One of the oldest humans performances is to detect faces. People would like to think that we don’t just look at body parts but indeed - that is how the eye-brain sees. Some of the built-in primitives are eyes, faces, secondary sexual characteristics and body dominance/submissive forms. (human morphology seems to be coded here - we don’t like things with the wrong number of legs like spiders and snakes) These have been shown to exist in the amygdala as part of your lizard brain wiring. These are “transferred” to the cortex over the course of becoming a grown human. This does shape your perception and emotional coding of these stored perceptions. This is really instinctual and there is not much anyone can do about it.
When you say you are viewing something in a mirror you are really saying swap the image left-right. Your left and right hemifields are split in the eye and routed separately into the brain stem on the way through the brain. The lower brain structures process these stream separately. One side (I forget which) responds to "fearful (whatever that is) or looming forms differently - more strongly. I could also see that shaping “pleasant” geometry. I suspect this colors your judgment that something is “right” and is mostly shared between humans.
It would be a shared human inclination that you could train out of - as you have stated. The underlying cortex can be retrained to counteract your original lizard programming.
As you noted - the papers talk about faces as these are easy to study.
If this emotional charge is hardwired into your subcortical structures than whatever these emotional triggers there are now known to be triggered differentially by a left-right orientation may possibly extend to other form recognition circuits.
What may be harder to find - but literature may exist in advertising circles - is what your lizard brain response to emotionally charged material aimed straight at your lizard brain. Certain forms such as curves that trigger your secondary sexual form filter makes shapes more interesting or “right” then they might otherwise be.
A chair’s curves can be angry or sexy. You see someone right-handed as more familiar and correct; due to this you subtly expect a person’s right side to be more developed - bulkier. I suspect that makes it so that when you draw something you feel that a slight bulkiness bias to larger on the left side seems “OK”.
More plainly: When it’s no longer sexy the spell is broken and it is just a form to be recognized as wrong.
I proofread by reading it backward; it’s not a coherent sentence anymore and suddenly I can see the errors. I see this as sort of the same thing.
Assuming I understand the phenom that you are describing, and certain facts gleaned from reading about this part of the brain, this general mix of facts makes sense to me. It is possible that if you structure your search correctly that much has already be learned about this line of inquiry.
Recognising something is a step in characterizing and controlling it. If you could characterize the way the lizard brain colors things then - yes - you can learn that with machine learning.