Applicability of Shannon’s Entropy Formula to Legal Entropy in Epistemological Analysis: A Study of Narco-Analysis Test Evidence in Indian Judicial Decision-Making

I. Introduction

Any field of study addressing rules, orders and norms will always be interspersed with uncertainty. Because of the fact that society is in a state of transformation and therefore continuously changing. All of these changes, contribute to creating different levels uncertainty or lack of clarity with respect the articulation and appreciation and enforcement etc. in Indian judiciary. Modifications in the current laws and orders are essentially not scary. On the other hand, what lies between existing norms and new implemented rules is a period of time where ambiguity reigns supreme. This lack of transparency can get in the way of justice being served best, as how beholden is always case-by-case.

Numerous scholars have sought to conceptualize this uncertainty in terms of entropy within legal texts, as well as in their interpretation and application. However, few, if any, have attempted to quantify this uncertainty through a mathematical framework to determine its degree.² In his seminal paper, "Prediction and Entropy of Printed English," Shannon³ pioneered the application of information theory to the study of natural languages. Drawing on methods from theoretical physics, Shannon utilized mathematical tools he had developed to analyze and understand information. This work has since inspired a vast body of literature focused on measuring the information content in both written and spoken language.

A. Scope of Research

Shannon’s Entropy formula, a fundamental concept in information theory, measures the uncertainty or unpredictability of information in a system. Its quantitative application extends beyond theoretical contexts and can be explored in the legal domain, particularly in the analysis of evidence derived from narco-analysis tests.

By examining how entropy reflects the reliability of information extracted through these tests, we gain insight into the uncertainties and potential distortions inherent in such procedures. This analysis is crucial for understanding the epistemological value of evidence obtained under narco-analysis, where the accuracy and truthfulness of the data remain questionable. The intersection of Shannon's Entropy and narco-analysis evidence opens a unique framework for evaluating the effectiveness and admissibility of such evidence in judicial processes.

B. REVIEW OF LITERATURE

1. Sichelman Ted⁴ : In the paper titled “Quantifying Legal Entropy” reiterates the various approaches to legal entropy as explored in the literature and its application, including delineative entropy, interpretive entropy, and applicative entropy. The concept of applicative entropy is particularly pertinent to this study, as it addresses the uncertainties inherent in the legal framework surrounding narcoanalysis tests and their epistemological implications.

2. Amato D’⁵ : In the paper titled “Legal Uncertainty” gives a quantifiable predictability level around 0.5 in reference with uncertainty. As more rules approach this 0.5 threshold, legal uncertainty increases. A more intricate dynamic occurs as the predictability decreases from 0.5 toward zero. When predictability falls below 0.5, legal certainty tends to increase as a rule becomes obsolete, yet it also decreases as new rules are relied upon or established to replace the extinguished ones. This Article contends that the ability of lawyers to predict how courts will resolve actual or potential cases is becoming increasingly uncertain.

3. Chipiga V⁶ : In the paper titled “Legal Entropy in International Law” reiterates a multidisciplinary approach which deals with the uncertainty in international law. The constrained variability in issues of interest defines the range of possible options for implementing international law norms and principles within specific countries, events, and decisions by international judicial bodies, as illustrated by the European Court of Human Rights (ECHR).

The literature concerning the study of entropy within the legal system underscores that uncertainty is not confined to a single legal domain but rather extends across multiple disciplines. This multidisciplinary approach has the potential to yield quantifiable probability variables. The various types of entropy elucidate the complexities inherent in this field of study, particularly regarding the admissibility of narcoanalysis evidence within the Indian judiciary.

II. Analysis and outcomes

A. Epistemological Analysis of Narco-Analysis Test

A case will only be deemed admissible in the eyes of the judiciary if it possesses evidentiary value. Absent such value, it remains a mere narrative. Various methods can be employed to uncover the truth in a case; however, methods deemed substandard or coercive are prohibited, particularly when they are administered without the accused's consent. In the context of narco-analysis tests, obtaining the consent of the accused is crucial and fundamentally relevant.

Usage as Evidence:

The admissibility of evidence obtained through narco-analysis is contingent upon several factors, including whether the court has granted permission for the test. Without judicial consent, such evidence cannot be admitted. Sections 24 to 30 of the Indian Evidence Act, 1872, govern confessions and encompass both written and oral statements. Consequently, statements made or recorded during a narco-analysis test fall within the definition of a confession. However, as per the proviso to Section 27 of the Indian Evidence Act, statements obtained through coercion or intimidation are inadmissible under Section 24. Nevertheless, if a voluntary narco-analysis test leads to the recovery of any information or material, such evidence may be admissible under Section 27 of the Indian Evidence Act, 1872⁷.

Shannon’s entropy, a concept from information theory, quantifies uncertainty or randomness in a system. In the legal context, this principle can be applied to measure the uncertainty or unpredictability inherent in evidence, testimonies, or legal outcomes. For instance, in the Indian Evidence Act, which governs the admissibility and evaluation of evidence, entropy could be used to assess the reliability or informational value of evidence. Evidence with high entropy (high uncertainty) might be deemed less reliable, whereas evidence with low entropy (more predictable or consistent) could be considered more credible. This statistical approach could provide a more objective framework for judges and legal practitioners to evaluate the weight of evidence.

Narcoanalysis, a controversial investigative technique, involves administering drugs to a subject to lower their inhibitions and elicit truthful information. While its admissibility in Indian courts is debated, applying Shannon’s entropy principles could offer a scientific lens to analyze the reliability of information obtained through this method. For example, the entropy of the subject’s responses could be measured to determine the consistency and coherence of the information provided. A lower entropy value might indicate a higher degree of reliability, whereas higher entropy could suggest inconsistencies or fabrication. This approach could help address concerns about the voluntariness and accuracy of narcoanalysis results, potentially making it a more scientifically valid tool in legal proceedings.

B. Shannon Entropy Principle

Shannon entropy, introduced by Claude Shannon in his 1948 paper on information theory, measures the uncertainty or unpredictability in a set of outcomes. The formula for Shannon entropy is:

Where:

• H(X) is the Shannon entropy of the random variable X,
• is the probability of the i-th outcome
• “n” is the total number of possible outcomes,
• , is the logarithm base b typically base 2 (for bits), but it can also be natural log (base e) or base 10 depending on the context.⁸

C. Shannon Entropy Principle Applicability in Narco-Analysis in reference with relevant Case Law

To apply Shannon's entropy to narco analysis, we can consider the following:

1. Random Variable: The outcome of a narco analysis test. This can be categorized as:

• Confession: The individual admits to involvement in a crime.
• No Confession: The individual denies involvement or provides irrelevant or inconsistent information.
• Inconclusive: The test results are inconclusive or cannot be interpreted reliably.

III. Historical data: case laws for deriving probabilities

In Townsend v. Sain⁹, it was held that evidence obtained from a person under the influence of drugs, such as a truth serum, is inadmissible in court. Similarly, in M.P. Sharma v. Satish Chandra¹⁰, the Supreme Court observed that the protection afforded by Article 20(3) of the Constitution, which prohibits self-incrimination, extends to evidence compelled outside the courtroom, emphasizing the phrase "to be a witness" rather than "to appear as a witness." In State of Bombay v. Kathi Kalu Ogha¹¹, the court clarified that self-incrimination involves providing information based on personal knowledge and does not encompass the mere mechanical production of documents. Lastly, in Dinesh Dalmia v. State¹² by SPE, CBI (2006), the Madras High Court ruled that subjecting a person to narco-analysis does not equate to testimony by conclusion, asserting that while a person may be taken for such tests against their will, the information revealed during the test is considered voluntary.

Probabilities: Based on historical data, expert opinions, and the specific circumstances of the case, we can assign probabilities to each outcome. For example:

• P(Confession) = 0.3
• P(No Confession) = 0.5
• P(Inconclusive) = 0.2

Calculating Entropy: Using the Shannon's entropy formula:

where:

• H(X) is the entropy of the narco analysis test outcome.
• p(x) is the probability of each outcome.

Therefore, the entropy of the narco analysis test would be:

H(X) = -[0.3 * log₂(0.3) + 0.5 * log₂(0.5) + 0.2 * log₂(0.2)]

Interpretation of Results:

• Higher Entropy: A higher entropy value indicates greater uncertainty or unpredictability in the outcome of the narco analysis test. This could suggest that the test results are unreliable or that the individual's state of mind during the test is significantly influencing their responses.
• Lower Entropy: A lower entropy value suggests more certainty in the outcome. However, even a low entropy value does not guarantee the accuracy of the confession. Factors such as the individual's mental state, the quality of the interrogation, and the reliability of the drug used can still influence the results.

Applying the Shannon entropy principle to the Selvi v. State of Karnataka (2010) case can offer a fresh perspective on the uncertainty and information flow in the context of forced narco-analysis tests, polygraph tests, and brain-mapping techniques. Here's how the principle can be conceptually applied:

1. Information as Entropy in Self-Incrimination

Shannon entropy measures the uncertainty or unpredictability of information in a system. Higher entropy corresponds to greater unpredictability, while lower entropy suggests more predictability. In the Selvi case, the core constitutional issue revolves around whether the accused can be forced to reveal personal knowledge (information) via Narco-Analysis or other investigative techniques. The accused's knowledge is the "information" in question. From an information theory standpoint, the accused's personal knowledge has high entropy because it's private and unpredictable by external forces. The state, through narco-analysis or other tests, attempts to reduce this entropy by extracting previously hidden or uncertain information about the crime.

2. Reduction of Entropy through Narco-Analysis

Narco-analysis tries to lower the entropy of the criminal investigation process by compelling individuals to provide otherwise uncertain or hidden information (confessions, details of the crime). This forced reduction of entropy violates the constitutional right against self-incrimination under Article 20(3), as the state artificially reduces the uncertainty (entropy) associated with the accused's knowledge without their voluntary consent.

3. Voluntariness and Entropy in Evidence

• According to the Shannon principle, the true value of information (entropy) comes from its inherent unpredictability. When the accused voluntarily discloses information, the entropy is reduced naturally and the information has high legal reliability.
• However, forced disclosures through narco-analysis or other tests artificially reduce entropy, creating an unreliable and coerced outcome, which may not accurately reflect the actual information content. The Selvi judgment acknowledges that forced tests may not yield truthful results, aligning with the idea that artificially reducing entropy may distort the quality of information.

4. Uncertainty and Legal Integrity

In information theory, artificially reducing entropy without consent can lead to distorted or noisy information. Similarly, in the legal context, extracting evidence through coercive means (e.g., narco-analysis) generates uncertainty in the reliability and admissibility of such evidence. The Selvi judgment reflects this by emphasizing the importance of voluntariness, ensuring that the information obtained is not the result of coercion, thus maintaining the legal system's integrity and avoiding the introduction of uncertain or unreliable evidence.

To apply the Shannon entropy formula:

in the context of Selvi v. State of Karnataka (2010)¹³, we can model the information extraction process during a Narco-Analysis test as follows:

Key Variables in Context:

• Random variable X: Represents the potential statements or pieces of information an accused might provide during interrogation, with or without a narco-analysis test. These statements could either be truthful, partial truths, or false information.
• p(x_i): The probability distribution of different outcomes (i.e., truthful or non-truthful information provided by the accused). For example:
• p(x₁) : Probability of obtaining truthful information.
• p(x₂): Probability of obtaining partial truth.
• p(x₃): Probability of obtaining false or unreliable information.

Applying Shannon Entropy:

Shannon entropy measures the uncertainty or unpredictability of outcomes (here, the information revealed by the accused).

1. Before Narco-Analysis:

Without any coercive testing, the information XXX that the accused holds is private and highly unpredictable. There is a high degree of uncertainty about whether the accused will provide truthful or false information.

High Entropy: The accused, protected by the right against self-incrimination under Article 20(3), may choose to stay silent or provide statements voluntarily, resulting in a high level of entropy because the state cannot predict what the accused will say.

The entropy in this scenario would be: BEFORE NARCO-ANALYSIS TEST

Since there is a high degree of uncertainty about the outcome. will be high.

2. During Narco-Analysis (Forced Test):

Narco-Analysis is intended to extract information by chemically inducing a relaxed state in the accused, making them more likely to reveal truthful or hidden information. However, the test does not guarantee truthful information and can lead to fabrications, misstatements, or unreliable results. Entropy in this scenario may decrease because the test attempts to force information out of the accused, theoretically lowering the unpredictability or uncertainty about obtaining some form of statement. However, due to the unreliable nature of the information extracted (since it could be influenced by the test's effects), the probabilities (partial truth) and (false information) may still be significant, preventing the complete reduction of entropy.

The entropy after the narco-analysis would be:

In this case, while (truthful information) might increase due to the test, the presence of uncertainty (false or partially reliable information) keeps the entropy from dropping to zero. The forced nature of the test also raises questions about reliability, which reflects the legal concerns emphasized in the Selvi judgment.

• Before narco-analysis: The entropy is high because the uncertainty is significant. There is a broad range of possible outcomes (truthful, partial truth, or false information) with high unpredictability.
• During/after narco-analysis: The test attempts to reduce entropy by forcing a reduction in the possible outcomes, but due to the unreliable nature of the test, this does not effectively reduce entropy to a meaningful or acceptable level for legal purposes.

Thus, may decrease slightly but remains non-zero, reflecting the uncertain and unreliable nature of the information obtained. The Selvi judgment recognizes this by deeming such tests unconstitutional, noting that they do not provide reliable reductions in uncertainty or valid information for legal purposes.

Intrusive yet Innovative Methodology

The use of statistical formulas like Shannon’s entropy in legal principles represents an intrusive method in the sense that it introduces quantitative, data-driven techniques into a field traditionally dominated by qualitative, interpretative reasoning. This shift challenges conventional legal paradigms by emphasizing objectivity and empirical analysis over subjective judgment. However, this intrusion is also innovative, as it opens up new possibilities for resolving legal ambiguities and enhancing the precision of legal decision-making. For example, in cases involving conflicting testimonies or complex forensic evidence, entropy-based analysis could provide a clearer, more systematic way to assess the probative value of such evidence.

IV. Limitation

The application of Shannon’s entropy principles to the legal domain, particularly in areas like the Indian Evidence Act or the admissibility of narcoanalysis tests, presents a fascinating yet inherently limited approach due to the fundamental differences between law and science. Law, as a discipline, deals with the complexities of human behavior, societal norms, and moral considerations, which are often ambiguous, context-dependent, and resistant to clear-cut quantification. Unlike scientific formulas, which thrive on predictability and exactitude, legal reasoning is inherently interpretive, relying on principles of justice, equity, and fairness rather than rigid mathematical frameworks.

Shannon’s entropy, while useful for quantifying uncertainty in information systems, may struggle to capture the nuanced and often subjective nature of legal decision-making. For instance, in the case of narcoanalysis, the test itself is controversial due to ethical concerns about voluntariness, the reliability of information obtained under altered states of consciousness, and the potential for misuse. Applying Shannon’s entropy to evaluate the consistency or informational value of narcoanalysis results might provide a statistical measure of reliability, but it cannot address the deeper ethical and legal questions surrounding the technique, such as the right against self-incrimination or the dignity of the individual.

Moreover, legal outcomes are not merely about the accuracy or consistency of information but also about the broader implications for justice, societal values, and human rights. Relying solely on Shannon’s principles risks reducing complex legal issues to a matter of statistical probabilities, overlooking the human and ethical dimensions that are central to the law. This approach could lead to an overemphasis on quantitative metrics at the expense of qualitative judgment, potentially undermining the flexibility and adaptability that are essential for addressing the unique circumstances of each case.

Furthermore, the legal system operates within a framework of precedent, policy, and evolving societal norms, which cannot be easily encapsulated by mathematical formulas. While Shannon’s entropy offers a novel perspective, its application must be tempered by an understanding of the limitations of scientific methods in addressing the inherently uncertain and value-laden nature of law. Ultimately, the integration of such principles should complement, rather than replace, the interpretive and humanistic aspects of legal reasoning, ensuring that the pursuit of scientific rigor does not overshadow the broader goals of justice and fairness.

V. Conclusion

The application of Shannon’s entropy formula to the legal domain offers a powerful conceptual framework to quantify epistemic uncertainty in judicial contexts. In the case of narco-analysis, this study reveals that rather than enhancing clarity, such techniques often amplify informational ambiguity, challenging the very foundations of evidence-based adjudication. The judicial system must exercise greater caution in accepting scientific evidence whose entropy levels remain high indicating unreliability and potential miscarriage of justice. As Indian jurisprudence evolves, embedding entropy-aware analysis can promote more rational, transparent, and constitutionally sound decision-making, particularly in the treatment of expert testimony and forensic interventions.

*****

Footnotes

1. Author is a student at St. Joseph’s College of Law, Bengaluru, India.

2. . Katz DM, Bommarito MJ. Measuring the complexity of the law (2014): the United States Code. Artif Intell Law. 22:337–74. doi: 10.1007/s10506-014-9160-8 ↩

3. C. E. Shannon. (1951). Prediction and Entropy of Printed English. Bell System Technical Journal 30, 1 ,50–64. DOI:http://dx.doi.org/10.1002/j.1538-7305. 1951.tb01366.x ↩

4. Ted Sichelman , ‘Quantifying Legal Entropy’ (2021) 9 Frontiers in Physics 1,4. ↩

5. D’Amato, A. “Legal Uncertainty”. (1983) California Law Review, 71(1), 1–55. https://doi.org/10.2307/3480139 ↩

6. Chipiga, Ilya. ”Legal Entropy in International Law” (2023). Russian Journal of Legal Studies (Moscow). 10. 113-120. 10.17816/RJLS322823. ↩

7. Sarkar, Soham and Singh, Shubham, Narco Analysis Test: Admissible OR Not? (October 5, 2018). Available at SSRN: https://ssrn.com/abstract=3687775 or http://dx.doi.org/10.2139/ssrn.3687775 ↩

8. C.E Shannon , “A mathematical Theory of Communication” (1948) , Vol. 27, pp. 379–423, 623–656 ↩

9. US 293 (1963) ↩

10. AIR 1954 SC 300 ↩

11. AIR 1961 SC 1808 ↩

12. Dinesh Dalmia v. State by Spe, CBI 2006 CriLJ 2401 ↩

13. AIR 2010 SC 340. ↩