Implicit Attitude Test: What Is It Good For?

Let’s imagine you are glucose intolerant.

We have two psychological tests that we want to use to predict the choice that you will make when you are presented with two items of food. One is the implicit attitude test (IAT), the other is a straightforward explicit test of preferences — i.e. which of two options would you select?

IAT test and results

In the IAT test, you are presented with two pictures: a chocolate cake and a salad.

Set 1: On one set of trials, pictures of chocolate cakes are paired with positive words (happy, tasty, joyful), and pictures of salads are paired with negative words (unhappy, tasteless, joyless).

When taking the test for this set of trials, you have to respond as quickly as possible in the following way: Press button X on the keyboard every time you see a picture of a chocolate cake, and press X when you see a positive word. Press button M whenever you see a picture of a salad, and when you see a negative word.

Set 2: In the second set of trials, chocolate cake is paired with negative words (unhappy, tasteless, joyless), and salad is paired with positive words (happy, tasty, joyful).

When taking the test for this set of trials you have to respond as quickly as possible in the following way: Press button X on the keyboard every time you see a picture of a chocolate cake, and press X when you see a negative word. Press M whenever you see a picture of a salad, and when you see a positive word.

The scores show that you are quicker to respond to positive word associations with chocolate cake, and are slower to respond when negative words were associated with chocolate cake (and vice versa for salad). The test suggests that you have a stronger positive association with chocolate cake and so the prediction is made that you will more likely pick it over the salad. In fact, when the options are presented for real, you pick the salad.

Explicit test and results

In the explicit test, when you are presented with two pictures, one of chocolate cake, and the other of a salad, and asked which one you would actually eat, you choose the salad. This is used to make the prediction that you will pick the salad over the chocolate cake, and this is indeed what happens when faced with those options for real.

What can we make of the results of each test and their predictions?

This is of course a fictional and somewhat exaggerated depiction of an IAT and an explicit test. Nonetheless, the idea is to illustrate some critical points.

The IAT is an indirect test, it is designed to reveal personal “unconscious” attitudes or personal unconscious beliefs. The reason for using this test, at least in this illustration, is to gauge the level of difficulty someone has in selecting options that are best for their health. It might be that we find it difficult to say out loud what we are really thinking because of how we will be judged, and so examining this indirectly is potentially more revealing of what we think than simply asking us directly.

But, in the illustration, and in actual work in the psychological sciences studying the IAT, it isn’t clear whether the IAT does accurately measure personal “unconscious” attitudes. Even where it might, it doesn’t reliably predict how people will actually behave (Oswald et al, 2013). So what does the IAT tell us?

Some argue that it acts like a mirror that reflects common associations that are present in the world (Fielder, Messner & Bluemke, 2006; Karpinski & Hilton, 2001), while others have contested this. Nonetheless, most agree that there are many factors that influence the way people respond on the IAT that has little to do with the purported “unconscious” biases the test is claimed to measure (De Houwer, et al, 2009). Also, if you ask the public what they think of the IAT, they reveal degrees of skepticism, and even if they are on board, they nonetheless want to know about its credibility (Yen, Durrheim & Tafarodi, 2018).

Does it make sense to use a test that is neither reliably reflective of an individual’s personal attitudes, or that isn’t reliably predictive of their actual behavior, in order to then make claims about an individual’s priors, their attitudes, and their behavior? The answer to this is based on whether this resides as an issue for basic science, or applied science.

Unconscious Essential Reads

How Your Unconscious Mind Guides Your Behavior Every Day

Human See, Human Do: We Imitate, Even if Others Dislike It

Psychological investigations of the IAT focus on refining and improving the tool so that it measures what it was designed to measure. This is what we expect. Science progresses towards improvements in measuring phenomena. But, what is concerning is how these tools are used in the real world. If the IAT is treated as an assessment tool in organizational settings, then there needs to be a disclaimer. It is not 100% accurate, in fact, it isn’t clear what percent accuracy we can attach to it. And if you take it and are shown to possess “unconscious” biases, then skepticism shouldn’t result in being labeled as having “blind-spot bias” or irrational denial.

We are all in possession of biases. They can be transitory as well as pervasive. Some are benign and some are consequential. The nature of our cognition is such that we operate with them, and to navigate the world successfully we learn to regulate their effects, particularly in light of negative consequences.