Assessing the Effects of Deep Brain Stimulation on Agency
By Adina L. Roskies
In the initial stages of development of the AAT we demonstrate proof of principle by showing that the data the AAT generates is sufficient to classify individuals in both nonclinical and neuropsychiatric populations as having certain agency-impacting disorders, such as depression, anxiety disorder, schizophrenia, eating disorder, etc. We hypothesize that psychological and neurological disorders can be characterized as identifiable regions in this agency space, so correctly predicting the presence or absence of particular illnesses is evidence of detection of relevant agentive properties.
At this time, we have completed the initial proof of concept of the AAT with a small-scale version of the survey. The main section of this version of the AAT had 223 statements that assay 13 different theorized dimensions of agency, such as emotion regulation, impulse control, volition, and risk aversion, among others. Subjects provided answers on a Likert Scale, with seven potential reactions to a given statement with responses ranging from “strongly disagree” to “strongly agree.” On the basis of machine learning models we found that we could predict with between 68 and 75% accuracy whether a subject self-reported any of six different disorders. On that basis we have moved on to developing a more comprehensive version of the AAT which integrates performance measures and insights from a wider variety of literature. We are working to intersperse performance measures to test for executive control, motor fluency, and other aspects of cognition that may be relevant to agency. Once the test is adequately honed and normed on nonsurgical populations, we will move to the clinic with DBS patients.
Want to cite this post?
Roskies, A. L. (2020). Assessing the Effects of Deep Brain Stimulation on Agency. The Neuroethics Blog. Retrieved on , from http://www.theneuroethicsblog.com/2020/02/assessing-effects-of-deep-brain_11.html
This post is part of a series featuring authors who have received the Neuroethics R01 (Research Project Grants) supported by the NIH BRAIN Initiative. These research projects specifically address prominent ethical issues arising from emerging technologies and advancements in human brain research.
![]() |
Image Courtesy of Pixabay |
“Agency” is a broad but ill-defined concept that nonetheless is of central importance to neuroethical thought. Among other things, philosophers have worried that brain interventions such as (but not limited to) deep brain stimulation (DBS) can compromise a patient’s agency. Although it plays a central role in philosophy of mind and in ethics, the concept of agency remains in need of elaboration and articulation.
As a recipient of one of the Neuroethics grants supported by the BRAIN Initiative, our group intends to provide a new approach to doing just that. The overall goal of our project is to articulate a framework for understanding agency, to enable us to better evaluate and respond to the neuroethical challenges raised by our developing abilities to alter brain function.
We conceptualize agency as a multidimensional concept, rather than as something that a person simply has more or less of. What the relevant dimensions are remains an open question. Both disease and neurotechnological intervention can affect aspects of agency, and in principle their effects can be measured. We are in the process of developing a computerized assessment tool (the Agency Assessment Tool, or AAT) to characterize where various people fall in this multidimensional space and, ultimately, to measure differences in agency along various dimensions. The AAT consists of a battery of survey questions interspersed with performance measures of various sorts that we think, informed by the philosophical and clinical literature, may be relevant to a more highly articulated conception of agency.
We have chosen to use DBS as a test case for the AAT. DBS has already raised ethical questions about ways in which direct brain stimulation may affect personal identity, autonomy, authenticity and, more generally, agency.1 DBS is approved for use with Parkinson’s Disease (PD), and it holds promise for treating some psychiatric disorders, including Treatment-Resistant Depression (TRD) and Obsessive-Compulsive Disorder (OCD).2 Although many DBS patients experience significant relief from debilitating symptoms that are recalcitrant to other, less-invasive types of interventions, that relief sometimes comes at a cost: Unwanted side effects can compromise a patient’s quality of life in ways that intuitively can be described as undermining agency. We aim to use the AAT, once developed, to measure changes in agency by comparing performance on the test before DBS electrodes are implanted and after stimulation.
Developing the AAT
In the initial stages of development of the AAT we demonstrate proof of principle by showing that the data the AAT generates is sufficient to classify individuals in both nonclinical and neuropsychiatric populations as having certain agency-impacting disorders, such as depression, anxiety disorder, schizophrenia, eating disorder, etc. We hypothesize that psychological and neurological disorders can be characterized as identifiable regions in this agency space, so correctly predicting the presence or absence of particular illnesses is evidence of detection of relevant agentive properties.
![]() |
Image Courtesy of Pixabay |
We formulate the AAT by canvassing the kinds of questions that are found in the philosophical literature on the metaphysics and ethics of agency as well as in the bioethics literature and by careful mining of case reports of patients with neurological and psychiatric illnesses. In addition, we sample judiciously from the wide variety of already normed and validated diagnostic tools that address specific kinds of challenges to agency, such as depression, anxiety, motor impairments, addiction, and so on. From these sources we develop a questionnaire of several hundred questions to assess different aspects of agency. The AAT includes catch trials in order to ensure that subjects are answering questions in a reliable and coherent manner. It also gathers information about pertinent demographics, such as age, gender, education, and race, as well as about any history of mental illness or other diseases or medical events that might impact agency.
During development, we will administer the AAT online via Amazon Mechanical Turk and Qualtrics to a broad spectrum of subjects. We aim to see whether patterns of answers on the AAT can predict whether subjects fall into specific clinical groups. We will use machine learning tools such as predictive modeling, clustering analysis, dimensionality reduction, and feature impact analysis to assess the diagnosticity of the overall test, the importance of individual questions for predictive modeling, and redundancy, to ultimately refine and streamline the questionnaire. On the basis of the results from the machine learning analysis, and through multiple iterations, the initial survey will be adjusted; this will be done both by removing less diagnostic or repetitive elements and by adjusting or inserting new questions in areas with high factor weighting to increase predictive power.
Current status
At this time, we have completed the initial proof of concept of the AAT with a small-scale version of the survey. The main section of this version of the AAT had 223 statements that assay 13 different theorized dimensions of agency, such as emotion regulation, impulse control, volition, and risk aversion, among others. Subjects provided answers on a Likert Scale, with seven potential reactions to a given statement with responses ranging from “strongly disagree” to “strongly agree.” On the basis of machine learning models we found that we could predict with between 68 and 75% accuracy whether a subject self-reported any of six different disorders. On that basis we have moved on to developing a more comprehensive version of the AAT which integrates performance measures and insights from a wider variety of literature. We are working to intersperse performance measures to test for executive control, motor fluency, and other aspects of cognition that may be relevant to agency. Once the test is adequately honed and normed on nonsurgical populations, we will move to the clinic with DBS patients.
Summary
![]() |
Image Courtesy of Pixabay |
The BRAIN Initiative puts a premium on the development of new neurotechnologies. As our
technological prowess in neuroscience increases it will be increasingly possible to intervene in or alter people’s neural function. However, we lack a clear framework in which to think of the ethical consequences of these interventions. The AAT aims to provide a broad-based assessment tool for evaluating changes in agency due to disease and neurointerventions. Our work will provide a framework for thinking about agency and the way it is affected by disease and by novel neurotechnologies -- not only DBS, but also Brain-Computer Interfaces (BCIs), optogenetics, and even pharmacological treatments. The data it yields will allow us to better identify, articulate, and measure changes on “dimensions of agency.” This is important for a number of clinically and ethically relevant considerations, including evaluation of the neuroethical implications of brain interventions, improved clinical decision-making and treatment, and informed consent.
The concrete objectives of our proposal are:
- To develop a new framework and comprehensive assessment tool for agency using both philosophical and empirical methods;
- To use this tool to assess changes in agency due to brain interventions using DBS patient populations as a test case;
- To develop a database to house the data that we acquire with these tools, and, going forward, to allow others to add to the database. This will enable us to catalogue the effects and side effects of DBS, and possibly to provide insight into the neurobiological mechanisms by which DBS works.
Research reported in this publication was supported by the NIMH’s BRAIN Initiative of the National Institutes of Health under award number 1RF1MH117813. Initial funding for this project came from the William H. Neukom 1964 Institute for Computational Science at Dartmouth College. The content is solely the responsibility of the author and does not necessarily represent the official views of the National Institutes of Health or the Neukom Institute.
References
- Baylis, F. (2011). “‘I Am Who I Am’: On the Perceived Threats to Personal Identity from Deep Brain Stimulation.” Neuroethics, 6(3), 513-526. doi:10.1007/s12152-011-9137-1; Goddard, E. (2017). “Deep Brain Stimulation Through the “Lens of Agency”: Clarifying Threats to Personal Identity from Neurological Intervention.” Neuroethics, 10(3), 325-335. doi:10.1007/s12152-016-9297-0; Johansson, V., Garwicz, M., Kanje, M., Schouenborg, J., Tingstrom, A., Gorman, U. (2011). “Authenticity, Depression, and Deep Brain Stimulation.” Frontiers in Integrative Neuroscience, 5. doi:10.3389/fnint.2011.00021; Roskies, A. L. (2015). “Agency and Intervention.” Philosophical Transactions of the Royal Society B: Biological Sciences, 370(1677), 20140215. doi:10.1098/rstb.2014.0215.
- Holtzheimer, P. E., Mayberg, H. S. (2011). “Deep Brain Stimulation for Psychiatric Disorders.” Annual Review of Neuroscience, 34(1), 289–307. doi:10.1146/annurev-neuro-061010-113638; Lee, D. J., Lozano, C. S., Dallapiazza, R. F., & Lozano, A. M. (2019). “Current and future directions of deep brain stimulation for neurological and psychiatric disorders”, Journal of Neurosurgery JNS, 131(2), 333-342.
______________
Adina Roskies is the Helman Family Distinguished Professor at Dartmouth College, Professor of Philosophy and chair of the Cognitive Science Program, and an affiliate of the Department of Psychological and Brain Sciences. She received a Ph.D from the University of California, San Diego in Neuroscience and Cognitive Science in 1995, a Ph.D. from MIT in philosophy in 2004, and anM.S.L. from Yale Law School in 2014. Prior to her work in philosophy she held a postdoctoral fellowship in cognitive neuroimaging at Washington University with Steven Petersen and Marcus Raichle, and from 1997-1999 was Senior Editor of the neuroscience journal Neuron. Dr. Roskies has received numerous awards and fellowships, including the Stanton Prize from the Society for Philosophy and Psychology, The Neuroethics Prize from the Italian Society of Neuroethics, a Mellon New Directions fellowship, and fellowships from the Princeton Center for Human Values and the University of Pittsburgh’s Center for Philosophy of Science. She was recently awarded grants from the NIH through the BRAIN Initiative, and the Templeton Foundation. Dr. Roskies’ research interests lie at the intersection of philosophy and neuroscience, and include philosophy of mind, philosophy of science, and ethics.She has coauthored a book with Stephen Morse, A Primer on Criminal Law and Neuroscience.
Want to cite this post?
Roskies, A. L. (2020). Assessing the Effects of Deep Brain Stimulation on Agency. The Neuroethics Blog. Retrieved on , from http://www.theneuroethicsblog.com/2020/02/assessing-effects-of-deep-brain_11.html