When the government can read your mind
We are now at a point where we can scan brain activity with fMRI, decode the patterns, and use the information to “read minds” or predict what a person is experiencing. For example, the Gallant Lab at the University of California, Berkley, published a paper in 20111 showing that by recording subjects watching a set of movies, they can estimate what visual features parts of the brain are encoding. Then, when they show you a new movie, their model can predict what you are seeing based on your brain activity.
This “mind-reading” has limitations: the reconstruction is primitive, worse on abstract or rare stimuli, and each subjects must be scanned many times to tune the model to his or her individual brain. However, this experiment proves the principle that we create models that use brain activity to predict dynamic conscious experience even with the low visual and temporal resolution and indirect measures of an fMRI. Other labs are progressing in different domains, for example Chang and colleagues were able to decode auditory cortex to reconstruct the individual words heard by subjects2.
This sort of technology has many implications, and provokes many questions—both technical and ethical:
fMRI for thought-reading Whether this approach of recording, modeling, and decoding will work with thoughts will depend on how similar our inner monologue is to actual audition. If our mental imagery uses the same pathways as sensory experience, and is encoded in a similarly, then we are not far away from it. However, this technique is unlikely to be able to extract thoughts the subject doesn’t want to share—it requires putting the subject in an fMRI (so could not be done surreptitiously), it requires compliance on the training taks, and even then inner-monologue can be consciously controlled.
fMRI as a Lie Detector Other groups are looking at lie detection—trying to use fMRI as a more accurate polygraph test. They work on the principal that certain patterns of brain activity will reflect the higher-order processing needed to suppress the truth. There are currently at least two private companies trying to market fMRI as a Lie Detector: Cephos Corporation and No Lie MRI. The sensitivity and specificity of these tests have not been well characterized.
Ethical use of thought-reading or lie detectors—Balancing benefits with risks In what ways is it appropriate to use these devices? Some are clearly beneficial, for example thought-reading may allow communication with otherwise inaccessible paralyzed “locked-in” patients.
Many possible uses are more ambiguous:
Even if we decide that some of these uses could be beneficial if used properly, how will we ensure they are used responsibly? Who will store the data produced and regulate that it will only be used for what it is meant to be?
These technologies will always be limited by the accuracy of our memories and our capacity for self-deception. Psychology studies have repeatedly shown malleability of memories and that subjective confidence isn’t a good correlate of accuracy. Even if a person thinks something is true and a lie detector verifies it, there is no insurance that it really is true.
As technology continues to improve, we must inform the greater public about what is currently possible, discuss these difficult questions, and make sure that our laws ensure the rights we want protected.
--Ben Kuebrich
Want to cite this post?
Kuebrich , B. (2012). When the government can read your mind. The Neuroethics Blog. Retrieved on
-->, from http://www.theneuroethicsblog.com/
Works Cited
1. Nishimoto NS, Vu AT, Naselaris T, Benjamini Y, Yu B, Gallant JL. (2011) Reconstructing Visual Experiences from Brain Activity Evoked by Natural Movies, Current Biology, 21(19): 10.1016/j.cub.2011.08.031
2. Pasley BN, David SV, Mesgarani N, Flinker A, Shamma SA, et al. (2012) Reconstructing Speech from Human Auditory Cortex. PLoS Biol 10(1): e1001251.doi:10.1371/journal.pbio.1001251
This “mind-reading” has limitations: the reconstruction is primitive, worse on abstract or rare stimuli, and each subjects must be scanned many times to tune the model to his or her individual brain. However, this experiment proves the principle that we create models that use brain activity to predict dynamic conscious experience even with the low visual and temporal resolution and indirect measures of an fMRI. Other labs are progressing in different domains, for example Chang and colleagues were able to decode auditory cortex to reconstruct the individual words heard by subjects2.
This sort of technology has many implications, and provokes many questions—both technical and ethical:
- Will the techniques extend beyond perception and allow us to read someone’s thoughts?
- Could they be used to create a perfect lie director (or at least one orders of magnitude better than the polygraph)?
- If such technology hypothetically existed, how would we use it?
- How could we balance its potential benefits with potential for abuse?
fMRI for thought-reading Whether this approach of recording, modeling, and decoding will work with thoughts will depend on how similar our inner monologue is to actual audition. If our mental imagery uses the same pathways as sensory experience, and is encoded in a similarly, then we are not far away from it. However, this technique is unlikely to be able to extract thoughts the subject doesn’t want to share—it requires putting the subject in an fMRI (so could not be done surreptitiously), it requires compliance on the training taks, and even then inner-monologue can be consciously controlled.
fMRI as a Lie Detector Other groups are looking at lie detection—trying to use fMRI as a more accurate polygraph test. They work on the principal that certain patterns of brain activity will reflect the higher-order processing needed to suppress the truth. There are currently at least two private companies trying to market fMRI as a Lie Detector: Cephos Corporation and No Lie MRI. The sensitivity and specificity of these tests have not been well characterized.
Ethical use of thought-reading or lie detectors—Balancing benefits with risks In what ways is it appropriate to use these devices? Some are clearly beneficial, for example thought-reading may allow communication with otherwise inaccessible paralyzed “locked-in” patients.
Many possible uses are more ambiguous:
- Could a theoretical thought-reading device ethically be used in questioning a criminal suspect, enemy combatant, or terrorist?
- Could a person be apprehended and held responsible for thoughts they had if they had not yet committed a crime?
- Is it ethical or legal for businesses to use lie-detectors in hiring, especially if there are of questionable efficacy?
- At what level of accuracy (if ever) should a lie-detector would it be admissible in court?
- Would results counted as testimony or biological evidence?
Even if we decide that some of these uses could be beneficial if used properly, how will we ensure they are used responsibly? Who will store the data produced and regulate that it will only be used for what it is meant to be?
These technologies will always be limited by the accuracy of our memories and our capacity for self-deception. Psychology studies have repeatedly shown malleability of memories and that subjective confidence isn’t a good correlate of accuracy. Even if a person thinks something is true and a lie detector verifies it, there is no insurance that it really is true.
As technology continues to improve, we must inform the greater public about what is currently possible, discuss these difficult questions, and make sure that our laws ensure the rights we want protected.
--Ben Kuebrich
Want to cite this post?
Kuebrich , B. (2012). When the government can read your mind. The Neuroethics Blog. Retrieved on
-->, from http://www.theneuroethicsblog.com/
Works Cited
1. Nishimoto NS, Vu AT, Naselaris T, Benjamini Y, Yu B, Gallant JL. (2011) Reconstructing Visual Experiences from Brain Activity Evoked by Natural Movies, Current Biology, 21(19): 10.1016/j.cub.2011.08.031
2. Pasley BN, David SV, Mesgarani N, Flinker A, Shamma SA, et al. (2012) Reconstructing Speech from Human Auditory Cortex. PLoS Biol 10(1): e1001251.doi:10.1371/journal.pbio.1001251
Comments
Post a Comment