What I've read is that even without facial expressions, the sense of the other person being there is surprisingly strong because of the subtleties of seeing how a person moves and reacts to you. I can't find any of the articles I've read saying that, but I'll transcribe a bit of this audio interview with two people right after they tried Oculus's Toybox demo:
"For me the biggest takeaway was how much is added when you're with another human being in VR and you're able to intuitively interact with them."
"I felt like this person that I'd never met before -- I felt at the end of it like we were bros. ... You're sharing this VR space but at the end it feels surprisingly intimate."
Note that it's already that good even without any representation of the other person's facial expression. All it would take to add facial expression are some sensors on the inside of the headset. There are even companies working on inferring what your eyes look like from the expression on the more-visible bottom half of your face.
"For me the biggest takeaway was how much is added when you're with another human being in VR and you're able to intuitively interact with them."
"I felt like this person that I'd never met before -- I felt at the end of it like we were bros. ... You're sharing this VR space but at the end it feels surprisingly intimate."
That's from here: http://voicesofvr.com/228-candid-reactions-to-the-oculus-toy...
Note that it's already that good even without any representation of the other person's facial expression. All it would take to add facial expression are some sensors on the inside of the headset. There are even companies working on inferring what your eyes look like from the expression on the more-visible bottom half of your face.