Beginners often look for visuals alone, without going deeper and looking for what is under the surface. Hourences argues that art is all about emotions and emotions aren't defined by mathematical equations. I can't dive into the matters of rational vs emotional, but every company that makes some product has to convey a message and the target audience has to be able to get it. This applies to games as well.
About emotions I'd like to make something clear. If the person has deficits in expressing emotions and/or feeling them and/or understanding them and/or recognizing them, we have a problem. This ought to produce consequences in level design or game design for that matter in some way. When we confront rational minds vs emotional minds we may have misunderstandings.
Composition. It's about how things feel and get your attention or not. Hourences doesn't go deeper here because there are numerous articles and work regarding composition in cinema, photography and arts in general. It'd be too much to add to his book. In here I'd like to say that it's funny how Hourences said that arts is about conveying emotions and to explain composition he resorted to graphs, which are a mathematical tool. I'd also add that this closely relates to cognitive functions because balance is a rather subjective topic and people have their own descriptions of what is balanced and what is not. Mark Rosewater in his lessons about designing cards for Magic tells that often people expect symmetry and that may or may not be related to any kind of mental health issue. There certainly some general rules, but there is always room for creativity.
Here is an example of the graphs that Hourences used in his book. I think that he forgot to label the axis. Vertical would be contrast, while horizontal would be space or time. I know the most common graphs that we should learn at school: log, exp, sine, etc. But I really don't think the mathematical theory applies in this case. The point that he makes in the book is that lack of contrast or too much contrast are to be avoided.
One consequence of having too much contrast is that it can misguide the player. The entrance to the next area is on one side, while contrast guides the player to the opposite site. This happens in Shadow Warrior 2013 for example.
About natural objects such as rocks and plants he says that objects should support each other. Support how? He doesn't provide an answer and I shall provide my own. I'd look for common sense and pictures of natural landscapes. In Discovery Channel there are some shows about unexplored places and search for lost civilizations. One thing that they look for is anything that doesn't look natural, anything that looks man made. For example: to place rocks in a circle would be close to impossible to happen out of nowhere, unless some person did it. I guess that's the kind of composition that he is talking about.
About moving objects he talks that motion can attract the player's attention to the right direction. In other words, we have the contrast of motion vs stillness. As I mention in this site, contrast really has a broad sense and there are multiple forms of contrast.
Vertical e horizontal lines. This is fairly intuitive for most people. The lines guide the player's eyes. If you have a wall with horizontal lines and a wall with vertical lines, the latter is going to appear taller even if it's the same height as the other. The former is going to look wider in comparison. Now extending this to psychology and we have that taller buildings and pointy structures convey a sense of grandiosity and maybe oppression. Horizontal lines would convey a sense of equality and maybe compassion. I really don't have knowledge about this other than some crude observations.
Lighting. Pretty much the same discussion about contrast and I have examples in this site. The most basic concept about colors that children learn at school at some point is warm vs cold. Pretty much every game has it.
Geometry and architecture. More or less the same discussion with textures and lights. One recommendation that Hourences gives and which I agree with is to study existing architecture by copying it. When you copy it you should be able to grasp even a tiny bit of the reasoning behind the structure and how and why it was built the way it was built. This isn't much different from children learning by copying or imitating something.
The same discussion about contrast with lights and textures can be made for geometry itself. Large flat surfaces are to be avoided if possible. In here I'd like to comment that we often have a hardware limitation. Adding more geometry means more triangles and we can't escape the hard limits of how many triangles the hardware can handle. Too much geometry detail can also lead to collision and too much contrast issues. I believe that engineers and architects face similar issues because sometimes what you want to achieve is not feasible due to one or more constrains (political interference, money, time, geographical location, resources, etc).
Unity and connectivity. Hourences himself tells that one huge issue is the lack of inter communication between gameplay, concept art, artists, designers, so on. Each piece must be made in a way that that the sum of all parts is additive, not conflicting with each other. The same can be said about architecture. Avoid making decisions where each piece is disjoint from the others. For example: a building which is mostly cubic with round windows; palm trees in the middle of the desert; a chinese portal in front of a modern building; etc.
Mark Rosewater talks a lot about this issue in Magic. The mechanics of a card have to match the card's theme and purpose and the card must be part of a bigger world. I could even argue that writing tasks and essays, scientific articles, have the same issue. What you write has to be meaningful. The paragraphs have to be cohesive and the whole text must be coherent. If you begin an essay with the argument that point A is true, but at the conclusion you end with "not so sure" or "point A is false". Your essay is lacking coherence and probably cohesion too. Extending this to architecture and we have issues such as visual pollution, undesired asymmetry, pillars that don't look and/or feel strong enough to support the structure's weight, brick walls that are too thin, so on. All that could be called a sort of an "uncanny valley" of architecture.
Stretching cohesion even further I could argue that if the game suffers from lack of cohesion between its elements, we could say that the people behind it lacked cohesion between themselves. Now this is a very hard task to handle because managing people is hard. Managing oneself is already hard for a lot of people.
The last point that Hourences makes is about thematic details. Great of examples of this are Max Payne and Bioshock. In both games great care was taken to make sure that the little details are all part of the game's setting. Star Wars games often have this in mind as well. A good example on how to spot inconsistencies here is looking for mistakes in movies. Lots of movies have mistakes related to, for example, using clothes from the 19th century when the movie takes place in the 18th century. Sometimes it can be a very small detail such as a painting from a 17th century artist placed in a room, but the movie takes place 100 years before that artist was even born. Some people are going to notice it sooner or later. Mark Rosewater makes the same comments about magic. When the fans are invested in the game, even obsessed with it, they are going to notice those little details.