On brain scans and data visualisations

A friend shared this image on their FB newsfeed.

Under the image was the following text

"
"This is the world’s first ever magnetic resonance image showing a mother and child’s bond.
The image is of neuroscientist Rebecca Saxe kissing her two month old son.
The child’s brain appears to be smoother and darker. This is because it has significantly less white matter. White matter is comprised of myelin, which is fatty tissue that acts as insulation for the wires that communicate messages inside your brain."

It looked wrong to me. It is unusual for an magnetic resonance image (MRI) to show activity. So I did a Google reverse image search, Google is not my search engine of preference I use one that doesn't track you for marketing purposes, but there is functionality in Google  not available elsewhere. I found several links to the same story.

Follow the link above and it is the scientist who made the image giving her account, she is also the mother in the image. Note the difference, no activity graph.

This is one of the most famous activity graphs from fMRI. This Wired article explains the story (note the website tracks you and only allows 4 free article views a month).

The researchers scanned the dead salmon's brain as the salmon "reacted" to various images of people interacting. They then compared blood flow as the fish 'looked' at different images. The graph above shows the statically significant differences they found.

The researchers deliberately did not use any techniques for doing multiple comparisons within a set of data, which is what gave the false positive result. This was done some time ago, and it was meant to be a wake up call to researchers in the field to do their statistics more carefully.

Conclusions

A data visualisation is not a scan. By calling it a 'scan' it makes (at least some) people think it is an unmediated picture of reality. When I used the dead salmon in a seminar at least one student could not see beyond the image, she asked if the researchers were sure the salmon was dead, as the 'picture' of the brain clearly showed activity.

I strongly believe that people in general have a bias towards biological explanations rather than psychological explanations. The image without the fake activity graph is cute, showing a proud scientist and her young child. The picture with the fake activity shows "a mother's and child bond" as biological reality.

The mixing of fakery with more factual text (everything after the first line, although the biological bias continues) seems like a particularly pernicious way of lying.