When you walk into an unfamiliar room, you look around. In a fraction of a second, you take in the shape of the room, the furnishings, the light, the ways out, and other clues; very quickly, you make some assumptions about what this room is and how it relates to why you walked in. Then you need to do what you came in to do. Where? How? You might be able to answer immediately—or not. Or maybe you're just distracted by other interesting things in the room.
Similarly, bringing up a web page or opening a window each incur a cognitive cost. Again, you need to figure out this new space: you take in its shape, its layout, its contents, its exits, and how to do what you came to do. All of this takes energy and time. The "context switch" forces you to refocus your attention and adjust to your new surroundings.
Even if you're already familiar with the window (or room) you just went into, it still incurs a cost. Not a large cost, but it adds up—especially when you figure in the actual time it takes to display a window or load a page.
This is true whether you're dealing with web pages, windows, dialog boxes, or device screens. The decisions that users make about where to go are similar, whether they use links or buttons—labels still need to be read, or icons decoded, and the users still will make leaps of faith by clicking on links or buttons they're not sure about.
Knowing that there's a cost associated with jumping from page to page, you can understand ...