November 15, 2026 21:42

Follow Lima Blog on

Can we use AI and neuroscience to detect lies in criminal proceedings?  A European perspective

Isadora Neroni Rezende, research fellow in Criminal Procedure at the University of Bologna.

Click here to download.

1. Introduction 

Neuroscience and artificial intelligence (AI) are increasingly used in criminal justice to obtain insights from the brain. For instance, brain diagnostics are used as evidence to discard criminal responsibility and assess recidivism risks (Greely & Farahany, 2019; Katwala, 2019). Lie and memory detection based on neuroimaging or emotion facial recognition (EFR) are also employed to evaluate the veracity of defendants’ and witnesses’ statements (Díaz Soto & Borbón, 2022, p. 2; Oravec, 2022, pp. 2–3). The use of these technologies is already a reality in Europe. Since 2020, the Belgian Code of Criminal Procedure foresees the possibility to submit suspects, victims, and witnesses to a polygraph test, on a voluntary basis[1]. The test does not necessarily foresee the use of AI methodologies[2], but the reform shows the interest of criminal justice authorities in spotting deceptive behaviour, which in the future may be done through more advanced technologies. In the immigration context, the Horizon2020-funded project iBorderControl was shortly trialled in Hungary, Latvia, and Greece to obtain faster border controls. The initiative employed EFR cameras to spot signs of deception in travellers responding to security agents (Sánchez-Monedero & Dencik, 2022). 

Neuroscientific and AI techniques are varied, and they show different degrees of intrusiveness. On the one hand, structural neuroscientific tools include magnetic resonance imaging (MRI) and computed tomography (CT), which provide a static image of the brain anatomy Ligthart et al., 2021, p. 192). That is why they are mostly used to prove brain damage and sustain claims of decreased responsibility for the defendant’s criminal conducts. On the other hand, functional neuroscientific instruments like functional magnetic resonance imaging (fMRI) and electroencelography (EEG) analyse the brain physiology, thus giving a more dynamic picture of how the brain works Ligthart et al., 2021, p. 192). Functional neuroimaging is even used to decode and reconstruct visual stimuli to determine whether subjects are thinking of a specific face or place (Haynes, 2012, p. 31; Shen, 2013, p. 687). While all neuroscientific tools are designed to draw information from the brain, these applications are what most resembles to actual mind reading, raising serious concerns for the respect of mental privacy. Similarly, emotional AI techniques draw inferences about someone’s mental state from external physiological inputs (e.g., facial expressions, respiration rates, blinking patterns, skin temperature, blood pressure), and are used in crime prevention or in forensic contexts as advanced versions of the polygraph (Podoletz, 2023, pp. 1067–1068) [3].  

Functional neuroscience and emotional AI are becoming popular tools for memory and lie detection, which are the focus of this analysis. Indeed, research shows that humans have poor skills when it comes to detect cues of deception, especially when non-verbal (Brennen & Magnussen, 2023, p. 2; Podoletz, 2023, p. 1073). If automated methods of lie detection hold the promise of increasing the accuracy of human assessments in criminal proceedings, neuroscientists have advocated against the premature use of these technologies in courtrooms (Greely & Farahany, 2019, p. 453). Nonetheless, the market is already pushing for these solutions; therefore, an assessment of their legitimacy in crime prevention contexts and criminal proceedings is topical.  

Unsurprisingly, the use of technologies for memory and lie detection in criminal proceedings poses many fundamental rights issues, in terms of privacy, freedom of thought, fair trial rights, and data protection. This contribution explores whether current memory and lie detection techniques, as used in different phases of criminal proceedings, are compliant with the abovementioned fundamental rights. The context chosen for the analysis is European human rights frameworks, including both the European Convention of Human Rights (ECHR) and the Charter of Fundamental Rights of the European Union (CFREU). Conversely, the so-called “neurorights” (Ienca, 2021) are not used as benchmarks in the assessment. While the contribution of this approach is acknowledged, the contention here is that introducing new fundamental rights is not necessary from a legal perspective (Bublitz, 2022; Hertz, 2023). Rather, it is interesting to see how existing provisions at the European level can be interpreted to address the risks raised by these applications.   

Given this, the article is structured as follows. Preliminary, we consider the relation between AI and neuroscientific-based lie and memory detection tools and the value of human dignity, an often-raised argument to exclude these technologies from criminal proceedings altogether (Section 2). Afterwards, we examine the issues raised by these systems in different “scenarios” of the criminal justice system: crime prevention (Section 3), investigations (Section 4), and the trial (Section 5). Lastly, Section 6 concludes the analysis.  

2. The human dignity conundrum 

An issue that underlies the use of memory and lie detection tools in criminal justice, regardless of the phase of the proceedings, is the potential violation of the value of human dignity. Notably, the Explanations to the Charter of Fundamental Rights of the EU seem to equate the need for respecting human dignity with the core essence of the rights protected[4]. If that were the case, any technology interfering with this value would be considered unacceptable in the European human rights system.  

 In the domain of memory and lie detection, some argue indeed that human dignity sets an absolute prohibition of using these tools against the person (Díaz Soto & Borbón, 2022, p. 3). Emotional AI systems (like EFR) pretend to objectify emotions, the most intimate contents of the mind, in a way that would degrade individual dignity. From a criminal procedural perspective, using neuroscientific methods of lie detection risks shifting the object of the ascertainment from the facts of the proceedings, to the persons participating in it, which could also result in a violation of their human dignity (Ferrua, 2020, p. 413).  

At the same time, however, others do not seem too keen on recognizing a clash between advanced lie detection and human dignity. In the neurolaw domain, for example, some argue against the exceptionalism of the issues pertaining to the neurolevel (Bublitz, 2022, p. 7). That concerns for instance the regulation of neurodata, which are often considered “special” in comparison to other legally recognized instances of sensitive data[5]. The particular nature of brain data would be explained by the so-called “inception problem”, meaning that the information to be protected cannot be distinguished from its source, i.e., neural activity (Ienca & Andorno, 2017, p. 14). According to others, however, it is not clear how the relationship between the sources and the information in the neural domain is any different from other forms of data (Ligthart et al., 2021, p. 200). In this sense, already existing evidence-gathering in criminal justice (e.g., sexual dysfunction assessed in psychiatric evaluation, ethnic origins revealed in DNA testing) could be considered equally invasive, as they reveal very intimate information about the subject (Ligthart et al., 2021, p. 200). Likewise, it is not explained how mental privacy should have a special regulatory treatment in comparison to other privacy interests (Bublitz, 2022, p. 7). In fact, when a memory detection machine determines that the witness actually recognizes an item from the crime scene, this information may not be considered sensitive in itself. What can be sensitive is rather the implication of that information for the subject, for example, how it may affect his or her position in the proceedings.  

Relying on human dignity only to reject the use of lie and memory detection is also problematic for the vagueness of the content of this principle. This is a very impalpable concept which is difficult to grasp from a legal perspective[6]. More than other fundamental rights, for which the EU set minimum standards of protection, diverging interpretations on the meaning of human dignity may cause fragmentation in suspects’ and accused’s prerogatives across Member States. Therefore, while the importance of this argument is not underestimated, it is useful to explore the implications of other fundamental rights, as outlined in the following sections.  

3. Crime prevention 

In crime prevention, technologies like EFR can be used to detect people acting suspiciously in public places (Crawford, 2021; Jupe & Keatley, 2020). An often-cited example is the US Transportation Security Authority’s 2003 Screening Passengers by Observation Techniques (SPOT) program, which aimed to find terrorists by scrutinizing airline passengers displaying fear or stress.  

Arguably, this practice impinges upon the right to mental privacy (Art. 8 ECHR) and the freedom of thought (Art. 9 ECHR), which includes the freedom not to manifest one’s thoughts (Koops et al., 2017, pp. 531–532). Because EFR aims to attribute emotional meaning to people’s demeanour and physiological data, indeed, these technologies try to reify the contents of their minds and use them for decision-making.  

In particular, a strong limitation can be found in the freedom of thought. Notably, this right comprises two dimensions, the external forum (subject to limitations under Art. 9(2) ECHR) and the internal forum (covered instead by absolute protection). Considering emotions as part of the forum internum is problematic, as the ECtHR adopts a narrow approach in its conceptualization. Indeed, it considers that only “views that attain a certain level of cogency, seriousness, cohesion and importance” belong to the core of this right[7]. However, such an approach underestimates the similarities between emotions and thoughts which have been uncovered in philosophy and cognitive sciences (Barrett, 2016; Nussbaum, s.d., p. 33). The conceptual link between emotions and thoughts suggests then that a wider protection of the “mind” is needed from a fundamental rights standpoint. If the internal forum of the freedom of thought is interpreted extensively, then a wider range of mental states (including emotions, ordinary thoughts, or any kind of brain activity) could be covered under the internal forum of the freedom of thought. In that case, inferring emotions revealing suspicious behaviour in public would not be considered acceptable under the Convention.  

Likewise, these applications could violate the presumption of harmlessness[8], and the fairness principle in data protection (Art. 4(1)(a) LED), which both aim to shield individuals from undue stigmatization (Neroni Rezende, 2022, p. 92). The conceptual link between these safeguards can be found in the CJEU surveillance case law. The Court has often raised concerns over the absence of personal limitations in surveillance systems implemented in the Union[9]. In this scenario, people having no connection whatsoever with criminal behaviour could be negatively affected by adverse inferences drawn by their “emotional demeanour” in public. Due to the aura of objectiveness surrounding scientific evidence, this kind of profiling could create presumptions of “suspiciousness”, which could be difficult to rebut for individuals. In fact, the techniques underlying EFR are opaque and have been defined as pseudoscience (Article19, 2021, p. 6; Jupe & Keatley, 2020, p. 62; McStay, 2020, p. 2)[10]. This entails that it might be difficult to challenge the outputs of these machines according to objective scientific criteria. In this sense, subjects would also struggle to effectively exercise their right to effective remedy. The negative implications for suspects are all the more serious if the data, gathered in the preventive context, enter in the proceedings and are used as evidence.  

4. Investigations 

During investigations, lie and memory detection tools can be useful in questionings. Often marketed as advanced versions of the polygraph, software like CM Cross, EmoKit, Miaodong and Sage Data rely on facial expressions, vocal tone, heart rate and other indicators of peripheral arousal to determine interviewees’ emotions during police questionings (Article19, 2021, p. 21). In the United States, Brain Fingerprinting Laboratories, Inc. is a company that sells a brain wave detection device that, combined with an algorithm, is claimed to recognize whether a person is recognizing a certain item (Moreno & Parashar, 2012, p. 177). 

Different variables impact the existence (and extent) of a fundamental rights interference prompted by these applications. For instance, while EFR can be used unobtrusively and covertly, functional neuroscientific methods need the cooperation of the people under examination. Moreover, in this scenario, the applicable regime also depends on whether the subject of the interview is witness or a suspect.  

Notably, witnesses have an obligation to tell the truth. Therefore, they might not always be in a position to reject the use of memory and lie detection tools altogether. Certainly, they have a right to privacy, freedom of thought, and occasionally they benefit from an anticipated application of the privilege against self-incrimination. Nonetheless, given their role in the proceedings, these rights may not be protected as strongly as for defendants. These prerogatives protect individuals from compulsions to express thoughts (Ligthart et al., 2021, p. 194), but giving statements is precisely what the law requires of witnesses to ascertain the facts. Usually, criminal procedural legislation does not refrain from using coercion to have witnesses cooperate (Redmayne, 2005, p. 211). Therefore, the use of these tools in witness interviews may raise questions of proportionality and evidentiary reliability, rather than of absolute prohibition.  

At the same time, using lie detection in witnesses’ questionings would raise a number of questions, for example, in relation to witnesses’ self-incriminating statements detected by the machine. Because these systems only generate inferences about mental states (Haynes, 2012, pp. 29–30), one should wonder about what probability threshold should be acceptable to consider that someone is lying, and should become a suspect. Also, if no specific statements are involved, interpreting the outputs of the machine may not be an easy task for those conducting the questioning. These issues remain unsolved, but applying these instruments in witnesses’ interviews does not seem as problematic as in suspects’ questionings.  

Indeed, suspects have a right to silence and not to incriminate themselves, two prerogatives that are strictly interconnected with the presumption of innocence. This principle is enshrined in Article 48 CFREU, which corresponds to Article 6(2) and (3) of the Convention. Even before the Charter, however, the CJEU had recognized the presumption of innocence as one of the fundamental rights protected in Union law[11]. At the level of secondary law, this right is explicitly recalled at Article 2 of the Directive on the presumption of innocence[12]. The instrument dedicates important provisions to the right to silence and not to incriminate oneself, which have different contents. On the one hand, the right to silence concerns the choice of the defendant to speak before the authorities (Article 7(1) of the Directive). On the other hand, the right not to incriminate oneself has a broader scope, and protects the suspect or accused against the forcible production of evidence that might be harmful for his or her defence strategy (Article 7(2) of the Directive) (Caianiello, 2020, p. 12). The ECtHR considers that this right does not hamper public authorities to use their compulsory powers to acquire evidence that exists independently of the suspect’s will[13]. This means that suspects remain free to keep a passive attitude in the proceedings against them. If they choose not to cooperate, however, public authorities can resort to their legitimate powers (e.g., issuing a judicial warrant) to obtain the evidence.  

 Against this background, it is doubtful that suspects can be subject to the assessment of covert lie detection systems, which a priori deny them the chance of making a reasoned defence choice. The same goes for functional neuroscientific tools, which always require the cooperation of the subject (Richmond, 2012, p. 186; Shen, 2013, p. 686). Even a judicial warrant may not legally compel them to undergo such an evidence-gathering act. Indeed, brain and physiological data analysed by these tools should qualify as testimonial evidence, triggering the application of the nemo tenetur. While static neuroscientific applications can be assimilated to other physical evidence (e.g., DNA, fingerprinting), the same cannot be said for AI and dynamic neuroscientific tools[14]. The data they analyse do not make any sense if not in the context of a question-answer scenario. Starting from raw data, they draw inferences about the mind’s contents, possibly revealing knowledge that would not otherwise have been shared by the suspect (e.g., if s/he recognizes an item found in the crime scene). In other words, this is not a case of mere physical evidence: giving a certain meaning to brain oxygen patterns or arousal indicators actually equates to attributing some kind of statement to the suspect.   

A contentious issue is, however, whether suspects can freely decide to submit themselves to the assessment of a lie or memory detection tool. In this respect, the right to the freedom of thought, as enshrined in the ECHR, may set a radical limit to the use of memory and lie detection tools for defendants. The preparatory works of the Convention show indeed that Article 9 was drafted to protect individuals not only from ‘confessions’ imposed for reasons of State, but also from those abominable methods of police enquiry or judicial process which rob the suspect or accused person of control of his intellectual faculties and of his conscience (Ligthart et al., 2021, p. 194; Schabas, 2015, p. 423). Memory and lie detection per se do not intervene on the suspect’s intellectual faculties directly[15]. However, there is room to argue that their use impacts subjects in a way that resonates the abominable police practices that the drafters of the Convention tried to address. The use of lie detection tools is currently criticised for failing to understand an anxiety heuristic, according to which liars show fear and truth tellers do not (Jupe & Keatley, 2020, p. 631). Being subject to such an assessment may generate feelings of anxiety that negatively impact the interviewee’s capacity for self-determination. Such anxious reactions may, in turn, induce the authorities that suspect is lying and push them to adopt more assertive questioning techniques, possibly leading to false confessions (Lasagni, 2021).

5. Trial 

At trial, a central question is whether the outputs of automated lie and memory detection machines can be used as evidence. This scenario mainly concerns witnesses, given that – as argued above – applying these tools to the accused’s statements appears highly problematic.  

If we focus on the European context, it is known that there are still no common rules on the admissibility of evidence. The ECtHR has adopted a self-restrained approach on the matter, and considers that unlawfully gathered evidence should be excluded from the proceedings only in the cases of the most flagrant human rights violations (e.g., evidence obtained through torture or inhumane treatment) [16]. Article 6 ECHR, instead, “does not lay down any rules on the admissibility of evidence or the way it should be assessed” [17]. In the EU, the issue is considered only in the perspective of mutual recognition, and Member States still hold a significant margin of appreciation on the matter (Lasagni, 2023). Recently, however, the CJEU has approached evidentiary issues through the lenses of privacy rights (Lasagni, 2023; Panzavolta & Maes, 2022). In La Quadrature du Net and Privacy International, for instance, the Court observed that it is for Member States to determine the requirements for the admissibility of and assessment of evidence. Nonetheless, national rules are not exempted from complying with the principles of equivalence and effectiveness in EU law, which require that suspects and accused are should not prejudiced from information obtained unlawfully. When evidence has been gathered in violation of privacy and data protection rights, national judges remain free to decide how to reach this objective, choosing to exclude the evidence altogether or to rely on rules governing its assessment and weighing. The violation of the adversarial principle, instead, imposes the more radical solution of excluding the evidence. For this to happen, according to the CJEU, two cumulative conditions should apply[18]. Firstly, the suspect should not have had the possibility to comment effectively on the evidence. Secondly, the evidence should pertain to a field in which the judge has no knowledge and can thus have a preponderant role in the outcome of the case.  

While the appropriateness of these rules has been questioned in the data retention field for the low technical expertise required for its understanding (Panzavolta & Maes, 2022), the same cannot be argued for AI and neuroscientific-based lie and memory detection tools. This is a field of highly specialized knowledge which remains obscure for the judge and may unduly influence the outcome of the proceedings. This means that if a lie detection machine determines that a key witness for the prosecution is telling the truth, it might be very difficult for the defence to challenge this assessment and instil a doubt in the judge’s mind about its reliability. And yet, the accuracy of these scientific methods has been questioned, although a difference is discerned between AI tools and neuroscientific ones.  

On the one hand, the functioning of AI models is known to be opaque, even to experts. In the case of emotional AI tools, such as EFR, a theory might explain their functioning, but this has been labelled as pseudoscience. Also, AI lie detection mirrors polygraph techniques, meaning that they measure physiological indicators of peripheral arousal. Problematically, however, these external markers of deception can be affected by other factors, or be manipulated by the subject (Haynes, 2021, p.35). As argued above, relying on poorly grounded scientific tools has worrisome implications for the defendant, considering that human decision-makers are often biased to believe the outputs of automated systems (Crawford, 2013). The combination of “junk science” and data fundamentalism may irremediably jeopardise the adversarial proceedings. That is why these tools should not be employed to assess the veracity of witnesses’ statements at trial (and even less of the defendants’, for the reasons states above).  

On the other hand, neuroscientific instruments appear more scientifically robust, but their accuracy is still lagging behind (Haut et. al, 2022, p. 138). Some retain that they have more development potential than polygraph-based techniques (Haynes, 2021, p.35), but there are still severe methodological limitations. For instance, fMRI shows brain activity with high spatial resolution, but scarce temporal resolution, which undermines its usefulness in real-time lie detection (Haynes, 2021, p.32). Moreover, this approach requires “fingerprinting” the brain, meaning that every thought should be associated with a unique brain pattern that varies from person to person (Haynes, 2012, p.30). Not all brain activities for specific thoughts are similar across people: therefore, the accuracy of the test depends on training the machine on a data sample from the specific subject to be heard. Current instruments are designed with very limited training data, considering that most research in deception detection is conducted on WEIRD (western, educated, industrialised, rich, and democratic cultures) students (Jupe & Keatley, 2020, p. 629). This arguably creates a false representation of human behaviour, which questions the applicability of these tools to a broader audience at this stage.  

Moreover, studies highlight that while lying stimulates similar brain activity, the patterns encoding visual images are very different (Haynes, 2021, p.33). At the same time, other studies suggest that the processing of distinct concepts (like perceiving or imagining a scene or an item) has similar neural substance, which may create confusion as to whether the subject recollects an event, or think s/he does (Brennen & Magnussen, 2023, p. 3). Reading “random” thoughts is also extremely problematic, considering that this task requires pre-training the system on the subject, and is effective only on a finite number of mental states (Haynes, 2012, pp.33-34).

 All of this currently poses strong limitations to developing “universal truth machines”. Likely, memory recognition and detection are the most problematic in this respect. Lie detection seems instead more plausible, although research is still needed to reach reliable standards of accuracy, if that will ever be possible. When this happens, their use to assess the veracity of witnesses’ statements may be admitted at trial, provided that room is left for the defence to comment effectively on the evidence based on trustworthy scientific standards.  

6. Conclusions 

This contribution explored how AI and neuroscientific lie detection tools impact fundamental rights in criminal proceedings in Europe. Starting from acknowledging the unclear implication of human dignity, the analysis covered different uses of these system in different phases of the proceedings. In crime prevention, spotting suspicious behaviour through EFR infringes the internal forum of the freedom of thought, and is at odds with the presumption of harmlessness, privacy and data protection rights, as interpreted in the CJEU’s surveillance case law. During criminal proceedings, applying these instruments to suspects and defendants finds a strong limitation in the freedom of thought and the privilege against self-incrimination. To a certain extent, the same cannot be said for witnesses, who are bound to tell the truth. Here, lie and memory detection tools appear less problematic in the investigation phase, where they can be used to obtain insights to steer further inquiries. Conversely, the currently available technologies do not reach adequate standards of reliability to be admitted as evidence at trial. The absence of clear scientific foundations, especially for emotional AI techniques, could indeed jeopardize the chances for the defendant to effectively comment on the evidence in an adversarial proceeding.  

For some, overcoming the technical limitations of advanced memory and lie detection tools seems to be a matter of time. If this ever happens, many questions are still open from a legal and ethical perspective. Does the brain really deserve special protection? Or are we already accepting equally invasive evidence-gathering acts in criminal proceedings? What is the role of human dignity, and how it relates to other fundamental rights, like privacy, the freedom of thought e defence rights? The intersections between some of the fundamental rights at stake is undoubtedly fuzzy at this stage. Clearly, more research is needed to see which is the place, if any, of lie detection tools in criminal justice.  

Bibliography  

Article19. (2021). Emotional Entanglement: China’s emotion recognition market and its implications for human rights. https://www.article19.org/wp-content/uploads/2021/01/ER-Tech-China-Report.pdf 

Ashworth, A., & Zedner, L. (2014). Preventive justice (First edition). Oxford University Press. 

Barrett, L. F. (2016). The theory of constructed emotion: An active inference account of interoception and categorization. Social Cognitive and Affective Neuroscience, nsw154. https://doi.org/10.1093/scan/nsw154 

Brennen, T., & Magnussen, S. (2023). Lie Detection: What Works? Current Directions in Psychological Science, 32(5), 395–401. https://doi.org/10.1177/09637214231173095 

Brkan, M. (2018). The Concept of Essence of Fundamental Rights in the EU Legal Order: Peeling the Onion to its Core. European Constitutional Law Review, 14(2), 332–368. https://doi.org/10.1017/S1574019618000159 

Bublitz, J. C. (2022). Novel Neurorights: From Nonsense to Substance. Neuroethics, 15(1), 7. https://doi.org/10.1007/s12152-022-09481-3 

Caianiello, M. (2020). Right To Remain Silent and Not to Incriminate Oneself in the European Union System. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3743329 

Chen, C., Crivelli, C., Garrod, O. G. B., Schyns, P. G., Fernández-Dols, J.-M., & Jack, R. E. (2018). Distinct facial expressions represent pain and pleasure across cultures. Proceedings of the National Academy of Sciences, 115(43). https://doi.org/10.1073/pnas.1807862115 

Crawford, K. (2013). The Hidden Biases in Big Data. Harward Business Review. https://hbr.org/2013/04/the-hidden-biases-in-big-data 

Crawford, K. (2021). Time to regulate AI that interprets human emotions. Nature, 592(7853), 167–167. https://doi.org/10.1038/d41586-021-00868-5 

Díaz Soto, J. M., & Borbón, D. (2022). Neurorights vs. neuroprediction and lie detection: The imperative limits to criminal law. Frontiers in Psychology, 13, 1030439. https://doi.org/10.3389/fpsyg.2022.1030439 

Ferrua, P. (2020). Neuroscientific Evidence and Criminal Trials. In A. D’Aloia & M. C. Errigo (A c. Di), Neuroscience and Law. Springer. 

Greely, H. T., & Farahany, N. A. (2019). Neuroscience and the Criminal Justice System. Annual Review of Criminology, 2(1), 451–471. https://doi.org/10.1146/annurev-criminol-011518-024433 

Haut, K. G., Sen, T., Lomakin, D., & Hoque, E. (2022). A Mental Trespass? Unveiling Truth, Exposing Thoughts, and Threatening Civil Liberties With Noninvasive AI Lie Detection. IEEE Transactions on Technology and Society, 3(2), 132–142. https://doi.org/10.1109/TTS.2022.3172556 

Haynes. (2012). Brain reading. In S. D. Richmond, G. Rees, & S. J. L. Edwards (A c. Di), I Know What You’re Thinking: Brain imaging and mental privacy. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199596492.001.0001 

Hertz, N. (2023). Neurorights – Do we Need New Human Rights? A Reconsideration of the Right to Freedom of Thought. Neuroethics, 16(1), 5. https://doi.org/10.1007/s12152-022-09511-0 

Ienca, M. (2021). On Neurorights. Frontiers in Human Neuroscience, 15, 701258. https://doi.org/10.3389/fnhum.2021.701258 

Ienca, M., & Andorno, R. (2017). Towards new human rights in the age of neuroscience and neurotechnology. Life Sciences, Society and Policy, 13(1), 5. https://doi.org/10.1186/s40504-017-0050-1 

Jupe, L. M., & Keatley, D. A. (2020). Airport artificial intelligence can detect deception: Or am i lying? Security Journal, 33(4), 622–635. https://doi.org/10.1057/s41284-019-00204-7 

Katwala, A. (2019). The race to create a perfect lie detector and the dangers of succeeding. The Guardian. https://www.theguardian.com/technology/2019/sep/05/the-race-to-create-a-perfect-lie-detector-and-the-dangers-of-succeeding 

Koops, B.-J., Newell, B. C., Timan, T., Chokrevski, T., & Gali, M. (2017). A Typology of Privacy. 38. 

Lasagni, G. (2021). La falsa confessione come causa di errori giudiziari. In L. Lupária, M. Cagossi, & A. Macrillò (A c. Di), L’errore giudiziario. Giuffrè Francis Lefebvre. 

Lasagni, G. (2023). Admissibility of evidence in criminal proceedings: Lessons (and problems) from the “data retention saga”. In Admissibility of Evidence in EU Cross-Border Criminal Proceedings: Electronic Evidence, Efficiency and Fair Trial Rights. Hart Publishing. 

Ligthart, S., Douglas, T., Bublitz, C., Kooijmans, T., & Meynen, G. (2021). Forensic Brain-Reading and Mental Privacy in European Human Rights Law: Foundations and Challenges. Neuroethics, 14(2), 191–203. https://doi.org/10.1007/s12152-020-09438-4 

Ligthart, S., Meynen, G., & Douglas, T. (2022). Persuasive Technologies and the Right to Mental Liberty: The ‘Smart’ Rehabilitation of Criminal Offenders. In M. Ienca, O. Pollicino, L. Liguori, E. Stefanini, & R. Andorno (A c. Di), The Cambridge Handbook of Information Technology, Life Sciences and Human Rights (1a ed., pp. 32–47). Cambridge University Press. https://doi.org/10.1017/9781108775038.005 

McStay, A. (2020). Emotional AI, soft biometrics and the surveillance of emotional life: An unusual consensus on privacy. Big Data & Society, 7(1), 205395172090438. https://doi.org/10.1177/2053951720904386 

Moreno, J. D., & Parashar, S. (2012). National security, brain imaging, and privacy. 

Neroni Rezende, I. (2022). Facial Recognition for Preventive Purposes: The Human Rights Implications of Detecting Emotions in Public Spaces. In L. Bachmaier Winter & S. Ruggeri (A c. Di), Investigating and Preventing Crime in the Digital Era (Vol. 7, pp. 67–98). Springer International Publishing. https://doi.org/10.1007/978-3-031-13952-9_4 

Nussbaum. (s.d.). Upheavals of Thought. The Intelligence of Emotions. 

Oravec, J. A. (2022). The emergence of “truth machines”?: Artificial intelligence approaches to lie detection. Ethics and Information Technology, 24(1), 6. https://doi.org/10.1007/s10676-022-09621-6 

Panzavolta, M., & Maes, E. (2022). Exclusion of evidence in times of mass surveillance. In search of a principled approach to exclusion of illegally obtained evidence in criminal cases in the European Union. The International Journal of Evidence & Proof, 26(3), 199–222. https://doi.org/10.1177/13657127221088328 

Podoletz, L. (2023). We have to talk about emotional AI and crime. AI & SOCIETY, 38(3), 1067–1082. https://doi.org/10.1007/s00146-022-01435-w 

Redmayne, M. (2005). Rethinking the Privilege Against Self-Incrimination. Oxford Journal of Legal Studies, 27(2), 209–232. https://doi.org/10.1093/ojls/gql001 

Richmond, S. (2012). Brain imaging and the transparency scenario. 

Sánchez-Monedero, J., & Dencik, L. (2022). The politics of deceptive borders: ‘Biomarkers of deceit’ and the case of iBorderCtrl. Information, Communication & Society, 25(3), 413–430. https://doi.org/10.1080/1369118X.2020.1792530 

Schabas, W. (2015). European Convention on human rights: A commentary (1st ed.). Oxford University Press. 

Schwartz, O. (2019). Don’t look now: Why you should be worried about machines reading your emotions. The Guardian. https://www.theguardian.com/technology/2019/mar/06/facial-recognition-software-emotional-science 

Shen, F. (2013). Neuroscience, Mental Privacy, and the Law. Harvard Journal of Law & Public Policy, 36(2), 653–714. 

US Government Accountability Office. (2013). Aviation Security: TSA Should Limit Future Funding for Behavior Detection Activities. https://www.gao.gov/products/gao-14-159 


[1] Art. 112duodecies. of the Belgian Code of Criminal Procedure (Code d’instruction criminelle), introduced by the Law of 4 February 2020, Loi modifiant le Code d’instruction criminelle en ce qui concerne l’utilisation du polygraphe.

[2] Art. 2 of the implementing royal decree identifies the techniques for the polygraph test, which certainly excludes the use of neuroscientific tools and EFR. Instead, the provision does not seem to explicitly exclude the use of AI models to interpret the physiological inputs gathered by the device’s sensors. See the Royal Decree of 28 June 2021, Arrêté royal portant exécution de l’article 112duodecies, § 4, alinéa 3, et § 7, du Code d’instruction criminelle déterminant les informations minimales devant figurer dans le procès-verbal de consentement et portant établissement des exigences techniques auxquelles l’appareil avec lequel le test polygraphique est effectué, doit répondre.

3 Reportedly, the UK police piloted an initiative using an AI lie detection software, EyeDetect, to measure the rehabilitation of sex offenders (Oravec, 2022, p. 6). For other examples, outside Europe, see below Section 3. 

[4] See Explanation on Article 1. Explanations relating to the Charter of Fundamental Rights OJ C 303, 14.12.2007, pp.17–35.

[5] While this reasoning has been explicitly purported for neurodata, it can easily be extended to emotional data too.

[6] The concept of human dignity has been compared to the essence of the right foreseen in Art. 52(1) CFREU, for which the jurisprudence has not been able to provide an abstract definition (Brkan, 2018). It is considered indeed that its content can only be determined in relation to a concrete case, and the same arguably applies to human dignity.

[7] ECtHR, 1 July 2014, S.A.S. v. France, app. nos. 43835/11, §55; ECtHR, 26 April 2016, İzzettin Doğan and others v. Turkey, app. nos. 62649/10, §68.

[8] This principle is closely related to the presumption of innocence, which formally applies in criminal proceedings only. In preventive contexts instead, the principle mandates respect for each individual’s status as a responsible agent in society (Ashworth & Zedner, 2014, p. 66).

[9] See CJEU, 8 April 2014, Digital Rights Ireland and Others, Joined Cases C-293/12 and C-594/12, §37; CJEU, 06 October 2020, La Quadrature du Net and Others v Premier ministre and o., Joined Cases C-511/18, C-512/18 and C-520/18, §143.

[10] Specifically, many EFR technologies are based on the theory of “micro-expressions”, formulated by the psychologist Paul Ekmann in the 1960s. The argument is that all human emotions can be reduced to six micro-expressions”. One of the criticalities of this thesis is its supposedly discriminatory nature, which would blatantly ignore social, cultural and contextual factors impacting on the display of emotions (Chen et al., 2018). The theory was also used to design the technologies used in the SPOT program, which eventually was discontinued for lack of results and potentially discriminatory effects (Schwartz, 2019; US Government Accountability Office, 2013).

[11] CJEU, 8 July 1999, Montecatini S.p.A, Case C-235/92, §175.

[12] Directive (EU) 2016/343 of the European Parliament and of the Council of 9 March 2016 on the strengthening of certain aspects of the presumption of innocence and of the right to be present at the trial in criminal proceedings OJ L 65, 11.3.2016, pp.1–11.

[13] ECtHR, 17 December 1996, Saunders v. UK, app. no. 19187/91, §69.

[14] With the only difference that, while current scientific standards allow us to interpret DNA evidence with sufficient reliability, the same cannot be said with regard to brain-based and external markers of emotions.

[15] Certainly, deception detection techniques can impact the target’s behaviors (e.g., causing stress), but their aim is measuring the individual’s reactions rather than determining a change in his or her mental state/brain. In this sense, they differ from “neurointerventions”, which are  “interventions that seek to alter a person’s brain and thus mind through direct (that is, not psychologically mediated) means, and that are performed without the valid consent of the person being targeted” (Ligthart et al., 2022, p. 34).

[16] ECtHR, 1 June 2010, Gäfgen v. Germany, app. no. 22978/05.

[17] See ECtHR, 21 January 1999, Garcia Ruiz v. Spain, app. no. 30544/96, §28;ECtHR, 12 July 1988, Schenk v. Switzerland, app. no. 10862/84, §§45-46; ECtHR, 26 September 1996, Miailhe v. France (No. 2), app. no. 18978/91, §43;ECtHR, 1 March 2007, Heglas v. the Czech Republic, app. no. 5935/02, §84; ECtHR, 19 July 2012, Hümmer v. Germany, appl. no. 29881/07; ECtHR, 11 July 2017, Moreira Ferreira v. Portugal (no. 2), app. no. 19867/12, §83.

[18] See La Quadrature du Net andOthers, §227;CJEU,6 October2020,Privacy International v Secretary of State for ForeignandCommonwealth Affairs and Others, Case C‑623/17,§44.The rule was inspired byECtHR, 18 March 1997,Mantovanelli v.France,App. no.21497/93

Leave a Reply

Your email address will not be published. Required fields are marked *

Facebook
Twitter
LinkedIn
WhatsApp
Email

Related Posts

Sign up for our Newsletter