This is a follow-up question to a previous one asking for the potential impact of the typesetting tool (at the review level) on the chances of acceptance for a given paper.
I'm interested in the general case where reviewers will get a "feeling" of the way the article is presented, be it the writing style, British vs. American English, English from non-native speakers, choice of font, or like in the linked question, typesetting and any other criterion that is not related to the content of the work presented.
To be clear, I'm not asking about the more obvious or well-documented sources of bias, like the sex, name or geographic location of the authors. This can possibly be addressed by double-blind review.
As an example, I'm used to journal submissions in the form of a PDF with the least possible amount of formatting: plain text, typically not justified on both sides, lines numbered, usually double-spaced, figure captions at the end, figures concatenated at the end by the electronic submission system. Where I publish, typesetting is a journal's thing and as an author I'm happy to delegate that hurdle.
I once reviewed conference proceedings where the submission was very informal and couldn't help but having a "feeling" from the looks of the submissions. Authors who used LaTeX, which is unusual in my field, didn't look like they had a lot of experience publishing in this field.
I think that ideally this feeling should be silenced and I tried to do so as much as I could. However, reading comments and answers on this other thread made me wonder if that was a sentiment that others shared. My question then boils down to:
Do reviewers have a duty to try to ignore their biases or is the "gut feeling" an important part of the review?
As a final note, I'm not looking for arguments in favor or against the use of LaTeX in general, or in the specific case I mentioned. I find LaTeX to be very well suited for the applications it's intended for and I simply used that example to illustrate something that is not content but could (and apparently does) induce a bias.