fMRI Primer

01 Introduction to the Problem

Question: What do these images have in common?

intro-to-the-problem.jpg

Answer: They could each be described by the same sentence in a published paper: “This image shows brain regions that respond to affective words during an emotional stroop (ES) task ( Z-score 1.65, cluster size p=.05).” In short, they were each produced, in separate analyses, by the same independent variable – affective words – applied to the same dependent variable – BOLD signal. If published individually, it would be difficult, if not impossible, for a reader to determine that any given one was flawed. As such, the experimenter would have significant discretion in which to publish, and might well decide to publish the one with ‘better blobs.’ Who would know? Almost nobody.

Follow-up question: What accounts for the differences between these images?
Follow-up answer: The inclusion of additonal, different regressors in the different analyses. As in all multivariate regression, the inclusion of different regressors leads to different findings. Note that the dependent variable – BOLD signal – is the same for each image.

Take home point: Determining which regressors to use in a study is subject to PI discretion, is difficult for people not involved in the study to detect, is open to simple errors (eg omitting a negative sign on a value in a regressor), and has a major impact on published images!

Advertisements

Leave a Comment »

No comments yet.

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Blog at WordPress.com.