Firstly its important to note that attention basically the outcome of automatic processes, although attention may easily be directed by conscious choices.
As a previous poster mentioned the reticular formation does have a mediating role in attention. However its doesn't really control attention, so much as it does direct it based on processing in other brain areas, Sturm and Willmes (2001) aptly encapsulate this as...
For an interpretation of the findings, we have proposed a network in which the anterior cingulate gyrus and the dorsolateral frontal cortex intrinsically control the brain-stem NA activation system via the reticular nucleus of the thalamus.
The vast majority of attentional processing at the most basic level is scanning the environment, in order to create a simulation in our minds. This simulation is not perfect by any means, in fact it appears that we only pay attention to a few key features and it is only when we focus or are guided to use a narrow area of the perceptual field that we attended to things we might not have seen, such as demonstrated by the phenomenon known as 'change blindness'.
From a perception perceptive, early attentional processing is influence by a number of factors relating to stimuli variability, which may guide the deployment of attention, such as motion, size and colour.
Fig 1. Unguided environment scanning is non selective, feature scanning which requires more processing is selective and aids quick recognition. Searching for an object requires use of both of these systems.
This early feature recognition is strongly associated with activity in the ventral visual system, see fig 2.
Fig 2. Diagram of the dorsal visual stream (non selective) and ventral visual stream (selective).
The excellent work of Prof. Jeramy Wolfe, over views much of this perceptual research. When looking for particular information an individual easily guide there attention to the relevant information, such information may even be primed rather than instructed. The pathways and areas of the brain associated with selective attention are relatively simple: the frontal eye (motor control), super parietal lobe (suppress normal attentional flow to specify new target), intraparietal sulcus (spatial orientation, salience mapping, and short-term visual memory).
Fig 3. Areas associated with attentional control, and orientation (Blue).
Luckily after years of research we understand attentional processing from a perceptual viewpoint rather well, however simple questions are far more complex than they at first seem. The current focus on research are narrowing down these areas, personally I think the more interesting questions regard saliency and hedonics. These questions look at reward processing in the brain, and how attention is implicitly guided, if your interested in this then you should check out Kent Berridge's research, which guides this area of research. Finally I'll just mention that acting on implicit processes seems to depend on executive functioning, and working memory capacity.