Start with why you are doing this test:
...just want to know if he notices it and reminds what he saw.
Sounds to me like the proposed design is intended to help users notice the elements and remember the content.
What's your research question? It should be in question form, e.g:
Which of two "Related Articles" visual design treatments will users more likely notice and/or find memorable?
You could do a split-run test remotely using something like VerifyApp.com, specifically their memory test format.
Set up at least 2 versions of your article page design.
Except for the visual design treatments of the "Related Articles" page elements, make your A and B variants absolutely identical.
(Some people test >1 thing at a time but that's called multivariate testing and I think it's much better to start very simple if you want to stay focused on why you're doing the test.)
Then, run an experiment with the same number of participants for each of your variants.
Important: know how you intend to measure "better."
In your case, I think the more accurate details test participants mention per variant , the more effective the variant's design approach...agreed?
Also, please remember that A/B testing is kinda pointless unless you:
- Test only 1 thing at a time, and
- Have tons of sample size to make your result a non-fluke.
For example, if your A and B version are absolutely identical except for that 1 one thing (different styling OR wording OR colors OR size of the "Related Articles" element, OR nearly any other variable you can think of) any differences in the test results between A and B may be connected to whatever it was that you switched up.
If you design a simple test, even a 50:50 tie tells you something new, i.e. that the details of the 1 thing you happen to be sweating may not really matter that much, in which case you can move on to futz with some other aspects in the design.