As Faust pointed out, that last one is nasty because it presents a chart and doesn't inform the reader how to interpret it. That isn't very nice as it requires one to re-arrange ones "mental grammar," as it were, to make sense of the "equation." To offer a linguistic equivalent, it would have been similar to a question asking information about the following sentence:

"Hates Zeality being tested on math"

Such a sentence is, effectively, no different than "Zeality hates being tested on math," but I switched the sentence construction (so it goes verb, subject, object, indirect object rather than the normal subject, verb, object, indirect object).

To relate that back to that problem, the problem reads "+" "X" "4" "1", with "=" being "gapped." One has to rearrange the "grammar" of the equation to make sense of it. "+" is essentially a verb for math, "x" being the subject, "4" being the object, and "1" being the indirect object.

Once that has been done, it is just simple addition (and subtraction); any 2nd grader should be able to do the majority of it. However, it also is a trick question. On a math section of a test, one has the expectation of being tested on math (silly expectation, I know, but blame the name). That problem is really just a thinking problem in the guise of math; it is testing how flexible a person's mind is, not how good they are at math. Thus, it is breaking a person's expectations. Very sneaksy.

Oddly, the test isn't uniform in testing one's "metathinking" apart from one's ability at math. Take the third problem, for example. With a 7th grade understanding of math, one could easily determine that the answer is around 350 (as the length of the hypotenuse is easy to roughly estimate, and the added length of the rest, to take into account the curve of the quarter circle, wouldn't that significant). If there had been a random distribution of answers, that is all one would need to do in order to find the answer. But no, the answers aren't randomly distributed; the designers specifically put two numbers that are unusually close together in the answer options (all other possible answers are around 50 apart, but two are only 10 apart). The problem would have been incredibly easy to solve if the test makers had played fair. But no, instead of being able to just change one's perspective, one really is forced to figure out one quarter the circumference of a circle to be sure one is getting the correct answer.

Indeed, from that page, I am not even sure if the designers were really aware of what they were testing in the first place.