The New Mind Readers: What Neuroimaging Can and Cannot Reveal About Our Thoughts by Russell A. Poldrack (review)

This book provides a good introduction to functional magnetic resonance imaging (fMRI). It provides a basic understanding of what fMRI is and what it has been able to accomplish. fMRI is certainly a valuable tool in the quest to understand the functioning brain. What it can do is identify neural structures that are involved in certain mental activities, a necessary first step.

The general finding is that a mental function typically involves different areas of the brain, interacting in various unknown ways. As a result, fMRI is a useful tool for mapping connections between structures and understanding the functional connectivity of areas of the brain. It is not possible to infer a one-to-one correspondence between activity in one brain area and a mental function. The author clearly recognizes this and describes very fully the problem of these kinds of false inferences from fMRI readings. In addition, he is justifiably critical of various simplified attempts to make commercial use of fMRI in lie detection and predicting rearrests of criminals. He also has a good discussion of the statistical power of experimental methods and its effect on finding significant effects, although he spends too much space presenting the problem in a social psychology context. While the author talks about statistical power in some fMRI studies, it could have been discussed more fully, when considering experimental fMRI results.

However, fMRI researchers have been able to predict, with better than chance accuracy, what a person is thinking about when the researchers have limited the possible thoughts the person can have, though still not perfectly and still not when there is an unlimited number of thoughts possible. They are also beginning to decode the fMRI data to reconstruct images, intents and other thoughts. However, there is still a long way to go. Therefore, fMRI researchers are not yet able to read minds in general. Whether the fMRI or its future refinements will provide the tools necessary to overcome these limitations is still far from clear. For me it would be better to emphasize mapping the brain rather than mind reading in the book title.

The book is written in a very personal style. Using this personal style, the author seems to be overly concerned with image—both his own and the field's. There is a kind of cheerleading tone to much of the writing. For example, he describes an experiment in which the intention of subjects to add or subtract a set of numbers was predicted using fMRI data. The author reports that they were able to predict what the subjects were going to do with 70% accuracy—certainly better than chance at 50%. However, the author describes it as "powerful" (pp. 69–70). I do not dispute the result, but some humility would appear to me to be in order. He seems to mention Harvard, MIT and similar prestigious institutions as well as papers appearing in journals such as Science and Nature more often than necessary. He even suggested that one researcher was sure to win a Nobel Prize. The author may feel that this cheerleading is necessary given the major statistical error that fMRI researchers, early on, made (now apparently corrected) where they failed to correct for the use of multiple correlations in the data analysis, which resulted in exaggerated effect size.

In Chapter 1 the author provides an overview of how neurons function. However, in this description he talks only about excitation and does not mention that the output impulse pattern of neurons, in general, results from the interaction of excitatory and inhibitory inputs, by means of different synaptic transmitters from different input nerve cells. Even though fMRI data do not seem to be able to decode this distinction between excitatory and inhibitory influences, mentioning these influences would help the reader appreciate the complexities involved in understanding brain functions and the limitations of fMRI...