International Journal of Law, Politics and Governance

DOI: 10.64823/ijlpg.2602003

Scientific Interrogation Techniques vs. Constitutional Safeguards: Narco-Analysis, Polygraph and BEAP in India

⚠️ This HTML version is automatically generated from the manuscript file and may contain formatting or data discrepancies compared to the original paper. Please refer to the PDF version for the authoritative, publisher-formatted record.

Introduction

Confessions have been an important tool of criminal investigation in India since time immemorial. Overstretched, understaffed and inadequately trained police forces have little choice but to take the quickest path to a solved case, and a statement from the accused is a convenient place to start. The law has long been uneasy about this tendency: the Bharatiya Sakshya Adhiniyam, 2023 (BSA) [1], like the Indian Evidence Act, 1872 [2] before it, renders confessions made to police officers inadmissible. The Constitution went one step further and made the right against self-incrimination a fundamental right [3].

Into this setting arrived a group of techniques that promised to sidestep the problem. If the accused would not talk, perhaps a drug would make him talk. If he lied, perhaps a machine could detect it. If he said nothing at all, perhaps his brain would betray him. Narco-analysis, the polygraph and BEAP were promoted as the “scientific and humane” alternative to the third degree. They were used in the Telgi stamp paper scam, the Nithari killings, the Aarushi Talwar murder and the Sister Abhaya case, and video recordings of drugged accused persons were broadcast on television.

The question these techniques raise is a simple one: can the State, in the name of investigative efficiency, intrude into a person’s mind without his consent? In 2010, the Supreme Court answered it with a “no” [4]. Fifteen years later the answer remains the same, but it continues to be tested by investigating agencies and, occasionally, by High Courts. With the new criminal laws in force since 1 July 2024 [1], [5], this is an appropriate moment to revisit the subject.

Understanding the Techniques

Narco-analysis involves the intravenous injection of a barbiturate, usually sodium pentothal and sometimes sodium amytal, in a carefully controlled dose that induces a semi-conscious or hypnotic state. The premise is that in this twilight state the subject’s capacity for imagination and deliberate deception is subdued, so that he speaks more freely. Pharmacology has, however, recognised for decades that a drugged subject is highly suggestible. He may fantasise, confabulate or simply agree with the questioner. A practised deceiver can hold on to his lie, while a suggestible innocent person can narrate events that never occurred.

The polygraph, popularly called the lie detector, records the subject’s blood pressure, pulse rate, respiration and galvanic skin resistance while he is questioned. The hypothesis is that deception generates anxiety and that anxiety produces measurable physiological change. Its central weakness is that it measures arousal, not deception: the same reading can be caused by fear, nervousness, anger or an underlying medical condition, and countermeasures can be taught [6].

BEAP, also known as P300 testing or “brain fingerprinting”, records the brain’s electrical activity while the subject is exposed to words, sounds or images connected with the crime. It rests on the premise that a particular brain wave, the P300, is generated when the brain encounters something familiar [7], [8]. Brain Electrical Oscillation Signature (BEOS) profiling is a related technique used in forensic science laboratories in Gujarat and Maharashtra. Unlike the polygraph, BEAP and BEOS involve no questioning at all. This feature was relied upon to argue that they fall outside Article 20(3), an argument the Supreme Court rejected [4].

What the three techniques share is their target. Fingerprints, blood samples and handwriting relate to the body. These tests are psychological: they are designed to uncover what a person knows, believes or remembers.

The Constitutional Framework

Article 20(3) and the Meaning of “To Be a Witness”

Article 20(3) provides that no person accused of any offence shall be compelled to be a witness against himself [3]. Three elements must coexist: the person must be accused of an offence, there must be compulsion, and the compulsion must be to be a witness against himself.

The earliest reading was expansive. In M.P. Sharma v. Satish Chandra [9], an eight-judge Bench held that “to be a witness” means to furnish evidence, whether oral or documentary.

An eleven-judge Bench in State of Bombay v. Kathi Kalu Oghad [10] narrowed that sweep. The majority held that giving specimen handwriting, signatures, fingerprints or thumb impressions does not make a person a witness, because such material does not convey personal knowledge of the facts of the case and is furnished only for comparison. To be a witness, the Court explained, is to impart knowledge of relevant facts through an oral or written statement. This distinction between testimonial and physical evidence has remained the cornerstone of Article 20(3) jurisprudence ever since.

In Nandini Satpathy v. P.L. Dani [11], Krishna Iyer J., speaking for the Court, carried the protection into the police station. The privilege extends to police interrogation and covers answers that may incriminate the person in other offences as well; and compulsion includes not only physical force but also psychological pressure, an intimidating atmosphere and overbearing methods of interrogation.

Article 21 and Mental Privacy

Since Maneka Gandhi v. Union of India [12], any procedure that curtails life or personal liberty must be fair, just and reasonable. In D.K. Basu v. State of West Bengal [13], the Supreme Court laid down detailed guidelines to protect the dignity of persons in custody, making it clear that arrest does not suspend dignity.

In K.S. Puttaswamy v. Union of India [14], a nine-judge Bench recognised privacy as a fundamental right and identified the privacy of the mind as one of its facets, a holding of direct relevance here. Thoughts, memories and mental processes now lie within a zone of constitutional protection, and any intrusion into that zone must satisfy the requirements of legality, legitimate aim and proportionality.

The Statutory Background

Two statutory provisions complete the picture. Section 161(2) of the Code of Criminal Procedure, 1973 (CrPC) [15], now Section 180(2) of the Bharatiya Nagarik Suraksha Sanhita, 2023 (BNSS) [5], requires a person examined by the police to answer truly all questions put to him, other than those whose answers would expose him to a criminal charge. Section 25 of the Evidence Act [2], now Section 23(1) of the BSA [1], makes confessions to a police officer inadmissible.

The High Courts Before Selvi

Before 2010, the High Courts largely sided with the investigating agencies. In Ramchandra Reddy v. State of Maharashtra [16], arising out of the Telgi stamp paper scam, the Bombay High Court held that the administration of narco-analysis, P300 and polygraph tests did not by itself amount to compelling the accused to be a witness, and that statements made during such tests could be used where they led to discoveries.

The Madras High Court followed suit in Dinesh Dalmia v. State [17], observing that narco-analysis is merely a step in the investigation and that the accused is not made a witness merely because the test is conducted. In Rojo George v. Deputy Superintendent of Police [18], arising out of the Sister Abhaya investigation, the Kerala High Court held that investigative techniques must keep pace with the sophistication of crime. The Gujarat High Court took the same view in Santokben Sharmanbhai Jadeja v. State of Gujarat [19].

The high-water mark of this phase came from a trial court. In June 2008, a Sessions Court in Pune convicted Aditi Sharma of murdering her former fiancé, relying in part on a BEOS test that purported to establish her “experiential knowledge” of the crime [20]. The conviction drew criticism from Indian and international neuroscientists [21], and the Bombay High Court later suspended her sentence and released her on bail. Meanwhile, an expert committee involving NIMHANS had seriously questioned the scientific validity of BEOS.

The reasoning in these decisions suffered from two flaws. First, they assumed that there is no compulsion unless physical force is used, a premise Nandini Satpathy had already rejected. Second, they asked whether the test produced a confession, rather than whether it extracted personal knowledge, which was precisely the question Kathi Kalu Oghad had identified as decisive.

Selvi v. State of Karnataka: The Turning Point

In Selvi v. State of Karnataka [4], a three-judge Bench led by Chief Justice K.G. Balakrishnan heard a batch of appeals challenging the use of all three techniques. Its principal holdings are as follows.

Testimonial character. Narco-analysis yields verbal statements and is plainly testimonial. In the polygraph and BEAP the subject does not have to speak, but the Court held that these tests still draw inferences about his knowledge or beliefs from his physiological responses. An inference about personal knowledge of a fact falls squarely within the meaning of testimony settled in Kathi Kalu Oghad.

Compulsion. A person who is drugged or wired up against his will has no choice whether or not to communicate. Compulsory administration therefore amounts to testimonial compulsion.

Persons protected. Read with Section 161(2) CrPC [15], the privilege extends not only to persons formally accused but also to suspects and witnesses whose answers could expose them to a criminal charge.

Article 21. Compelling a person to undergo these tests is an unwarranted intrusion into personal liberty and mental privacy and may amount to cruel, inhuman or degrading treatment. The Court rejected the argument that the compelling public interest in solving grave crimes such as terrorism could justify the intrusion.

Statutory interpretation. The Explanation to Sections 53, 53A and 54 CrPC, inserted by the 2005 amendment, covers certain enumerated tests and “such other tests” as the registered medical practitioner considers necessary in a particular case. Applying the rule of ejusdem generis, the Court held that these general words take their colour from the enumerated tests, all of which are physical. A testimonial technique cannot be smuggled in through them.

Voluntary tests. The Court did not ban the techniques outright. They may be administered with the informed consent of the subject. Even then, the results are not admissible as such, because the subject does not exercise conscious control over his responses. However, information or material subsequently discovered with the help of a voluntarily administered test may be admitted under Section 27 of the Evidence Act [2].

Safeguards. The Court directed strict compliance with the National Human Rights Commission’s 2000 Guidelines for the Administration of Polygraph Test on an Accused [22], and extended them to narco-analysis and BEAP. Under these guidelines the subject must be told that he may refuse; he must have access to a lawyer and be informed of the physical, emotional and legal implications of the test; his consent must be recorded before a Judicial Magistrate; the test must be conducted by an independent agency in the presence of his lawyer; and a complete record must be maintained. Any statement made during the test is to be treated as a statement to the police, not as a confession.

Selvi’s real achievement was to carry the distinction between testimonial and physical evidence into a new technological setting without allowing the form of the technology to dictate the result. A test that reads the mind is testimonial, whether the person speaks or sits in silence while his brain waves are recorded.

After Selvi: Consolidation and Pressure

Courts Directing Tests

Selvi did not settle matters in practice. Agencies continued to apply for the tests, and courts sometimes granted them. In Sangitaben Shaileshbhai Datana v. State of Gujarat [23], the Gujarat High Court, while deciding a bail application, had directed lie-detector, brain-mapping and narco-analysis tests. The Supreme Court set aside the direction as contrary to Selvi and pointed out that a bail application must be decided on considerations relevant to bail.

The issue arose again in Amlesh Kumar v. State of Bihar [24], decided on 9 June 2025 by a Bench of Sanjay Karol and Prasanna B. Varale JJ. The Patna High Court, while hearing a bail application in a dowry case, had accepted the investigating officer’s request to subject all the accused to narco-analysis. The Supreme Court set that order aside. It reiterated that involuntary tests violate Articles 20(3) and 21, and that nothing derived from an involuntary test may be admitted in evidence in any proceeding. It added, citing Manoj Kumar Soni v. State of Madhya Pradesh [25] and Vinobhai v. State of Kerala [26], that even the report of a voluntary test cannot by itself form the basis of a conviction.

The Accused’s Request for a Test

Amlesh Kumar also took up a question that Selvi had left open: can an accused demand a narco-analysis test to establish his innocence? The Supreme Court did not go as far as the Rajasthan High Court in Sunil Bhatt v. State of Rajasthan [27], which had read such a right into the right to lead defence evidence under Section 233 CrPC. It held that the accused has no indefeasible right to a narco-analysis test, but may apply for one at the appropriate stage of trial, and the court may permit it after considering the circumstances as a whole, including free consent and proper safeguards. This is a sensible compromise. An unrestricted right would allow a well-resourced accused to manufacture an exculpatory record from an unreliable test, while an absolute bar would sit uneasily with the right to a fair defence.

Related Developments

In Ritesh Sinha v. State of Uttar Pradesh [28], the Supreme Court held that a Judicial Magistrate may order a person to give a voice sample, treating it as physical evidence comparable to handwriting. Parliament has since codified this power in Section 349 BNSS [5], which allows a Magistrate to order any person, whether or not arrested, to give specimen signatures, handwriting, voice samples or finger impressions. What Parliament left out is significant. In codifying Ritesh Sinha, it made no provision for narco-analysis, polygraph or BEAP. That silence confirms that these techniques do not belong to the realm of physical evidence.

In practice, the tests continue to surface in sensational cases, now generally with recorded consent. In the 2022 Shraddha Walkar murder case in Delhi, the accused underwent polygraph and narco-analysis tests after consenting before the court. Consent is thus present in form, though not necessarily in substance.

The Position under the New Criminal Laws

The BNSS and the BSA replaced the CrPC and the Evidence Act with effect from 1 July 2024. The relevant provisions have been largely carried over.

Section 51 BNSS corresponds to Section 53 CrPC and permits the medical examination of an arrested person at the request of a police officer. The Explanation retains the residuary words “such other tests which the registered medical practitioner thinks necessary in a particular case”. Since the enumerated tests remain physical in nature, the ejusdem generis reading adopted in Selvi applies with equal force. Section 180 BNSS reproduces Section 161 CrPC, including the protection in sub-section (2).

Section 23 BSA consolidates the bars on confessions to the police and in police custody, and its proviso reproduces the discovery rule of the old Section 27. That proviso is therefore the route by which material discovered after a voluntary test can enter the record. Expert opinion is governed by Section 39 BSA, which corresponds to Section 45 of the Evidence Act.

A more serious concern is the Criminal Procedure (Identification) Act, 2022 [29], which replaced the Identification of Prisoners Act, 1920 [30]. Section 2(1)(b) defines “measurements” to include “behavioural attributes including signatures, handwriting or any other examination referred to in section 53 or section 53A” of the CrPC. The expression “behavioural attributes” is not defined, and critics have suggested that it could be read to cover polygraph or BEAP-type examinations. Any such reading would collide head-on with Selvi, since the Act permits measurements to be taken from convicts, arrested persons and persons under preventive detention, and makes resistance an offence. The provision ought to be read down to exclude testimonial techniques.

The Problem of Scientific Reliability

Even if the constitutional objections are set aside, the scientific case for these techniques is weak. Statements obtained under “truth serum” have long troubled comparative jurisprudence. In Townsend v. Sain [31], the United States Supreme Court held that a confession obtained after the accused had been given a drug with truth-serum properties could not be treated as the product of a free will and was therefore involuntary.

Polygraph evidence has a long and unhappy history. Frye v. United States [32], the case that gave American law its “general acceptance” test, concerned an early systolic blood pressure deception test, and the evidence was excluded. In United States v. Scheffer [33], the Supreme Court upheld Military Rule of Evidence 707, which bars polygraph evidence in courts-martial altogether, reasoning that there was no scientific consensus on the reliability of polygraph evidence. A comprehensive review by the National Research Council reached a similarly sceptical conclusion about the scientific basis of polygraph testing [6].

BEAP and BEOS rest on an even slenderer foundation. A P300 response indicates recognition, not guilt [8]. An innocent person may recognise the stimuli because he read about the crime in a newspaper, was taken to the scene or was shown photographs by investigators. BEOS has not undergone independent peer review, and no published validation study has answered the doubts raised by the NIMHANS committee. A method of unproven reliability, whatever it is called, cannot satisfy the requirements for expert evidence under Section 39 BSA.

Comparative Perspectives

The reasoning in Schmerber v. California [34] is closely analogous. The United States Supreme Court held that the compelled extraction of blood did not violate the Fifth Amendment, because the privilege protects only communications or testimony. Significantly, the Court noted that lie-detector tests, although they measure physical responses, may be essentially testimonial because they seek to elicit evidence of guilt. Selvi drew on this observation.

Article 14(3)(g) of the International Covenant on Civil and Political Rights guarantees the right not to be compelled to testify against oneself or to confess guilt [35]. India is more protective than many jurisdictions in extending the bar to suspects and witnesses, but lags behind others in leaving the safeguards governing voluntary tests to judge-made rules rather than legislation.

Critical Appraisal: How Real Is Consent?

The framework built in Selvi rests on consent. That choice is principled: it respects individual autonomy and leaves an innocent person free to take the test if he wishes. Its weakness lies in the circumstances in which consent is sought.

It is difficult for a person in police custody to give consent freely in any meaningful sense. He may be told, expressly or by implication, that refusal will be treated as a sign of guilt, that it will make bail harder to obtain, or that the test is a mere formality that will clear his name. In its 180th Report, the Law Commission opposed permitting adverse inferences to be drawn from silence [36], even though the Malimath Committee later recommended that courts be allowed to draw them [37]. Whatever the formal legal position, the practical inference that investigators, the media and sometimes the public draw from a refusal is far harder to eliminate.

The Judicial Magistrate is the principal safeguard, yet the guidelines say little about the inquiry the Magistrate must conduct. Unless the Magistrate is satisfied, in the absence of the police, that the subject understands the nature of the drug or procedure, its risks, and the fact that refusal carries no legal consequence, recording consent can easily become a routine formality.

The second concern is derivative use. Selvi held that facts discovered as a result of a voluntary test may be proved under Section 27 of the Evidence Act, now the proviso to Section 23 BSA. This creates an unfortunate incentive: the test result is inadmissible, but its fruits are not. Amlesh Kumar’s insistence that a voluntary test cannot by itself sustain a conviction helps, but the framework still invites agencies to use the test as an investigative shortcut.

Third, there is the problem of leakage. Recordings of narco-analysis sessions have in the past been broadcast publicly. This damages reputations and prejudices fair trial, yet there is no specific penal consequence for it.

The fourth concern is the use of these tests on victims and witnesses. In the 2020 Hathras case, the State Government drew criticism after it announced polygraph and narco-analysis tests for those connected with the case, including members of the victim’s family.

Suggestions

The following reforms are proposed:

  1. Parliament should enact dedicated legislation on forensic psychological techniques that places the NHRC Guidelines and the Selvi conditions on a statutory footing. Consent should be recorded by a Judicial Magistrate after a private interaction with the subject, in the presence of counsel.
  2. The law should make clear that these techniques fall outside both the Explanation to Section 51 BNSS and the definition of “measurements” under the Criminal Procedure (Identification) Act, 2022.
  3. There should be a reasonable interval between consent and the test, and the subject should be entitled to withdraw consent at any time.
  4. Tests should be conducted only in accredited laboratories using independently validated protocols, and the subject should be entitled to a complete recording of the test.
  5. Unauthorised disclosure of recordings or results should be made a specific offence.
  6. The rule laid down in Sangitaben and Amlesh Kumar, that a court hearing a bail application or a petition for quashing must neither direct nor rely on such tests, should be enshrined in statute.
  7. Finally, investment in DNA laboratories, digital forensics and trained investigators will secure more convictions than any technique that claims to read minds.

Conclusion

The three techniques discussed in this paper share an aim that the Constitution regards with deep suspicion: they seek to extract from a person’s mind what he has chosen not to disclose. Article 20(3) is not a mere rule of evidence. It embodies the principle that the State must prove its case through its own efforts and cannot conscript the accused into that enterprise. After Puttaswamy, Article 21 recognises the mind as the most intimate zone of a person’s life.

Selvi translated these principles into a workable rule, and Amlesh Kumar has kept that rule firmly in place. The task now lies with Parliament. A consent framework that exists only in judgments and NHRC guidelines, sitting beside a residuary clause in Section 51 BNSS and an undefined expression in the Identification Act, remains exposed to erosion every time a sensational case generates public pressure for quick answers. Until that framework is written into law and the science behind these techniques is subjected to rigorous independent scrutiny, the machine that claims to read the guilty mind should remain where it belongs: at the margins of Indian criminal procedure, and well outside its core.

Funding

This research received no external funding.

Conflict of Interest

The author declares no conflict of interest.

Data Availability Statement

Not applicable. This is a doctrinal study; no new data were created or analysed.

AI Usage Disclosure

[Author to state any use of generative AI or AI-assisted tools, including the tool, version and purpose, or state: “No generative AI tools were used.”]

Author Contributions

Vankdoth Naveen Nayak Raj is the sole author and was responsible for conceptualization, research, writing—original draft, and writing—review and editing. The author has read and agreed to the published version of the manuscript.

References

  1. Bharatiya Sakshya Adhiniyam, 2023, No. 47, Acts of Parliament, 2023 (India).
  2. Indian Evidence Act, 1872, No. 1, Acts of Parliament, 1872 (India).
  3. Constitution of India, 1950, art. 20(3).
  4. Selvi v. State of Karnataka, (2010) 7 SCC 263 (India).
  5. Bharatiya Nagarik Suraksha Sanhita, 2023, No. 46, Acts of Parliament, 2023 (India).
  6. National Research Council, The Polygraph and Lie Detection. Washington, DC, USA: The National Academies Press, 2003.
  7. L. A. Farwell and E. Donchin, “The truth will out: Interrogative polygraphy (‘lie detection’) with event-related brain potentials,” Psychophysiology, vol. 28, no. 5, pp. 531–547, 1991.
  8. J. P. Rosenfeld, “‘Brain fingerprinting’: A critical analysis,” The Scientific Review of Mental Health Practice, vol. 4, no. 1, pp. 20–37, 2005.
  9. M.P. Sharma v. Satish Chandra, AIR 1954 SC 300 (India).
  10. State of Bombay v. Kathi Kalu Oghad, AIR 1961 SC 1808 (India).
  11. Nandini Satpathy v. P.L. Dani, (1978) 2 SCC 424 (India).
  12. Maneka Gandhi v. Union of India, (1978) 1 SCC 248 (India).
  13. D.K. Basu v. State of West Bengal, (1997) 1 SCC 416 (India).
  14. K.S. Puttaswamy v. Union of India, (2017) 10 SCC 1 (India).
  15. Code of Criminal Procedure, 1973, No. 2, Acts of Parliament, 1974 (India).
  16. Ramchandra Reddy v. State of Maharashtra, 2004 All MR (Cri) 1704 (Bombay High Court, India).
  17. Dinesh Dalmia v. State, 2006 Cri LJ 2401 (Madras High Court, India).
  18. Rojo George v. Deputy Superintendent of Police, 2006 (2) KLT 197 (Kerala High Court, India).
  19. Santokben Sharmanbhai Jadeja v. State of Gujarat, 2008 Cri LJ 68 (Gujarat High Court, India).
  20. State of Maharashtra v. Aditi Baldev Sharma (Sessions Court, Pune, Jun. 12, 2008) (India).
  21. A. Giridharadas, “India’s novel use of brain scans in courts is debated,” The New York Times, Sep. 14, 2008.
  22. National Human Rights Commission, “Guidelines for the administration of polygraph test (lie detector test) on an accused,” NHRC, New Delhi, India, 2000.
  23. Sangitaben Shaileshbhai Datana v. State of Gujarat, (2019) 14 SCC 522 (India).
  24. Amlesh Kumar v. State of Bihar, 2025 INSC 810 (India).
  25. Manoj Kumar Soni v. State of Madhya Pradesh, 2023 SCC OnLine SC 984 (India).
  26. Vinobhai v. State of Kerala, 2025 SCC OnLine SC 178 (India).
  27. Sunil Bhatt v. State of Rajasthan (Rajasthan High Court, 2022) (India).
  28. Ritesh Sinha v. State of Uttar Pradesh, (2019) 8 SCC 1 (India).
  29. Criminal Procedure (Identification) Act, 2022, No. 11, Acts of Parliament, 2022 (India).
  30. Identification of Prisoners Act, 1920, No. 33, Acts of Imperial Legislative Council, 1920 (India).
  31. Townsend v. Sain, 372 U.S. 293 (1963).
  32. Frye v. United States, 293 F. 1013 (D.C. Cir. 1923).
  33. United States v. Scheffer, 523 U.S. 303 (1998).
  34. Schmerber v. California, 384 U.S. 757 (1966).
  35. International Covenant on Civil and Political Rights, Dec. 16, 1966, United Nations Treaty Series, vol. 999, p. 171.
  36. Law Commission of India, “Article 20(3) of the Constitution of India and the right to silence,” Report No. 180, Government of India, New Delhi, India, 2002.
  37. Committee on Reforms of Criminal Justice System, Report, vol. I. New Delhi, India: Ministry of Home Affairs, Government of India, 2003.