Interpreting negative connectivity-behavior correlations

Submitted by jiansh on

Hi everyone,


I am working on a seed-based resting-state analysis, particularly focusing on the relationship between intrinsic connectivity of seed ROIs and a behavioral measure. I was wondering if anyone has experience interpreting negative connectivity-behavior correlations.


Firstly, if for instance, I found that connectivity between seed A and region X is negatively correlated to a behavioral measure, is the appropriate next step to run an analysis without the behavioral measure and mask it with region X to examine the direction of intrinsic connectivity between the seed A and region X (because the intrinsic connectivity could a positive correlation or anticorrelation)?


Then this leads me to the next question: As I obtained several negative connectivity-behavior relationships for my seed ROIs, I was wondering if the strength of connectivity is primarily interpreted uni-directionally (i.e., stronger connectivity always means positive connectivity, and anticorrelations are not meaningfully interpreted) or bi-directionally (i.e., stronger connectivity could imply stronger positive connectivity or stronger anticorrelations). So, which is more appropriate -  "The greater the negative intrinsic connectivity between seed A and region X, the greater the behavior score" or the lower the intrinsic (positive) connectivity between seed A and region X, the greater the behavior score"? - or are both acceptable ways to interpret such negative connectivity-behavior correlations?






YAN Chao-Gan

Mon, 08/31/2015 - 05:42

Hi Darren,

I think investigating how seed A correlates with region X is an important step to interpret the FC-behavior relationship.

I think both are acceptable. 

It's better to make sure if it's negative -> less negative, negative -> positive, or positive -> more positive.