Resolving conflict in the medial frontal cortex

 Image courtesy Wikimedia Commons.

Image courtesy Wikimedia Commons.

What does any part of the brain do? This simple question remains largely unanswered in cognitive neuroscience, where researchers are charting out the functional territories of the human brain. As reports of activation have accumulated for the medial frontal cortex (MFC), a swath of gray matter buried in the frontal lobe, a key challenge to the mapping efforts has come to light. From holding letters in memory to viewing images of spiders, an assortment of processes that are seemingly unrelated have been pinned down within its bounds.

The apparent conflict between functional maps of the MFC is what we should have expected all along from the history of the field. The parts list for the human mind was defined in psychology before much was known about brain organization. Likewise, anatomical maps of the brain were drawn without respect to mental functioning. In the case of the MFC, it is possible that the diverse functions mapped to this region involve a shared process that is common to the structure as a whole. For example, cognitive tasks may be so dull or demanding that they draw out the same not-so-pleasant feeling as hearing a baby cry or touching a hot stove. Alternatively, the MFC may need to be sliced into finer sub-structures before a neater pairing with mental functions can be achieved – but where to draw the lines?

  Figure A.  The hierarchical design employed by Kragel  et al.  compares the similarity of activation patterns in the medial frontal cortex (MFC) between subjects within two studies (S1 and S2), studies within subdomains, and subdomains within domains. The anatomical parcellation from Kragel  et al.  illustrates how the MFC was split into six sub-regions.

Figure A. The hierarchical design employed by Kragel et al. compares the similarity of activation patterns in the medial frontal cortex (MFC) between subjects within two studies (S1 and S2), studies within subdomains, and subdomains within domains. The anatomical parcellation from Kragel et al. illustrates how the MFC was split into six sub-regions.

  Figure B.  Neurosynth topics involving pain, negative affect, and cognitive control all predict activation within the MFC (blue), but only pain and negative affect are predicted by activation in the same areas (red). Topics were selected from the 50-topic solution. The MFC is outlined in white.

Figure B. Neurosynth topics involving pain, negative affect, and cognitive control all predict activation within the MFC (blue), but only pain and negative affect are predicted by activation in the same areas (red). Topics were selected from the 50-topic solution. The MFC is outlined in white.

In a cleverly designed meta-analysis, Kragel et al. sought to resolve which mental functions have spatially separable representations within the MFC. They began by pooling together 18 fMRI studies of three mental functioning domains – cognitive control, pain, and negative emotion – which had all been previously related to MFC activity (Figure 1a; Figure A). To make their design hierarchical, each domain included studies that spanned three subdomains. For instance, within the domain of negative emotion, studies in the visual subdomain showed their subjects unpleasant pictures, while studies in the auditory subdomain used sounds. By looking for activations common to a lower level of the hierarchy, Kragel et al. could determine whether a domain had a generalizable representation within an MFC region, meaning that the pattern of activations was shared across its subdomains. Furthermore, through comparisons at the upper level, they could identify mappings that were distinctive of each domain.

Kragel et al. found that some domains but not others could be mapped uniquely to areas of the MFC. The middle cingulate displayed an activation pattern that was shared across studies of pain and not the other domains (Figure 2). In the ventromedial prefrontal cortex, activations were common across studies of negative emotion and were domain-specific. There was no region in the MFC that showed a consistent pattern of activation across cognitive control studies. This may mean that even though we experience processes like remembering letters and choosing a button to press as relatively similar, they are supported by differing neural systems. It may also be the case that these mental functions involve regions beyond the MFC, or that they were not well isolated by the cognitive tasks.

The work by Kragel et al. illustrates how combining fMRI scans in a well-controlled meta-analysis can help settle discrepancies in the field. Thanks to online platforms for automated meta-analysis, it is now possible for anyone with an internet connection to summarize findings from thousands of neuroimaging studies. On the Neurosynth site, you can generate probability maps of mental functions similar to those assessed by Kragel et al. (Figure B). The data are in your hands to determine where cognitive control (or a more neurobiologically valid subdomain) is localized in the brain.

Edited by Kristin Muench

References:

Kragel, P. A. et al. “Generalizable representations of pain, cognitive control, and negative emotion in medial frontal cortex.” Nat. Neurosci. 21, 283-289 (2018), dx.doi.org/10.1038/s41593-017-0051-7.