The Fight for the Mind: On Balancing Neural Evidence in Criminal Justice
- TULJ
- Nov 13
- 12 min read
Eleanor Chou
Edited by Jordan Perlman, Marissa Ambat, Mac Kang, and Sahith Mocharla
I. The Rise of Neuroscience in the Courtroom
What if the court could read your mind? This question, previously thought to be rooted in fantasy, is quickly becoming a reality as the field of neuroscience rapidly advances. Through technological innovations, neuroimaging continues to improve in accuracy. Modalities such as electroencephalograms (EEG), positron emission tomography (PET) scans, and functional magnetic resonance imaging (fMRI) have opened up new avenues for measuring the neural underpinnings of complex human behaviors [1]. Noninvasive EEG data records the electrical activity of the brain, identifying specific neural signatures such as P300 waves that signal familiarity in humans [1]. PET scans and fMRI data measure chemical and blood-oxygen changes, respectively, and map which neural regions are active, providing visible indicators for behaviors, emotions, and structural information about the brain [2].
However, the proliferation of neurotechnology––while offering novel applications in medicine––creates significant challenges when considering its potential uses in a legal context. These neuroimaging methods have advanced the understanding of functional regions—such as the anterior cingulate cortex, frontal lobe, and amygdala—illuminating traits like aggression, rational decision-making, and memory in individuals [3]. In courtroom contexts, proof of these neural patterns––linked to thought, intent, and guilt––could serve as compelling evidence drawn directly from an individual’s mind. However, this ability to translate neural data into traits and behaviors raises questions about violations of liberty [4].
Neuroimaging has the potential to reveal not just physical information, but direct correlations to an individual’s thoughts and emotions, forcing courts to confront how the inner workings of the mind will be protected. Specialized research within fields like neurophysiology has deterministically tied lesions in specific regions of the cortex to impairments of behavior, such as damage to the frontal lobe and subsequent proven influences on rational decision-making [5]. However, in both research and courtroom contexts, complicated processes like intent remain difficult to define.
In complex domains like intent, measurements through neural data may be made to prove correlation, but any conclusions drawn remain probabilistic due to scientific limitations [6]. Due to imperfections in the validity of neural data, its use in criminal trials must be heavily regulated, as attempting to define underlying thoughts in an individual may disproportionately affect sentencing and determinations of guilt if misinterpreted or unfairly admitted into evidence.
While neural evidence is excluded in cases that attempt to infer intent, guilt, or criminal predispositions, it is increasingly admitted in cases that limit its use to mitigation. As this issue of privacy evolves, courts must remain cautious as the decisions they make today define privacy protections in the future. The prevalence of neuroimaging in criminal courts has ignited a profound modern debate: Does the mind have the legal right to privacy?
II. Cases Excluding Neural Data
The encroaching threat to mental privacy posed by the usage of neural data is best demonstrated by its rejection in criminal trials. In such cases, courts have refused to admit evidence intended as probabilistic proof of an individual's mental traits or behavioral predispositions.
One of the earliest cases ruling on neural data was a case involving the attempted use of evidence on a gene claimed to be associated with aggression in humans. In 1994, Stephen Mobley was convicted of murder and aggravated assault before being sentenced to death by the trial court [7]. Mobley later appealed on the grounds of genetic predisposition to violent actions; essentially, Mobley argued he was predisposed to violent or impulsive behavior and thus less culpable. However, in Mobley v. State (1995), the Georgia Supreme Court affirmed the lower court’s decision, ruling on neurotechnology’s lack of verifiable scientific accuracy.
In Mobley, Mobley’s defense argued that the trial court violated his Eighth and Fourteenth Amendment rights to a fair trial and protection against disproportionate punishment when it refused to admit the evidence to excuse criminal culpability. The evidence employed studies on monoamine oxidase A (MAOA) gene deficiencies, suggesting that aggressive behavior resulted from alterations in neurotransmitter processing [8].
The Georgia Supreme Court rejected the defense’s claims, asserting that the MAOA gene deficiency could not be linked to Mobley’s behavior with certainty [9]. They concluded that given the complexity of the human brain, dysfunction in a single gene cannot alone account for predispositions to violence.
This case established that biological determinism was insufficient evidence to negate criminal intent, setting an early limitation on the application of neurogenetics in criminal courts. Their refusal to admit the evidence served as a barrier against improper interpretation of Mobley’s genetic profile, creating a precedent of legal accountability when considering neural evidence to prove behaviors in criminal courts. While the defense introduced the evidence in this case, the court’s ruling made it so that neural information must meet a scientific standard before being publicly interpreted, thereby safeguarding probabilistic biological data about an individual’s character from unjust speculation..
Following this case, a standard of evidence was delineated in Daubert v. Merrell Dow Pharmaceuticals, Inc., where evidence on a chemical causing birth defects was ruled unreliable due to high error rates and a scientifically unsubstantiated link to the psychological states of truth and deception [10]. In this case, the court ruled that judges must serve as gatekeepers to scientific evidence by scrutinizing the data before presenting it to the jury. Daubert was the first case to create specific requirements for scientific data and serves as a precedent for evaluating the reliability, accuracy, and validity of evidence before admission into court.
Building on Daubert’s standard of evidence, restrictions on the specific use of fMRI evidence were created. In United States v. Semrau, the defense attempted to introduce fMRI lie detection evidence from Cephos Corporation to exonerate psychiatrist Dr. Lorne Semrau, who was charged with Medicare fraud [11].
The evidence consisted of three brain scans that recorded Dr. Semrau’s psychological states, where increased blood flow indicated active deception or a lack thereof, measuring truthfulness. The defense claimed the Cephos Corp scans had 80-90% accuracy in detecting deception, and the visible inactivity in the scans was consistent with their assertion that Dr. Semrau did not intend to commit fraud.
Semrau placed an explicit restriction on probabilistic neural evidence on the basis that existing technology cannot assert definitive ties to subjective domains of the human mind. They drew from the Daubert precedent, ruling that the fMRI data in this case could not verifiably prove the complex multinodal aspects of Dr. Semrau’s cognitive state and were therefore excluded from the case [12]. This ruling establishes a boundary to the use of neural scans within its objective scope, meaning that for all subsequent cases, fMRI data cannot establish or infer patterns of thought that indicate intent because such scientific evidence would fail to pass existing standards (Daubert).
Therefore, even in cases like Semrau, where the evidence was presented by the defense to prove truthfulness, allowing its admission would invite larger violations of cognitive liberty, as using weakly-validated associative data to infer intent invites the risk of misinterpretation and undue encroachment on mental states.
Beyond the United States, international courts have taken steps to protect cognitive states by preventing the use of neural signals in proving a defendant’s knowledge of a crime.
In India, an appeal of the State of Maharashtra v. Sharma resulted in the Bombay High Court overturning the decision that had accepted the use of Brain Electrical Oscillation Signature (BEOS) data to prove experiential knowledge of a crime [13]. Initially, the prosecution successfully introduced EEG-based data, which measured electrical activity in the brain linked to memory recall, to convict Aditi Sharma of murder, marking the world’s first criminal conviction backed by brainwave evidence. However, Sharma appealed this decision, and it was overturned with BEOS evidence being ruled inadmissible in criminal court. In a two-part argument, the Bombay High Court held that the evidence proving experiential knowledge could not be admitted based on its lack of peer-reviewed acceptance and violation of the law protecting against self-incrimination under Article 20(3) of the Constitution of India.
As a whole, Sharma highlights a sustained global effort to protect individuals from unjust use of probabilistic neural data in the courtroom while simultaneously preserving the concept of cognitive liberty. The decision in Sharma reveals an international leaning toward safeguarding individual thoughts over careless state intrusion in the pursuit of justice, raising the question of ethical limitations on brain-based evidence in courtrooms.
III. Cases Admitting Neural Data
While extensive regulations exist regarding the use of neural data in court, it has been successfully admitted under specific, highly restricted standards. In these cases, criminal trials have allowed neural data to draw limited conclusions, primarily for mitigation.
Defendant Herbert Weinstein in People v. Weinstein (1992) was charged with second-degree murder after strangling his wife and throwing her body from a 12-floor window [14]. In this case, the defense argued that the defendant suffered from an arachnoid cyst, damaging his frontal lobe and impairing rational judgment. They presented MRI scans showing the cyst pressing against his frontal lobe, and PET scans showing abnormally low glucose uptake in the same region—both of which have been scientifically aligned with behavioral disinhibition in humans. The research linking lesions to impulse control had been empirically studied through an abundance of reproducible tests at the time, mapping the direct influence on behavior to exact neural circuits–such as the ventromedial prefrontal cortex’s link to limbic structures and its role in computing emotional and rational information that influences decision-making.
The Court accepted the neural evidence within a limited scope. They allowed the MRI scans to be used to provide a reason for Weinstein’s neural impairment, but not as a direct link to his violent behavior [15]. Weinstein’s sentence was reduced to manslaughter after a guilty plea, and the admission of this evidence shows the potential use of brain scans in trial courts, while also demonstrating the strict standards for accuracy and scope of use [16].
In this case, the scans were restricted to prove only extreme emotional disturbance under New York Penal Law § 125.25(1)(a), and not complicated processes like intent, knowledge, or guilt. Furthermore, disinhibition due to lesions in the frontal cortex had been extensively studied at the time, and the link presented between the coupled PET and MRI scans and Weinstein’s impairment further validated specific influences on impulse control.
Weinstein defined a new standard of empirically supported deterministic data. The acceptance of direct scans coupled with scientific measurements to link verifiable neural malformations to behavioral states redirects the trajectory of the legal sphere into a new frontier.
More modern cases like People v. Adams (2014), where the MAOA gene was used as evidence for sentence mitigation, show the continued willingness of criminal courts to admit neuro-genetic evidence within its defined scope of use [17].
This trend was further solidified in the Italian case of Abdelmalek Bayout (2009), where the defendant Bayout killed a man in a heated confrontation. After being sentenced to nine years and two months in prison, Bayout appealed, and neural evidence coupled with genetic testing was used to mitigate his sentence by a year [18].
Similar to People v. Weinstein, MRI evidence was used to prove abnormalities in Bayout’s brain structure to prove a lack of impulse control [19]. These abnormalities were coupled with genetic tests discovering low activities of the MAOA gene, which had been studied more extensively by this time and linked to antisocial behaviors and aggression in affected individuals. While similar genetic evidence had failed in cases like Mobley v. State, the MAOA gene had failed due to its use to prove intent.
While Bayout is independent of Mobley’s jurisdiction, it used similar MAOA data to prove predisposition. However, it differs from Mobley’s use of the MAOA as a standalone proof of aggression, but instead couples the genetic test with neuroimaging scans, instead of considering it as a standalone proof for aggression. Furthermore, Bayout’s defense distinctively aimed to prove increased impulsiveness to mitigate his sentence, as opposed to Mobley, where the genetic data was used to disprove criminal responsibility completely [20]. The direct evidence proving brain abnormalities and low-activity MAOA variants established a credible tie to dysfunction and behavior, making the evidence admissible.
The admission of neural data has largely been restricted to direct structural deformities causing specific behaviors and sentence mitigation in the court. While widely accepted in sentence mitigation, the admissibility of neural data raises questions about where future boundaries will be drawn as neurotechnology and research continue to progress. In the discussed cases, the neural data were admitted in isolated situations where the presented cortical deformities and abnormal neurogenetics have been confirmed through scientific scrutiny to give rise to a specific behavioral outcome. In future cases, as research on neural linkages to complex functions becomes more accurate and the span of behavior that can be scientifically verified increases, the question of cognitive liberty remains paramount.
IV. Defining Cognitive Boundaries in Courts
As a whole, neural data has opened up doors to expansive possibilities in criminal courts. As research continues to venture into new frontiers of the human brain, the legal sphere must adapt to new types of evidence. Precedent has regulated both the type and use of neural evidence presented to the jury. Exclusionary standards outlining the importance of validity before have been established by decisions like Mobley and Semrau, where the evidence must be thoroughly scrutinized [21]. Furthermore, protections against the threat of self-incrimination are outlined in international cases like Sharma, showing a global push for cognitive liberty [22].
These decisions have served as safeguards for individuals’ private intents, motives, and emotions from unfair speculation in criminal courtrooms, ensuring that all uses of neural information avoid complex private realms.
As cases such as Weinstein and Bayout establish, courts have not completely excluded neural data [23]. While brain scans proving structural deformities in the human brain were admitted in these cases, neural data use is limited to both sentence mitigation and to proof of impairment. The court allows for discussion of neural evidence within set (objective) boundaries. Scans may be used to display physical proof of lesions, and the reasoning they provide is strictly restricted to proof of the state of impairment at the time of the relevant crime. For example, the evidence in Weinstein was not used to justify the defendant’s violent actions, but rather to show cognitive dysfunction that inhibited rational decision-making [24]. Bayout, an international case, followed this same reasoning, using genetic data and fMRI scans to mitigate the defendant’s sentence [25]. Data was admitted under the scope of proving a neurologically rooted loss of control, giving context for his aggressive nature in that certain circumstance; however, crucially not mitigating the ‘guilt’ of the defendant for the crime.
In both of these decisions, the evidence was used for verifying existing impairments and proving a disinhibition at the time of the committed crimes. This data, however, was not allowed to be used as causal reasoning for behavior, showing the nuanced perspective of the legal sphere when considering complex neural evidence.
V. The Future of Neural Data
As neurotechnology becomes increasingly sophisticated, the courts must continue to scrutinize evidence for validity while also considering questions of consent and self-incrimination. Existing barriers to neural data in the courtroom build their foundation on previous scientific shortcomings. But modern research on potential neural evidence, like P300 brain signals that strengthen markers like familiarity and awareness, calls into question the ruling in Sharma, which proclaimed experiential knowledge was inadmissible, due to weak scientific linkage [26].
Furthermore, conclusions about patterns of thought proving intent in cases like Sharma and Semrau will soon require further scrutiny. Sophisticated fMRI applications in semantic and visual decoding now allow for high-precision translation of an individual’s thoughts and neural perceptions of visual content through computational modeling, rendering arguments over excluding imprecise neural data obsolete [27].
New technology has the potential to provide deterministic evidence of courtroom focuses such as intent, guilt, and even personal thoughts. When these applications become widely available, the protections on privacy placed by scientific barriers will soon falter, steering the discussion toward balancing objective justice and individual liberties in criminal trials. Questions of justice and fair use will continue to arise as neuroscience and law intersect. Ultimately, court decisions on neural evidence today set the foundation for the boundaries of the future, as well as the trajectory of the court toward becoming either an objective seeker of justice or a robust defender of internal domains of thought.
[1] Neuroelectrics, Exploring Brain Imaging Techniques: EEG, MRI, NIRS, and PET, Neuroelectrics (Aug. 27, 2024), https://www.neuroelectrics.com/blog/exploring-brain-imaging-techniques-eeg-mri-nirs-and-pet.
[2] John Polich, Updating P300: An Integrative Theory of P3a and P3b, 118 Clin Neurophysiol 2128 (2007), https://pmc.ncbi.nlm.nih.gov/articles/PMC2715154/
[3] Ji Hyun Ko, Chris C. Tang & David Eidelberg, Brain Stimulation and Functional Imaging with fMRI and PET, 116 Handb Clin Neurol 77 (2013).
[4] Maja Nikolic et al., Brain Responses in Aggression-Prone Individuals: A Systematic Review and Meta-Analysis of Functional Magnetic Resonance Imaging (fMRI) Studies of Anger- and Aggression-Eliciting Tasks, 119 Progress in Neuro-Psychopharmacology and Biological Psychiatry 110596 (2022), https://www.sciencedirect.com/science/article/pii/S0278584622000884
Eyal Aharoni et al., Neuroprediction of Future Rearrest, 110 Proc. Natl. Acad. Sci. U.S.A. 6223 (2013), https://www.pnas.org/doi/full/10.1073/pnas.1219302110
[5] Nita A. Farahany, Paul W. Grimm, The Battle
for Your Brain: A Legal Scholar’s Argument for Protecting Brain Data and Cognitive Liberty, JUDICATURE VOL. 107 NO. 3 (October 14, 2025)
[6] Kieran O’Driscoll & John Paul Leach, “No Longer Gage”: An Iron Bar through the Head, 317 BMJ 1673 (1998), https://pmc.ncbi.nlm.nih.gov/articles/PMC1114479/
[7] John-Dylan Haynes et al., Reading Hidden Intentions in the Human Brain, 17 Current Biology 323 (2007), https://www.sciencedirect.com/science/article/pii/S0960982206026583
[8] Mobley v. State, Justia Law, https://law.justia.com/cases/georgia/supreme-court/1995/s94p1271-1.html (last visited Oct. 15, 2025)
[9] H. G. Brunner et al., Abnormal Behavior Associated with a Point Mutation in the Structural Gene for Monoamine Oxidase A, 262 Science 578 (1993)
Deborah W. Denno, Revisiting the Legal Link Between Genetics and Crime, 69 Law & Contemp. Probs. 209 (2006)
[10] see [8]
[11] Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), Justia Law, https://supreme.justia.com/cases/federal/us/509/579/ (last visited Oct. 15, 2025).
[12] United States v. Semrau, 693 F.3d 510 (6th Cir. 2012)
[13] see [11]
[15] People v. Weinstein – Case Brief Summary – Facts, Issue, Holding & Reasoning – Studicata, https://www.studicata.com/case-briefs/case/people-v-weinstein (last visited Oct. 15, 2025).
[16] PEOPLE v. WEINSTEIN, 156 Misc.2d 34 | N.Y. Misc., Judgment, Law, Casemine.Com, https://www.casemine.com/judgement/us/5914bee4add7b049347aaec2?utm_source=amp&target=amp_impara (last visited Oct. 15, 2025).
[18] People v. Adams, 59 Cal. 4th 1081, 330 P.3d 301, 177 Cal. Rptr. 3d 107 (2014).
[19] Emiliano Feresin, Lighter Sentence for Murderer with “Bad Genes,” Nature (2009), https://www.nature.com/articles/news.2009.1050
[20] see [14]
[21] see [8] and [19]
[22] see [8] and [10]
[23] see [14]
[24] see [15] and [19]
[25] see [15]
[26] see [19]
[27] see [13] and [2]
[28] Alexander G. Huth et al., Decoding the Semantic Content of Natural Movies from Human Brain Activity, 10 FRONT SYST NEUROSCI 81 (2016), https://pmc.ncbi.nlm.nih.gov/articles/PMC5057448/ and see [10]
Comments