The New Mind Readers: What Neuroimaging Can and Cannot Reveal about Our Thoughts Russell A. Poldrack Princeton University Press (2018)

Since the advent of neuroimaging in the 1980s with positron emission tomography (PET), the sight of a living human brain in action has captivated scientists and the public. The emergence of functional magnetic resonance imaging (fMRI) in the early 1990s was a watershed. MRI scanners were already common in hospitals and, unlike PET, fMRI does not expose people to radioactivity. By measuring activity in the brain at the scale of a few millimetres, these scans seem to promise profound insight into the workings of the brain. That has led to wild claims that the technique could enable mind reading — actually knowing a person’s precise thoughts.

Russell Poldrack tackles these claims head on in The New Mind Readers. The experimental psychologist and neuroimaging pioneer takes readers through three decades of fMRI, its promise and limitations. From the race between groups in Minnesota, Massachusetts and Wisconsin in 1991 to show that MRI measures of blood oxygenation can reflect functional brain activity, to the development of techniques for decoding what someone is looking at, Poldrack surveys the history and biological basis of the technique and its potential application in areas as diverse as law and psychiatry.

Poldrack is an ideal guide. As director of the Stanford Center for Reproducible Neuroscience in California, he actively advances fMRI methods. His enthusiasm for them is clear, as is his frustration at how their data have been misinterpreted and abused.

The technique has revolutionized neuroscience. Thousands of fMRI studies are published each year on topics ranging from perception to decision-making. For example, we now know that the pattern of blood flow to the fusiform face area in the temporal lobe can indicate that a person is looking at a face instead of a ball; and that imagining playing tennis or walking around your house, say, elicits activations in different brain regions. That is a major advance for neuroscientists and physicians who work with people in apparently non-responsive states after brain injury. It means they can identify patients with conscious awareness simply by asking them to engage their imaginations.

But some claims for fMRI are exaggerated. In 2007, The New York Times published an article based on fMRI data collected while people viewed images of candidates in US presidential primary elections, such as Barack Obama and John McCain. A group of neuroscientists at the University of California, Los Angeles, and political scientists had interpreted the results, alleging that they revealed how swing voters felt about the candidates.

As Poldrack explains, the trouble is that activations of particular brain regions — such as the amygdala and the insula, which have been associated with fear and disgust, respectively — are not uniquely associated with particular mental states. One region, the anterior cingulate cortex, was found to be active in about one-quarter of thousands of studies that Poldrack and his colleagues examined, including those involving pain, short-term memory and cognitive control. So, ‘reverse inference’ of thoughts from brain-activation patterns can be very misleading. The potential for outlandish claims is high, Poldrack shows, when scientific data are used to support political and commercial interests — for example, when companies promote the ability to detect lies or to evaluate how viewers respond to advertising without sufficient scientific rigour. The New Mind Readers is a valuable example of how science can be discussed clearly and even-handedly, without sensationalism.

One of Poldrack’s key themes is that interpreting fMRI findings demands an understanding of the underlying data and how they were produced. These scans do not measure neural activity directly. They rely on changes in the magnetic properties of haemoglobin (depending on levels of oxygen), to reveal local differences in blood flow. These reflect neural activity and are associated with different mental states, such as increases in the activity of motor cortex while tapping the fingers. When a technique involves hundreds of thousands of measurements across the brain, it is challenging to distinguish between a real change and a chance observation. Concerns over reproducibility are prominent. Moreover, many experiments use only a small sample, of fewer than 20 participants, often university students, in a laboratory setting.

Caution is needed about generalizing to more complex, real-world situations. These include driving a car down a busy motorway, or moving from average activation patterns to single brains in a much more diverse general population, ignoring the importance of the individual variability that is part of being human.

The New Mind Readers is personal and selective. Poldrack gives short shrift to some methods, including brain stimulation, in which magnetic pulses are used to alter brain function directly to probe the specific role of the region targeted. He also skimps on neuroimaging techniques such as magnetoencephalography, which directly measures changes in magnetic fields produced by electrical signals in the brain. This is not an exhaustive account, and Poldrack focuses only on key developments and pioneers close to his own work. Yet his idiosyncratic approach is deeply engaging.

I was fascinated by Poldrack’s description of why he decided to scan himself more than 100 times over 18 months to investigate how the brain changes over time — despite enduring a panic attack the first time he went into an MRI scanner. This intensive study uncovered much about the stability of brain function and the factors that affect it (including caffeine, food and mood). Yet Poldrack reveals that he learned “depressingly little” about himself during the experiment, highlighting the challenges of using fMRI for personalized medicine.

At times, Poldrack loses focus. His brief forays into topics such as the nature of mental illness are unsatisfying: they are too brief and lack the clarity of the rest of the book. Nevertheless, this is a compelling introduction that lucidly spells out the risks of taking media reports at face value, and urges readers to dig into the details. fMRI is evolving rapidly and researchers are just starting to map brain activity at sub-millimetre resolution, revealing activity — both in different regions and in different layers of cortex within a region.

Happily, despite the book’s title, Poldrack makes it clear throughout that ‘mind reading’ as most people would imagine it remains in the realm of science fiction. What is much more exciting is the potential of fMRI for providing insight into brain function that will ultimately lead to clinical applications.