With the development of functional neuroimaging has come a host of ethical issues. Some of these are classical bioethical issues, including safety (e.g., for scans involving radiation or high magnetic field strengths) and researchers’ obligations when incidental findings of abnormal brain structure or function are observed in research scans. Other issues arise from the unprecedented and rapidly developing ability to correlate brain activation with psychological states and traits.

One of the most widely discussed new applications of functional neuroimaging is based on correlations between brain activity and intentional deception. Most experts believe that fMRI-based lie detection is not feasible in real world situations, although a number of different research groups have identified fMRI correlates of intentional deception in laboratory tasks. Among the brain areas most reliably activated in association with lying are the anterior cingulate cortex, which is also typically involved in tasks that evoke cognitive conflict, and prefrontal areas important for holding task contexts in working memory and retrieving long-term memories. This makes sense, in that deception requires making responses that conflict with what the liar knows to be true, and may also require holding the fabricated version of the truth in working memory or retrieving it from long-term memory. Whether deception has a specific brain signature, distinct from other effortful processes involving cognitive conflict and memory, remains to be determined. Nevertheless, two companies have been formed to offer fMRI-based lie detection commercially: Cephos and No Lie MRI.

An older technique for studying brain function, event-related potentials (ERPs), has come closer to detecting deception in the real world, although most experts believe that it too stands in need of additional validation research. The technique is based on the detection of “guilty knowledge,” measuring correlates of familiarity with items that only the perpetrator would know are associated with the crime. Brain Fingerprinting, as the technique is called by its developer, has been admitted as evidence in court and is being promoted as a means of screening for terrorists.

Brain imaging can also be used to measure correlates of people’s desire for certain products, an application that has been called “neuromarketing.” To the extent that neuroimaging can measure unconscious motivation to buy, it provides a valuable new kind of information for marketers.

The use of brain imaging to “read” mental states has been greatly enhanced by the application of pattern classification techniques developed in a branch of computer science known as machine learning. Conventional image analysis involves spatial smoothing, that is, averaging the signal strength measured in nearby voxels (3-dimensional pixels) to obtain a more reliable measure of the overall level of activity in a region. An alternative approach is to analyze the pattern of variability across voxels in the unsmoothed image. Although some of that variability is noise, some of it is the result of activity in different groups of neurons, and as such it bears information about the state of the brain. The utility of this approach has recently been demonstrated in various ways, for example by reading subjects’ intentions to perform one of two tasks and by discriminating true statements from lies.

Brain imaging can also provide information about more enduring mental traits, an application that is in many ways analogous to genetic information. Like genotyping, “brainotyping” can reveal information about mental health vulnerabilities and predilection for violent crime. Unconscious racial attitudes are manifest in brain activation. Sexual preferences can in principle be determined based on the finding that sexual attraction and even the attempt to suppress feelings of attraction have neuroimaging correlates. A growing body of literature has investigated the neural correlates of personality using brain imaging, including extraversion and neuroticism, risk-aversion, pessimism, persistence and empathy, and functional imaging correlares have also been found for various aspects of intelligence.

Of course, current functional neuroimaging cannot determine personality or intelligence with any precision -- nor can genotyping for that matter. Brain imaging is at best a rough measure of personality. But this is not to say it is uninformative even in its current state of development, or that it will not improve in the coming years.

These capabilities of brain imaging, actual and potential, raise a number of ethical issues. The most obvious concern involves privacy. For example, employers, marketers, and the government all have a strong interest in knowing the abilities, personality, truthfulness and other mental contents of certain people. This raises the question of whether, when, and how to ensure the privacy of our own minds.

Another ethical problem is that brain scans are often viewed as more accurate and objective than in fact they are. Many layers of signal processing, statistical analysis and interpretation separate imaged brain activity from the psychological traits and states inferred from it. There is a danger that the public (including judges and juries, employers, insurers, etc.) will ignore these complexities and treat brain images as a kind of indisputable truth.

Martha J. Farah