I don't know when, but at some point in my long career of reading fiction, I started to become annoyed at the omniscient-by-cheating narrative mode. That's a term I just made up. What it means is that the story's point of view is nominally first person or third-person subjective (meaning that the narrator only knows what one particular character knows), but they give other information by cheating. The most common form of cheating is the old "his eyes narrowed with suspicion". It tells you that the other guy is suspicious without leaving the limited knowledge of one character. In real life, you can't tell reliably when someone is suspicious by looking at their eyes. You might be suspicious that they are suspicious, but you can't really know. Similar lines: "the flat, dead eyes of a killer", "As he walked, Joe could tell from the smoothness of his movements that he was an experienced and dangerous fighter", "the bulge under his jacket showed that he carried a gun", etc. In the real world, his eyes might have been flat and dead because he was bored, he might have walked smoothly because his hemorrhoids made it painful to walk any other way, and the bulge under his jacket might be his lunch.
This narrative mode is different from the Sherlock Holmes trick. When characters do the Sherlock Holmes trick, they have some special ability to observer minute details and draw conclusions from them. There is in principle no reason that the character could not be wrong, any more than any other conclusion of the character might be wrong. By contrast, in the omniscience-by-cheating mode, the character doesn't have any special abilities and there is no chance that he is wrong, it is just the author telling you that this guy is a trained killer, that guy has a gun, etc.
You can see why authors do this. It lets them retain the intimacy and immediacy of the third-person subjective mode, while giving them the ability to reveal more information, to increase suspense, and such. I usually write in first person or third-person subjective but I don't do the point-of-view cheating because, well, I just don't think that way. If I want to reveal that this guy is suspicious then I have to come up with a way that the character could figure it out. If I really have to, and I'm in third-person subjective, I might step out of subjective mode for a bit because that seems less jarring to me than the idea that the character has sporadic god-like powers to read thoughts from facial expressions.
But even stepping outside of personal mode seems like cheating to me, so I avoid even that. And I think this explains the tunnel-vision quality of most of my fiction. What I mean by that is "tunnel vision" is that the stories seem to constrain the reader, keeping his view of events narrow, even claustrophobic. The reader feels that there are things going on that he doesn't know or doesn't understand because he just isn't getting all of the information. I've noticed this quality before, but never really understood what the source was. I've noticed it also in two authors, Jack Vance and Lawrence Watt-Evans. It's not necessarily a bad thing, but it is a bit constraining.
Part of this tunnel-vision quality is that I tend not to dwell on what the p.o.v. character is feeling. If it is mentioned at all, it is passed over quickly. But as I've been thinking about this point-of-view cheating, I think another aspect is that a strong unwillingness to cheat on the point of view, either by stepping out of it or by inventing signs that aren't there in real life, leads to this constrained view of the world. The reader doesn't know any more in the book than he would know in real life, and this gives a slight sense of claustrophobia and tunnel vision.