It's dangerous to take a result like this and speculate on reasons, without developing proof that supports those reasons. The research mentioned in the article measured one particular trait which seems to be correlated with expressed political views, and provides the evidence for that (eye-tracking data); however, the article goes on to propose reasons for that trait and correlation, without justifying the
reasons with any evidence. Perhaps the underlying research gives reasons, or perhaps it doesn't, but either way the article seems irresponsible to report those reasons without supporting evidence.
This article also assumes a single axis, single-variable political spectrum from liberal to conservative, which is an absurdly simplistic view.
The one thing that does seem notable in the article is the conclusion, though. Put that together with Duverger's Law (https://en.wikipedia.org/wiki/Duverger%27s_law), which explains why we have two political parties in the first place, and perhaps the article's research explains where we got the rather odd party division we have, rather than a more useful one. I dislike the conclusion that this might make compromise easier (https://en.wikipedia.org/wiki/Argument_to_moderation), but perhaps if we can recognize the current liberal/conservative distinction (as expressed in politics today) as a difference that has little to do with useful political views, perhaps we can talk less about the hot-button items that currently make up news and campaigns, and start getting issues that matter into the political spectrum.