Brain and politics 2.1: Russell Poldrack replica a Marco Iacoboni

Puedes seguir la discusion sobre Neuropolitica en el blog Neuroethics and Law

First, Dr. Iacoboni argues that it is possible to use brain imaging data as "probabilistic markers of brain states." I agree in principle, and in fact I have argued this point myself in print (Poldrack, 2006, Trends in Cognitive Science; http://dx.doi.org/10.1016/j.tics.2005.12.004). The critical question is: What are the probabilities? If they are very high then the strategy that Dr. Iacoboni advocates is reasonable, whereas if they are low then that strategy will fail. I asked exactly this question in my TICS paper and tested it using real data. What I found is that while engagement of an individual region does provide some probabilistic information regarding the engagement of a mental process, the added information is relatively weak. I have no doubt that cognitive neuroscience will eventually identify ways to detect mental states using brain imaging data; indeed, there is currently a growing body of research that uses machine learning and pattern recognition techniques to "read" mental states. This work suggests that, to the degree that neuroimaging data are predictive of mental states, this prediction will come both from fine-grained patterns of activity (which are not evident to the human eye) and from the coordinated activity across many brain regions, not from localized activity in specific brain regions.

Dr. Iacoboni's defense of his group's use of a specific reverse inference (the association of amygdala activity with anxiety) depends largely upon the frequency of citation of specific terms in the literature. However, the distribution of terms in the literature is biased by past theories that have driven work in particular directions and which may not reflect current thinking. Dr. Iacoboni cites the fact that there are many more citations for amygdala and anxiety in the literature than amygdala and happiness. However, this largely reflects the fact that 30 years of work has examined the relation of anxiety and amygdala activity, whereas only recently has interest accrued in the role of amygdala in positive emotional responses. In addition, the results of such an analysis are heavily dependent upon the specific search terms; for example, whereas a search on 'amygdala AND happiness' in PubMed (as of June 3, 2008) yields 88 hits, searching on 'amygdala AND reward' yields 585 hits, in comparison to the 1314 hits for 'amygdala AND anxiety'. These results undercut the argument that amygdala activity is as strongly predictive of negative emotion as the reverse inferences in the Op-Ed would require.

Second, Dr. Iacoboni notes that reverse inference is widespread throughout the neuroimaging community, and my colleagues and have used it in published papers. In fact, I published a paper with Anthony Wagner in Current Directions in Psychological Science in 2004 which argued in favor of the careful use of reverse inference. My 2006 TICS paper also attempted to analyze what kind of information reverse inference provides (though in the end arguing that it is relatively weak). Cognitive neuroscience is still a young field, and I do not think that we should forestall the use of any means to better understand how mental processes are implemented in the brain. Reverse inference is one such means, and I still believe that it can be useful, but my analyses over the last few years have made me increasingly skeptical of many of the reverse inferences presented in the literature. The fact that I made such claims in papers published in the past should not discredit my more recent arguments against reverse inference.

Finally, I agree with Dr. Iacoboni that we should use our best scientific methods to address questions of relevance to the larger community, and that we should translate our research in a way that makes it accessible to non-experts. However, although sensationalist presentation of scientific results may result in greater press coverage, it does not advance (and may even hinder) public understanding of scientific research. Our goal should instead be to present research findings to the public in a way that expresses excitement about our discoveries but also acknowledges the fundamental uncertainty of our results and the limitations of our methods. The reaction that my colleagues and I had regarding the NY Times Op-Ed was so strong not because of its use of reverse inference per se, but because it was used so carelessly and presented with absolutely no caveats or qualifications. The portrait of neuroimaging research that this Op-Ed provided to the public bears little similarity to the kind of methodical, thoughtful science that so many researchers in the field are engaged in.