February 24, 2014, by Robin Scaife
What is “implicit” about implicit biases?
The discussion following Jules’ post about “awareness of implicit bias” prompted me to spend some time looking into how the term implicit is defined in the literature. What I found was a lack of consensus and often a lack of clarity surrounding the use of the term implicit.
The term “Implicit” in psychology originally comes from research on memory where it refers to the agent not being aware of the memory which is the source of a particular piece of information. In this origional sense the proper use of the term implicit when referring to biases would be the agent being unaware of the source of the bias. However, this is only one of a large number of ways in which implicit is now used when discussing biases.
The two features which are typically considered to be the central features of implicit are Lack of Control and Lack of Awareness. At this level the issue does not seem to be that complicated. However there is little agreement over whether one (and if so which) or both of these features are required for a bias to count as implicit. Furthermore, there are multiple ways in which to unpack each of these central features.
Lack of control might refer to:
a) the agent not being able to prevent themselves from having a particular bias
b) the agent not being able to prevent a particular bias from influencing their behaviour
Note that “not being able” in the definitions above does not necessarily mean that it is impossible for the agent to control the bias just that the agent cannot control the bias directly via simply exercising their willpower.
When we consider lack of awareness there are even more options:
a) the agent lacks awareness of the bias
b) the agent lacks awareness of what caused them to have the bias
c) the agent lacks awareness of the fact that the bias influenced their behaviour
d) the agent lacks awareness of the way in which the bias influenced their behaviour
While some authors are precise about what they mean by lack of control or lack of awareness many of the papers I have read make no attempt to clarify between these different options. I get the impression that the Implicit Association Test is so well established as a methodology that they do not feel the need to explain what it is which makes the attitudes it measures implicit. In fact some authors state that it is just the way in which the attitudes are measured which makes them implicit. So for them an attitude counts as implicit as long as the way in which it was measured does not rely on self-report.
I am certainly not the first person to notice this lack of clarity in the literature on implicit attitudes. For example, De Houwer and Moors (2007) point out that despite the increase in popularity of research on implicit attitudes and processes “relatively little attention has been given to clarifying the meaning of the concept implicit.” They claim that researchers are using implicit as a synonym for automatic which is itself defined in terms of a number of features that do not always co-occur. For this reason they “recommend specifying the automaticity features one has in mind when using the term implicit.” From my experience of the literature this sensible recommendation has not caught on nearly as much as I would like.