To the Editor:

I am not a social scientist, nor am I a practitioner of functional magnetic resonance imaging (fMRI), but I am expert in MRI, so I feel qualified to offer unbiased criticism of your essay in the July 19, 2004, issue of The Scientist. You have written a factually inaccurate and dangerously oversimplified bit of fluff that, for the sake of a moment's amusement, holds up to ridicule the work of many serious scientists. Certainly, there have been abuses and overinterpretation of results, but we all remember when sequencing the genome was going to answer all of our questions about medicine.

Let me be explicit about the ways in which you were wrong or wrong-headed:

1) Putting "fake" in the title is simply irresponsible, and I hope you get the avalanche of mail that you deserve about this.

2) You make the blanket statement that fMRI is not science, but this judgment is critically dependent upon how you define science. You are presumably using your own personal definition of science, but you never make that definition clear. In contrast, my American Heritage Dictionary defines science as "the observation, identification, description, experimental investigation, and theoretical explanation of natural phenomena," or any "methodological activity, discipline, or study." By these far better established criteria, fMRI is certainly a science, even if it sometimes lacks explanatory power.

3) Reading fMRI results is no more interpretative than reading the results of a microarray analysis or a photographic plate of stars in the night sky. In fact, the statistical basis of fMRI is far more advanced than the statistical methods that are used in many other fields that you might acknowledge as "science."

4) The example you use, of measuring blood flow in the brain while someone is placing a bid for a Hello Kitty lampshade on eBay, is idiotic. A typical fMRI experimental paradigm involves a simple task repeated many times, so that many separate replicates can be summed. Perhaps placing a bid on eBay is simple-minded, but it would need to be repeated many times in order to yield a detectable signal.

5) The fMRI method does not "measure blood flow." It measures a disturbance to a magnetic field that is thought to arise from the differing magnetic properties of oxygenated and deoxygenated blood. But this is just a hypothesis.

6) Many scientists face the problem of being unable to see their subject matter. That doesn't make the subject matter any less real. Particle physicists cannot directly see their nucleons, molecular biologists cannot directly see their genes, chemists cannot directly see their transition states, geologists cannot directly see continental drift; psychologists are no different in being unable to see thought processes.

7) The methods of fMRI are only about 15 years old, so it seems premature to call fMRI a "plain old" method. As we move forward and delineate the strengths and limitations of fMRI, I believe that it will mature into a plain old method, because it provides insights that are unavailable by any other means.

R. Grant Steen, PhD

Associate Professor, Department of Psychiatry

University of North Carolina

Grant_Steen@med.unc.edu

To the Editor:

Sam Jaffe points out the cold hard truth about the recent spate of functional magnetic resonance imaging (fMRI) research in which researchers attempt to link all manner of human abilities with specific areas of the brain: It is questionable science. Even though this research is often touted as the new frontier of behavioral science, there is a small, but growing chorus of critics. For example, Steven Faux of Drake University has referred to many of these efforts as "chasing ghosts with Geiger counters." William Uttal of Arizona State University has gone further and called the general trend of trying to localize cognitive functions within the brain "the new phrenology."

Jaffe implies that the subject matter of all behavioral scientists is the brain. In reality, their subject matter is not the brain, but rather behavior—hence behavioral science. Of the many possible explanations of behavior are neural processes. And for scientists interested in such processes, there are ways, as Jaffe suggests, of more directly studying the brain than using fMRI.

However, the ultimate causes of our behavior (and of the neural structures and processes that underlie it) are to be found in our shared evolutionary history (as coded in our genes) and in our unique learning histories as individuals. The latter has been studied as scientifically as the subject matter of any other science—by systematically experimenting on observed and precisely measured behaviors and experiences called learning.

Although it is not as sexy or profitable for grant-funded researchers, basic learning researchers toiling quietly with relatively little notoriety have discovered numerous laws that adequately explain much human behavior and that have been used to effectively solve behavior problems with treatments derived from such laws.

Jaffe and other critics rightly cast reasonable doubt on whether fMRI research will ever attain the same level of scientific success.

Hank Schlinger, PhD

Lecturer, Department of Psychology

California State University, Northridge

hschling@csun.edu

To the Editor:

Sam Jaffe's recent editorial raises doubt that functional magnetic resonance imaging (fMRI) studies of cognitive function reflect real neuroscientific outcomes. He rightly recognizes that to some degree, those colorful brain pictures in the crème de la crème of journals are open to interpretation and subject to the biases of those doing the experiment. But functional neuroimaging has been criticized previously as simply representing nothing more than "technicolor phrenology," so his complaints are nothing new.

However, one issue not raised is that those pretty brain pictures, and the accompanying lists of cognitively induced activation hotspots, are often all the research community sees out of the gigabytes worth of volumetric time series data collected as part of the fMRI study. Its no wonder Mr. Jaffe feels the results from these studies are fake. What happened to all that data? How can one see other potential activated brain areas not shown in the published article? If one does the analysis differently, does he/she obtain different results? How much of that figure is Photoshop and how much is real data?

What might be helpful for fMRI believers and skeptics alike would be if, with the published article in hand, researchers also had access to the raw fMRI data itself—then the claims of the original authors could be validated, novel analytic and visualization approaches could be applied to the data, and the study could be directly compared and contrasted with other data sets to identify homologies across disparate cognitive domains. Moreover, the data could be made available to everyone at the time of publication. With their data open to public scrutiny, cognitive scientists might be less willing to embellish on their results or prone to overspeculate on their meaning.

Mr. Jaffe might be interested to know about the fMRI Data Center (fMRIDC) project at Dartmouth College. Founded in 2000, in conjunction with the Journal of Cognitive Neuroscience and the encouragement of other leading periodicals, this NSF/Keck/NIMH funded project is a publicly available archive of the complete fMRI time series, anatomical MRI volumes, and accompanying study metadata from peer-reviewed articles of human cognition. Researchers worldwide can request the data from a growing collection of studies and have it sent to them at no cost for use in new research and education. Several new study re-analyses have been published in the last few years using data from this resource, underscoring the value of the original research findings and broadening their scope by generating new, testable hypotheses worthy of further examination.

Efforts like the fMRIDC are an initial step in the organization and cataloguing of the large quantities of imaging data collected in these studies. It represents one attempt to help researchers be more critical about their work and leverage the data of others into new and exciting brain research.

Cognitive neuroimaging is still a young endeavor, and there is no doubt that there is still considerable ground to cover. The field is gaining in its maturity as a science, and the researchers in this field recognize this more than anyone. Despite Mr. Jaffe's grievances, fMRI is a remarkable technology that is improving understanding of fundamental mental activities in normal and patient subjects. However, by backing up published fMRI articles with the complete data going into them, openly available to all, the maturation of cognitive neuroscience toward legitimate "hard science" status is off to a good start.

John Darrell Van Horn, PhD, MEng

Research Associate Professor

Operations Director, The fMRI Data Center

Department of Psychological and Brain Sciences

Dartmouth College

John.D.Van.Horn@Dartmouth.EDU

To the Editor:

Thank you for your article in The Scientist. I needed the laugh, and the perspective. Although I am not a scientist, and never was, I was rather rigorously trained in medical technology as an undergraduate, and I developed a skeptical eye and a thirst for good, hard data. I learned that even good, hard data had to be interpreted and applied with caution. Now, as a graduate student in liberal studies, I am struggling with a thesis, choking on every assertion I make, looking for something solid, looking for real science. You have reminded me that I am doing something different now and that it is just as slippery as I thought it was back in the days when, test tubes in hand, I chuckled at the very idea of "social sciences."

I will return now to the jello of my thesis and support it as best I can, doubting my data with an untroubled mind.

Cheers!

Lin Edge-Gulick

University of Delaware

lgulick@UDel.Edu

R. Grant Steen et al.