Rather simple, short paper - I will present it as such without going too much into fMRI methods for estimating correlated spontaneous noise.
Introduction
It is well known that the brain is organized into large functional networks, and that these networks exhibit correlated activity during both task and rest. Additionally, we know that in primary sensory cortices, there are patterns of local representational homogeneity. For example, recall the “splotchy” patterns of orientation preference across primary visual cortex. Both animal models and fMRI have revealed small scall functional networks in sensory cortex.
Goal
The goal of this paper is to use fMRI to show that the prefrontal cortex is also organized into local heterogeneous functional networks.
Methods
Working in sensory cortex is a lot “easier” to demonstrate local networks, as stimulus parameters can be finely tuned and manipulated across many presentations. The flexibility of the PFC in representing many different task goals and cognitive features is a challenge. Authors focus on inferior frontal sulcuss (IFS), due to previous experiments which suggest IFS represents “task context”.
Subjects were thus scanned while preforming a task with two contexts - either to pay attention to the direction of movement or the predominate color of a collection of random dot stimuli. Thus two contexts: Motion or Color. In experiment 2 another set of tasks are used but the results are qualitatively identical.
The main crux of this paper is that they are going to be preforming all of their analyses on the “residual noise”, left over after estimating task responses. Previous work suggests this is the best way to reveal functionally organized areas of cortex.
Results
- Demonstrate above chance decoding in in IFS for task context in both experiments
- Demonstrate reliable (permutation-tested) population of voxels that preferentially represent one context over the other
- note: the majority of voxels show no strong preference, but there were enough that did to support their analyses
- Demonstrate in both experiments that there is reliable spatial clustering in IFS
- proximal voxels were able to reliably predict changes in “residual” activity above distal voxels
- this analyses was model driven
- MDS (a form of PCA) analysis of context preferences again confirm the presence of functionally distinct subnetworks
- Furthermore, MDS of task and resting state (exp. 2) show that these networks existed prior to task and the overall strength could not be explained by learning effects
- this is strong evidence!
Conclusions
This paper provides evidence for distinct functional subnetworks in the PFC. One interesting implication of this finding is the connection to attention and working memory. The authors suggest ceiling present in general online cognitive capacity could be a resulting feature of set number of functional subnetworks availble in the PFC.