The Implicit Association Test  is a sorting task which reveals something about our automatic, non-deliberate, associations .
The part of the test which betrays our automatic associations is a combination of two simpler sorting tasks. Both simple tasks involve sorting words and pictures into categories which are assigned to the left and right (by pressing the E and I keys, which are on the left and right of your keyboard). One task is to sort words (like ‘love’, or ‘failure’) into the categories ‘good’ and ‘bad’. The other task varies depending on what you want to detect automatic associations about. In the ‘race IAT’ the task is to sort pictures of the faces of white americans and the faces of black americans. The race IAT isn’t the only version, but it is the most (in)famous (you can also do the IAT on fat vs thin, arab-muslim vs non-arab-muslims, for different US presidents and in many other variations). The compound task involves sorting both words and pictures to the left and right where each side has two categories assigned to it – so ‘good’ and ‘black american’ on the left, and ‘bad’ and ‘white american’ on the right, for example.
What the IAT test does is compare your times for sorting good words when the ‘good’ side is also the ‘white’ side to when the ‘good’ side is also the ‘black’ side (and vice versa for sorting bad words, and for sorting white and black faces to the good and bad sides). By doing these comparisons the test can detect any evaluation of ‘white’ or ‘black’ as positive or negative that is affecting your time to classify the words or faces to the correct side. So, for example, if you take significantly longer to sort good words to the ‘black’ side than you do to the ‘white’ side then the result is an automatic preference for ‘white americans’ over ‘black americans’ 
What the Racial IAT indicates is that most Americans have an automatic preference for whites over blacks. Two things are important about this. First it isn’t really clear what mechanisms lie behind the effects found in the test (‘Voodoo’ is one suggestion!), nor is it clear what they mean . Second, the automatic preference shows up for most people, even in those who consciously express no race preferences and even in many black americans.
Now where did this automatic preference come from? It certainly can’t be deliberate attitudes, since the bias shows up in people (including many black americans) who have explicitly anti-racist attitudes. Some suggestions have been made, like they are the residual of previously held explicit attitudes, or the result of a ‘cultural bias’ (whatever that means) , but I think a strong, and more likely causal , possibility is that that these preferences are the result of systematic exposure to particular associations (i.e that white = good and black = bad). Associations can become established in memory merely by the repeated co-presentation of two things (conditioning), there doesn’t need to be any logical connection between the two. So if on television the adverts for flash cars and happy domestic scenes always feature white folks and the the crime shows more often have black folks as the bad guys you’re going to absorb those associations.
The researchers running the project imply as much in an answer in their FAQ
…it is very possible to possess an automatic preference that you would rather not have (and the researchers who developed this test are convinced that they, too, fall into this category). One solution is to seek experiences that could undo or reverse the patterns of experience that could have created the unwanted preference. But this is not always easy to do. A more practical alternative may be to remain alert to the existence of the undesired preference, recognizing that it may intrude in unwanted fashion into your judgments and actions. Additionally, you may decide to embark on consciously planned actions that can compensate for known unconscious preferences and beliefs.”
The interesting thing for me about the hypothesis that these automatic preferences develop from repeated exposure to particular associations is that you do not need to believe the associations on any deliberate level, nor do you need particularly to pay attention to them, all you need to do is to have them as part of your environment. In that way our Implicit Associations reflect a part of our minds which belongs as much to the environment of our experience as to ourselves – and, additionally, is as much common to everyone who has shared our environment as it is unique to our individual minds.
And this relates to advertising. Adverts are ubiquitous. Advertising shapes the statistical content of the stimuli we are exposed too, however much we decide to give ourselves certain experiences. Does the IAT give us a glimpse of the consequences we reap from an unclean mental environment? 
References below the fold
 You can get all the research papers here. How wonderful
 I nearly used the word ‘unconscious’ here but couldn’t quite bring myself to. I’m afraid that if i say it three times the ghost of Freud will appear!
 Here’s one example of an interpretation
 That’s the problem with much psychology research. You can find factors associated with some phenomenon, but it’s far hard to find what is truly causing it