Skip to main content

Lessons from a Dead Fish

This article was published in Scientific American’s former blog network and reflects the views of the author, not necessarily those of Scientific American


The revolution in neuroscience is often characterized as a revolution in new imaging technology.

A long overdue reassessment of  neuroimaging machines—in particular the functional magnetic resonance imager—has underlined that what you see is not always what you get.

A study published this year in Perspectives on Psychological Science noted that many papers in social neuroscience, the field that examines the neurobiology of social behavior, suffered from faulty analyses that produced "voodoo correlations" in their data. Separately, an analysis by researchers at the National Institute of Mental Health found that almost half of the neuroimaging studies published in well-known journals, including Nature and Science, contained unintentional biases capable of distorting conclusions.


On supporting science journalism

If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.


The latest entry for what's-wrong-with-this-picture arrived literally with a splash earlier this month when a number of blogs trained their attention on a poster presentation about a dead fish that took place at the Human Brain Mapping conference in June. The poster went under the deadpan name: "Neural correlates of interspecies perspective taking in the post-mortem Atlantic Salmon: An argument for multiple comparisons correction."

Whaat? 

The poster asserted that a salmon placed in an fMRI scanner activated a portion of its brain when the fish was shown photos of humans and asked to identify emotions on their faces, a typical task for human research subjects.  "The salmon was approximately 18 inches long, weighed 3.8 pounds., and was not alive at the time of scanning," the text of the poster remarks.

So are government dollars going to pay for perverse pranks carried out by researchers from Dartmouth, Vassar and the University of California, Santa Barbara?

Not in the least. The researchers used the defunct salmon to make a salient point.  When statistical tests are made of the 130,000 tiny three-dimensional pictures (voxels) that make up an MRI image, some of them are bound to give "false positive" results purely by chance: in this case, a cluster of  pixels just happened to illuminate in the region of the dead salmon's brain.

The researchers used dead-salmon neuroimaging to issue a clarion call for statistical adjustments, "mutliple comparisons correction" or similar techiques, to rid images of false positives. These corrective measures are still ignored by some researchers.

Needless to say, the  blog world went wild. A few writers actually contended that the fish was deep in thought, or that the image was capturing light from its piscine soul, nay sole. But most bloggers just had fun, as Craig Bennett, one of the researchers, pointed out on prefrontal.org. Two comments he received:

"..measurements were right off the scales."

"...the fish wasn't dead, it was just tenured."

The researchers are now trying to get their work published in an academic journal, proving that a picture worth 130,000 voxels is more than just a red herring.

 Courtesy of Craig Bennett of the University of California at Santa Barbara and colleagues

 

 

Gary Stix, Scientific American's neuroscience and psychology editor, commissions, edits and reports on emerging advances and technologies that have propelled brain science to the forefront of the biological sciences. Developments chronicled in dozens of cover stories, feature articles and news stories, document groundbreaking neuroimaging techniques that reveal what happens in the brain while you are immersed in thought; the arrival of brain implants that alleviate mood disorders like depression; lab-made brains; psychological resilience; meditation; the intricacies of sleep; the new era for psychedelic drugs and artificial intelligence and growing insights leading to an understanding of our conscious selves. Before taking over the neuroscience beat, Stix, as Scientific American's special projects editor, oversaw the magazine's annual single-topic special issues, conceiving of and producing issues on Einstein, Darwin, climate change, nanotechnology and the nature of time. The issue he edited on time won a National Magazine Award. Besides mind and brain coverage, Stix has edited or written cover stories on Wall Street quants, building the world's tallest building, Olympic training methods, molecular electronics, what makes us human and the things you should and should not eat. Stix started a monthly column, Working Knowledge, that gave the reader a peek at the design and function of common technologies, from polygraph machines to Velcro. It eventually became the magazine's Graphic Science column. He also initiated a column on patents and intellectual property and another on the genesis of the ingenious ideas underlying new technologies in fields like electronics and biotechnology. Stix is the author with his wife, Miriam Lacob, of a technology primer called Who Gives a Gigabyte: A Survival Guide to the Technologically Perplexed (John Wiley & Sons, 1999).

More by Gary Stix