>Functional MRI (fMRI) is 25 years old, yet surprisingly its most common statistical methods have not been validated using real data
>In theory, we should find 5% false positives (for a significance threshold of 5%), but instead we found that the most common software packages for fMRI analysis (SPM, FSL, AFNI) can result in false-positive rates of up to 70%
>a 15-year-old bug was found in 3dClustSim while testing the three software packages (the bug was fixed by the AFNI group as of May 2015, during preparation of this manuscript)
>It is not feasible to redo 40,000 fMRI studies, and lamentable archiving and data-sharing practices mean most could not be reanalyzed either
Loads of "this part of the brain is responsible for X" rubbish.
Oliver Price
You guys know that this bug was patched over a year ago, right? Talk about slow poke news.
Liam Russell
What's going on here?
James Diaz
Edited by Emery N. Brown, Massachusetts General Hospital, Boston, MA, and approved May 17, 2016 (received for review February 12, 2016)
Benjamin Long
> over a year ago > 20 years of "studies" are "not" effected at all
Luis Hill
>Talk about slow poke news The news is that the majority of results generated in the field have been wrong for over a decade and there's no feasible way to go back and fix them.
And the bug is just one problematic factor in the article. Other more fundamental problems are also discussed.
Elijah Morris
Sunk costs, doesn't really matter except maybe to funders.
In fact this is an opportunity for the field to get more funding and do it right this time.
Levi Bennett
So everything we learned about that is bullshit ?
Isaiah Jackson
Some might still be right, the problem is you can't tell which. Basically we thought the error rate (or rather, false positive rate) was 5% but that was based on faulty assumptions and crappy software. The false positive rate was actually closer to 70%. Whoops.
Things which were suggested by fMRI and later successfully validated by other methods might be safe, though I'd be (based on my own experiences) wary of confirmation bias and double check even those.
Chase Rivera
The excitement here is ridiculous. fMRI being shit is well known, and there are no serious fMRI studies that don't use other methods to validate the results, or claim any specific "this part of the brain is responsible for...", only shitty news sites that have no understanding of the implications of a paper.
Christian Reed
Ok thanks user
Robert Clark
Hey user, is a master degree in CS and prior experience with artificial intelligent good enough to do PhD in computational neuroscience?
Aaron Cooper
Generally yes, depends on where of course. In my university it's so common there's actually a "basic training" class in the neuroscience department where they give CS M.Sc.s and B.Sc.s several month long crash courses in biology, physics, etc. before they start their PhDs
Levi Mitchell
thank user, do you think there will be any disadvantage moving from CS to neuroscience?
Camden Cox
Clueless biologists whose stats and compsci knowledge begins and ends with Excel (i.e. most biologists) in your department will swarm you with their problems. But hey it might get your name on some of their papers plus it's a good networking opportunity.
t. bioinformatician
Kevin Robinson
Also prepare to cry over the quality of scientific software.
Which I suppose is another opportunity to create something less shitty. Perhaps commercially after quitting academia? Food for thought.
Aaron Hill
nice to know I'll get the chance to get laid :^) . thanks user.
Jacob White
Becoming somebody's p
James Thompson
> not taking the chance to ask the qt out for a date are you kidding me user?
David Carter
Just make sure to check which part of your brain lights up when they reply ()
Carson Wilson
See >Building on his interests in belief and religion, Harris completed a PhD in cognitive neuroscience at UCLA.[22][27] He used fMRI to explore whether the brain responses differ between sentences that subjects judged as true, false, or undecidable, across a wide range of categories including autobiographical, mathematical, geographical, religious, ethical, semantic, and factual statements.[82]
>In another study, Harris and colleagues examined the neural basis of religious and non-religious belief using fMRI.[83] Fifteen committed Christians and fifteen nonbelievers were scanned as they evaluated the truth and falsity of religious and nonreligious propositions. For both groups, statements of belief (sentences judged as either true or false) were associated with increased activation of ventromedial prefrontal cortex, a region of the brain involved in emotional judgment, processing uncertainty, assessing rewards and thinking about oneself.[27] A "comparison of all religious trials to all nonreligious trials produced a wide range of signal differences throughout the brain," and the processing of religious belief and empirical belief differed in significant ways. The regions associated with increased activation in response to religious stimuli included the anterior insula, the ventral striatum, the anterior cingulate cortex, and the posterior medial cortex.[83]
Jose Phillips
There are plenty, and it will really change depending on what you plan on doing. You will have a lot of learning to do probably, both to get deeper understanding of what you are working on (clinically, anatomically, molecularity etc.) and until you learn how to plan experiments and do medical/biological research. Those are problems every PhD faces, but moreso when you come from a different field. From what I've seen from computer scientists in my lab, you will might only get handed down missions for a while. I'm a bioinformatician as well (MD-PhD) and I can tell you that is definitely not my experience, since there are plenty of other bioinformaticians where I work.
Carter White
it's part of life. men get fucked either way though.
Jack Morales
Also, remember a few things: 1. Most girls are already engaged, married or in a long relationship at this point (at least in my country). 2. If you need people asking for your professional help in order to meet girls, you are probably not their first choice. 3. In the long run, everyone agrees workplace romance is horrible. Horrible.
Landon Rodriguez
you user just poured cold water on my enthusiasm. trash talking aside thank for the insights. I have no problem running errands at first, at least that's 70% of my first year in MSc.
Sebastian Robinson
Fair enough. In my case (I'm ) when I started my PhD four years ago there were only two bioinformaticians in the department (my advisor and a postdoc who left a year later).
Now I'm one of two postdocs at the same lab with three PhD students around besides.
Cooper Bell
>(based on my own experiences)
do tell user
Owen Morris
not that user but people tend to bend over backwards to make data 'fit' and fiddle much more with results which don't fit but by all expectations should
Jeremiah Smith
>new shit
lel dead-fish/voxel should have been the wake-up call.
Sebastian Brown
The dead fish article was exhorting everyone in the field to use a Gaussian Random Field based approach to reduce false positive rate. Which people did and it is now standart practice (it was in use in the dead fish days too but not by everyone).
THIS article shows that the GRF was applied incorrectly.
If anything the dead fish publication contributed to the problem by pushing the use of a spurious veneer of statistical legitimacy.
Jaxson Collins
>pushing the use of a spurious veneer of statistical legitimacy
Isaac Hernandez
lololol is it considered fedora to use obscure langauge?
Hudson Thomas
So fMRI studies are all invalidated now? What are the implications for the field of cognitive neuroscience? Will they be the laughing stock of science? Or did they not have anyhting to lose in the first plcae?
Nicholas Carter
Yes when it's completely unnecessary to get your point across.
John Ward
See If anything it's a great opportunity to push the field forwards.
Wyatt Sullivan
>lamentable archiving and data-sharing practices That's pretty harsh language for a publication.
Jace Gray
fMRI has spawned a plethora of useless neuroscience.
We expose the subject to stimulus X and this part of the brain lights up. Hurr durr we understand consciousness now.
Neuroscience needs mathematicians and engineers if they want to get anywhere, not glorified Bio Majors
Perhaps the most globally appreciated prank to ever make use of an fMRI scanner was brought to the attention of the neuroimaging community at the Organization for Human Brain Mapping’s annual conference. During the final lecture on June 22, the past chair, Rainer Goebel, delivered his “closing comments and meeting highlights” to a full auditorium in San Francisco. After reviewing many of the emerging directions in the field, he displayed what he described as one of his favorite posters from the conference: “Neural Correlates of Interspecies Perspective Taking in the Post-Mortem Atlantic Salmon: An Argument for Multiple Comparisons Correction.” It was greeted with a cathartic laughter of recognition. The abstract, by Craig Bennett, Michael Miller and George Wolford (Bennett, Miller, & Wolford, 2009: later to include Abigail Baird on the poster), described a study of social cognition in “one mature Atlantic Salmon ( Salmo salar ).” In keeping with scientific punctiliousness, and no doubt to thwart any appropriation of their study by the overzealous, the authors then noted: “The salmon was approximately 18 inches long, weighed 3.8 lbs, and was not alive at the time of scanning.” The task paradigm was delivered with the familiar laconic methods section: The task administered to the salmon involved completing an open-ended mentalising task. The salmon was shown a series of photographs depicting individuals in social situations with a specified emotional valence. The salmon was asked to determine what emotion the indi- vidual in the photo must have been experiencing. Stimuli were presented in a block design (Bennett et al., 2009)
Liam Price
Daaaamn. Someone bad a SHITLOAD of money on this short sale. I bet it was the same people who released the info. I would have shorted the shit out of any companies using this data.
Adam Torres
See
Jaxson Torres
>lamentable archiving and data-sharing practices mean most could not be reanalyzed either This is what pisses me off. We generate all this data but it is almost never shared. There could be tons of value here.