American Journal of Bioethics Neuroscience Publication

30 04 2011

Paul Boshears of the Europäische Universität für Interdisziplinäre Studien and I published an Open Peer Commentary in the American Journal of Bioethics this April 2011.  The article addresses important issues and warning with regard to over-interpreting neuroimaging data.

Here is a draft of the article “Ethical Use of Neuroscience,” but the final publication can be found here.

*************************************************************************************************

Levy’s essay (2011) claims that some intuitions leading to one’s moral judgments can be unreliable and he proposes the use of a more reliable, third party, empirical measure. It is commendable that Levy attempts to work beyond traditional bounds; however, the author’s use of fMRI data is questionable in supporting an argument about intentionality. As neuroscientists, we rely upon evidence-based thinking and conclusions to create generalizable knowledge, and while fMRI data can be informative in broad correlational accounts of behavior, to rely upon these data as reliable measures of intuition is arguably just as speculative as the first-person account. It is deeply concerning that society may attempt to apply these data in the manner Levy describes. Indeed, alarming misappropriation of fMRI and EEG data for commercial purposes and as evidence in criminal cases–thereby establishing legal precedents–has already begun.

Levy brings into question the appropriate context for which to use neuroscience as a tool–specifically in illuminating moral decision making. We share with Levy an enthusiasm for neuroscience, and it is enticing to think that in learning how the brain operates we will thereby better understand how the mind also operates. Problematic is Levy’s belief that fMRI studies demonstrate how brain regions, “function to bring the agent to think a particular action is forbidden, permissible, or obligatory.” This is something that fMRI simply cannot do as it is a technique developed to represent mathematical constructs, not detail physical mechanistic processes. We believe the essay depicts scenarios beyond the limitations of what is truly testable by neuroscience, and this could facilitate unintended unethical applications of neuroscience.

As imaging data has become so compelling and headline-grabbing, we focus on addressing these data. Our concern is that one more professional-sounding voice will influence society with scientifically-unfounded claims of what current technology can do leading to unethical exploitation of neuroscience findings. This is especially important given evidence that simply referencing neuroimaging data can bias the public’s evaluation of papers (McCabe and Castel, 2008, Weisberg et al., 2008). It is, therefore, necessary to outline the limitations of fMRI brain imaging and EEG technologies. What is brain imaging actually? What can it tell us? What are the limitations of how these data can be interpreted?

Functional magnetic resonance imaging (fMRI) and electroencephalography (EEG) are noninvasive techniques that indirectly measure brain activity (for an extensive review see Shibasaki 2008). Magnetic resonance imaging uses electromagnetic fields and radio waves to reconstruct images of the brain. Functional MRI relies on detected changes in blood flow by tracing oxygenated and deoxygenated blood. Changes in blood flow are calculated by statistical software and then colorized in a constructed brain image based upon mathematical modeling. When imaging a brain of a person making a moral decision, for instance, one might identify gross changes in blood flow in some areas versus others. This may be called activation, more oxygenated blood brought in to the brain area of interest, or deactivation, less oxygenated blood. The spatial resolution of fMRI allows fairly accurate reconstruction of activated structures. However, the actual neural activity generating these changes and the origins of the blood flow changes are not identified and could arise from areas centimeters away from the activated or de-activated region (Arthurs and Boniface, 2002).

EEG utilizes a series of sensors affixed to the scalp and these scalp recordings are used to describe electrical fields emanating from the cortex. EEG cannot detect activity of deep brain structures, unlike fMRI, which can detect changes in cortical and deep brain structures. Functional MRI can detect changes within seconds; EEGs can detect changes within fractions of seconds for better time resolution of actual neuronal firing rates. Electroencephalographs have relatively poor spatial resolution, but can be combined with other higher resolution techniques, such as fMRI, to give more informative data. Neither technique has the spatial resolution to detect the activity of individual or specific types of neurons. Rather, these techniques detect networks and groups of neurons on the order of thousands to millions. Knowing which types of neurons are activated can give us more mechanistic information. Based on our anatomical and functional knowledge of specific types of neurons, we can predict where these neurons project in the brain and to what extent as well as what kind of neurotransmitters these neurons release. Knowing the chemical phenotype of a neuron gives us important distinctions.  For example, an activation of excitatory neurons (which release excitatory transmitters) would not have the same effect as an activation of inhibitory neurons. In addition, recent data (Koehler et al., 2009, Attwell et al., 2010) have suggested that fMRI detects blood flow regulation by glia, not neurons (the brain cells classically known to mediate synaptic transmission) bringing to question how fMRI data can alternatively be explained, and what fMRI actually tells us about brain function. Importantly, it is unclear whether the changes neuroimaging data depict are indicative of causative factors or simply after-effects.

Overall, we do not doubt the statistical rigor and analyses of researchers, and our simplified description of these techniques is not meant to devalue or undermine the contributions of neuroimaging data. However, we must remind ourselves that the brain is composed of much more than blood vessels and electrical fields, having more complexity than can be described with neuroimaging techniques alone. We must caution against over-interpretation of these exciting data and call for the responsible incorporation of these studies into interdisciplinary pursuits that aim to describe the human mind.

When considering how any scientific data might translate into something as complex as moral behavior, we can do little more than show correlation. While some brain areas may show some degree of specialization such as in “Reward Pathways,” it is apparent that these brain areas work in concert to serve multiple functions and could not accurately describe exclusive rights to consequentialist-based or emotion-based moral decision making. When interpreting areas of brain activation, we must also consider a variety of functions that each brain region can have. If we could: (1) identify individual cells in the brain as the smallest unit of moral processing that were (2) active exclusively during consequentialist and not emotion-based moral decision making, and (3) describe all of the requisite circuitry–then these data might have applications as the author describes. However, this is not the case. The studies cited in Levy’s paper are purely correlational of a behavior and in no way directly describe the biological construction of specific thoughts, intuitions, or morally (ir)relevant processes. Research has not demonstrated that these brain regions are the sites of generating moral intuitions, nor identified that the essence of morality or intuition is stored somewhere in these neurons. We can make some broad conclusions about what brain regions might be involved in mental states or thought processes from neuroimaging data, but we cannot draw conclusions about moral constitution.

Levy’s invocation of a future constructed “neural signature of intention” from fMRI or EEG data ignores what the brain is designed to do best and what artificial intelligence engineers have the most difficult time re-creating: the brain’s plastic, ever-changing, and adaptive nature. Experimental models of learning in brain cells have repeatedly shown that experience can strengthen or weaken connections between cells and between cells connecting different areas of the brain, making it unrealistic, and potentially ethically dangerous, to imagine a fixed moral signature. Researchers should take care to avoid interpretations that conspire with assumptions that the mind is the brain, thereby implying the brain itself is the moral agent. One should also question the accuracy of stating that a brain region or set of cells is the seat of moral agency.

While Levy is concerned about expanding the ethicist’s toolkit through using neuroscience findings, we wonder about the ethical implications of using neuroscience in a manner that seems to ascribe moral agency to the brain alone. Levy describes scenarios where neuroimaging could be used to discriminate intent. What is to say that these data would not be used to predict mal-intent as the Department of Homeland Security’s (DHS) and Transportation Security Association’s (TSA) Future Attribute Screening Technology (FAST) aims to do? Indeed, neuroimaging technologies have already been (mis)appropriated in the courtroom (Brown and Murphy, 2010) and in some cases for questionable commercial activity (Farah, 2009) such as those in the business of lie detection (Greely and Illes, 2007).

We applaud Levy’s creativity and concern for expanding and improving interdisciplinary ethical discourse. However, we suggest that caution be exercised to avoid using neuroscience beyond its limitations. We also advise that scientists must be the ethical stewards of their work. While neuroscience continues to deliver exciting findings the considerable beauty and complexity of the brain has yet to be fully understood.

References

Arthurs, O. J. and Boniface, S., 2002. How well do we understand the neural origins of the fMRI BOLD signal? Trends in Neurosciences. 25: 27-31.

Attwell, D., Buchan, A. M., Charpak, S., Lauritzen, M., Macvicar, B. A. and Newman, E. A., 2010. Glial and neuronal control of brain blood flow. Nature. 468: 232-243.

Brown, T. and Murphy, E., 2010. Through a scanner darkly: Functional neuroimaging as evidence of a criminal defendant’s past mental states. Stanford Law Review. 62 1119-1208.

Farah, M. J., 2009. A picture is worth a thousand dollars. Journal of Cognitive Neuroscience. 21: 623-624.

Greely, H. T. and Illes, J., 2007. Neuroscience-based lie detection: The urgent need for regulation. American Journal of Law & Medicine. 33: 377-431.

Koehler, R. C., Roman, R. J. and Harder, D. R., 2009. Astrocytes and the regulation of cerebral blood flow. Trends in Neurosciences. 32: 160-169.

Levy, N. 2011. Neuroethics: A new way of doing ethics. AJOB Neuroscience.

McCabe, D. P. and Castel, A. D., 2008. Seeing is believing: The effect of brain images on judgments of scientific reasoning. Cognition. 107: 343-352.

Shibasaki, H. 2008. Human brain mapping: Hemodynamic response and electrophysiology. Clinical  Neurophysiology. 119(4): 731-43.

Weisberg, D. S., Keil, F. C., Goodstein, J., Rawson, E. and Gray, J. R., 2008. The seductive allure of neuroscience explanations. Journal of Cognitive Neuroscience. 20: 470-477.

Advertisements

Actions

Information

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s




%d bloggers like this: