of sites like Ready Player Me. Check out the example below, which I threw together quickly based on a photo of me:

But why aren’t our avatars more similar to us? For a couple of interconnected reasons. One, compute power isn’t quite yet at the level required for us to design and run true-to-life digital representations of humans in the metaverse in real time. Two, even if it were close enough (and it could be very soon), we wouldn’t want this. Unless the avatar was a perfect match, users would be uncomfortable — even the subtlest of dissimilarities could have the effect of sending them into sudden meltdown. This is what we call ‘uncanny valley’.

Uncanny valley is a phenomenon originally developed by renowned Japanese robotics professor Masahiro Mori in the 1970s to describe the creepy, disturbed sensation people usually experience when they encounter humanoid robots that closely — but imperfectly — resemble human beings. Mori explained in a seminal paper on the topic at the time:

“I have noticed that, in climbing toward the goal of making robots appear human, our affinity for them increases until we come to a valley, which I call the uncanny valley”.

The roboticist’s theory is just as applicable today as it was five decades ago thanks to the rapid proliferation of not only artificial intelligence and increasingly realistic automations but also augmented reality and virtual reality. And so, as we look to a future in the metaverse, uncanny valley is once again on our minds.

We’ve even seen examples of the concept crop up in mainstream culture. For instance, during early test screenings of computer-animated movie Shrek, a lot of kids were incredibly unsettled by the lifelike appearance of Princess Fiona, and the filmmakers consequently decided to edit her to look much less realistic.

Conversely, when the balance between real and fake is perfect, the use of avatars can elicit a positive emotional response in people similar to human-to-human interaction. We’ve seen this, for example, in online therapy, where digital representations often facilitate a more meaningful text dialogue between psychologists and their clients, helping establish a better rapport between the two parties than if there were no visual representation at all.

They’re also capable of evoking an even more positive emotional response than a real human face during video communications. Returning to the online therapy illustration, replacing people with avatars on a Zoom or Skype meeting can encourage a client to open up and share more candidly.

Avatars have proven beneficial in other areas too, including business. A recent LinkedIn post from Gymshark (founded by EY’s Entrepreneur of the Year UK 2020, Ben Francis) detailed the firm’s foray into metaverse meetings, with CEO Ben and Chief Brand Officer Noel Mack saying they loved their digital conflab.

After the pair switched to a larger, brighter virtual space, Noel called the experience “a step ahead of Zoom that feels much more authentic”. He then joked about Ben threatening virtual fisticuffs before reflecting on his avatar — to be specific: the avatar his avatar could see on his virtual laptop screen. A metaverse within a metaverse — how terribly meta, even for the metaverse!

I definitely advocate digital twins (the other kind), but for the reasons I’ve outlined here, I’m happy to stick with the cartoonish Catriona you see above. If there’s one thing nobody wants (perhaps apart from professional wrestlers and Gordon Ramsay), it’s for their appearance to paralyse people with fear.



Catriona Campbell

Behavioural psychologist; AI-quisitive; EY UK&I Client Technology & Innovation Officer. Views my own & don't represent EY’s position. catrionacampbell.com