Guest Post: Flawed Vision — Rethinking Eyewitness Identification in Indian Criminal Trials

(This is a guest post by Viraj Thakur)

How much weight should a 7-year-old memory carry in a murder trial? Courts in India have often answered: enough to convict.

On 29 July 2025, dealing with a case concerning the Godhra Riots of 2002, the Gujarat High Court (“HC”) in Sachinbhai v. State of Gujarat had the opportunity to elucidate on the law surrounding Test Identification Parades (“TIPs”) i.e. identification of suspects by eyewitnesses, outside of court, and Dock Identifications (“DIs”) i.e. the identification of suspects in court by eyewitnesses. While the HC, on facts, rightly refused to rely on identification in my opinion, it failed to engage with foundational issues: how time, stress, and suggestibility undermine visual memory, despite scientific consensus on these. This gap between evidence law and memory research poses a direct threat to fair trial rights for an accused, who may be convicted solely on the basis of a DI.

 

I argue that the effect of delays on eyewitness memory and the impact injuries/weapons have on accuracy of eyewitness identification are overlooked by courts, by referring to key precedents on TIPs. Consequently, I argue that courts must actively bear in mind that the standard to admit DI without a TIP ought to be very high. While courts do occasionally recognise this, a more well-reasoned rationale coupled with scientific evidence may make for a fairer process.

Introduction to TIPs and DIs

TIP is an eyewitness identification out of court made relevant under Section 7 of the Bharatiya Sakshya Adhiniyam (“BSA”). It is relevant for purposes of corroborating substantive evidence of eyewitness identification in court i.e. DIs (Daya Singh v. State of Haryana, ¶ 11). A TIP in particular is primarily meant to help investigators check if they are on the right track regarding the accused’s identity (Dana Yadav v. State of Bihar, ¶ 37). They are ordinarily held at the instance of the investigating officer (Dana Yadav, ¶ 5) and the accused has no right to ask for a TIP to be held, though courts may grant TIPs on an application if they believe it prudent to do so (ibid, ¶ 14). However, DIs have low probative value if uncorroborated, unless DI is of “sterling quality,” especially when the accused is a stranger to the witness (Sachinbhai, ¶ 44).

Reconsidering Time and Anxiety in Assessing Eyewitness Identification Reliability

Time elapsed between the crime and the identification and consequent memory decay

It is well-established that delays can affect the reliability of eyewitness identification. The delay may arise in holding the TIP or in the DI ultimately held in court. In Raja v. State, it was held that the impact of delays is a fact-based inquiry (¶ 14). However, if it can be demonstrated that a witness has an “enduring impress” of the identity of the suspect, then DI, even seven years after the crime, can be considered reliable (Daya Singh, ¶ 14). This “impress” often arises from sustained interaction or noticing distinctive features of the accused (¶ 23). However, if the testimony is unreliable or the suspect had changed their appearance, even a two-year delay would be fatal (¶ 25).

However, I argue that the theory of “enduring impress” overlooks the impact delays have on eyewitness memory.

 

Memory research shows that recollections fade and distort over time, unlike video recordings.[i] The greater the time between the crime and identification, greater the chance of false identification (ibid, pp. 23). As Gambell argues, repeated questioning, multiple recounts of the event, and multiple identifications, lead to changes in the memory of an eyewitness. Eyewitnesses may add details under the assumption that this “must have happened” (pp. 190-1). This is done by the eyewitness simply to make sense of the event and to avoid appearing as untrustworthy.

 

Consequently, given that verbal and pictorial information tend to be forgotten as time passes[ii] and that recollections of physical attributes (such as age, hair colour, complexion, height, etc.) are usually inaccurate (pp. 298-9), it is clear that delays have an adverse impact on the reliability of eyewitness identifications.

 

This becomes especially concerning given that with time, eyewitnesses grow more susceptible to suggestive cues – whether from police, environment, or internal bias – making delayed identifications less reliable (pp. 323-4).

 

This suggestiveness may be inherent in the process, such as in DIs, wherein only one person is shown to the eyewitness for identification. This is suggestive because the identity of the police suspect is inherently obvious, if there is only one “option” for an eyewitness to choose from (Perry v. New Hampshire Amicus Brief, pp. 24).

 

Suggestiveness may also arise due to the actions (subconscious or conscious) of the police, who may present the suspects differently from the fillers. For instance, in Tahir Mohammed v. State of MP, the suspects were put up with fetters connected by rods, while the fillers in the line-up were unfettered (¶ 8). Or, in case of TIPs, police personnel may indicate their approval / disapproval of a particular identification by the eyewitness through verbal or non-verbal cues (pp. 273).

 

Suggestiveness may even arise due to the psychology of the eyewitness – the desire to bring a criminal to justice, to exact vengeance upon them, or to find a scapegoat (pp. 367). It may arise due to the behaviour of the suspect themselves. Suspects are incentivised not to stand out and therefore may appear to be most anxious, standing further apart from the rest of the group, or receiving the most “looks” from the others in the lineup (pp. 319). Hence, the greater the delay, the more susceptible the eyewitness to suggestive cues.This further compromises the reliability of their identification.

 

Therefore, convicting a suspect based on a DI seven to eight years later, as in Daya Singh, is arguably incorrect. It would be a conviction based on unreliable evidence.

Stress & Anxiety due to Weapons and Injuries

The SC has held that eyewitnesses being injured would allow them an opportunity to observe a suspect, therefore overcoming problems posed by delays.[iii] In other words, being injured is a relevant factor in holding that a particular eyewitness had an “enduring impress” of the accused. In Pargan Singh v. State of Punjab, it was held that being in a near-death experience and sustaining injuries would mean seeing the suspect for merely 90 seconds would make for a sufficiently reliable DI, even with a delay of 7½ years (¶ 17). The SC specifically noted that “[t]he testimony of an injured witness requires a higher degree of credibility” (ibid, ¶ 19).

 

This goes against well-established evidence to the contrary. If a weapon is used to threaten / injure the eyewitness in committing a crime, most of the eyewitnesses’ attention is focused on the weapon and not the suspect (pp. 276). Stress and anxiety adversely affect the ability of a witness to recall, narrowing their attention – not the other way round (pp. 694). This becomes all the more concerning when “various factors that affect eyewitness accuracy often interact to compound the risk of mistaken identification” (Commonwealth of Pennsylvania v. Walker Amicus Brief, pp. 18). Therefore, in Pargan Singh and Daya Singh, delays coupled with injured eyewitnesses ought to render eyewitness identification, whether through TIPs or DIs, extremely unreliable – to the point, I suggest, that the results of such identification ought to be disregarded.

Conclusion

I have argued that Indian courts often overestimate the reliability of eyewitness identification. Long delays, and high-stress circumstances – especially where weapons or injuries are involved – distort memory. Scientific research suggests that such identifications are prone to error, yet courts continue to rely on them without adequate safeguards. Courts must adopt stricter evidentiary standards for visual identification – particularly DIs – or risk legitimising wrongful convictions based on flawed memory and implicit bias.

 

---

[i] Arne Trunkell, Reliability of Evidence (Beckman 1972) 21.

[ii] Jessica Lee, ‘No Exigency, No Consent: Protecting Innocent Suspects from the Consequences of Non-Exigent Show-Ups’ (2005) 36 Columbia Human Rights Law Review 755, 759-60.

[iii] See Dana Yadav [8]; Raja [22].