Confidence After the Feedback
During a night-time street robbery, a 19-year-old witness sees the offender for about 12 seconds under street lighting. Two days later, police conduct a photo lineup of six faces. The administering officer knows the suspect and has already told the witness, 'We have the person, we just need you to confirm.' The witness first says, 'It could be number three, maybe four out of ten.' After the officer says, 'Good, that's the one we expected,' the witness says, 'Yes, number three, nine out of ten.' The suspect and witness are of different ethnicities. The defence has applied to exclude the identification and to call an expert on eyewitness memory. The prosecution argues the jury can use common sense.
1. Evaluate the identification procedure against established system-variable principles. What is the likely effect on the reliability of the identification?
Read the worked answer
The procedure departs from good practice on several system variables that police can control. The officer knows the suspect, so the lineup is not double-blind; the pre-lineup comment 'we have the person' is a biased instruction that implies the culprit is present. A proper instruction would say the offender may or may not be in the lineup and that the officer does not know. The post-identification feedback 'that's the one we expected' is a classic confirming feedback manipulation: it inflates confidence and can retrospectively distort the witness's recollection of how certain they were. The witness's initial 'four out of ten, maybe' is weak and should have been recorded verbatim before any feedback. The cross-race identification adds a well-documented accuracy cost, especially with brief exposure. Confidence after feedback is not a good guide to accuracy; the initial confidence and the procedure quality are more informative. The identification is therefore contaminated and should not be treated as a strong identification, regardless of the later 'nine out of ten'.
2. Should the expert evidence be admitted, and how should the jury be directed?
Read the worked answer
The prosecution's 'common sense' argument is weak because the relevant effects are counterintuitive. Jurors often believe that confidence predicts accuracy and that a confident witness is reliable; the research shows that post-feedback confidence can rise sharply while accuracy does not, and that cross-race identification is poorer. Expert testimony can assist the jury on those specific points without usurping its role. The expert should not say whether this witness is right; the expert should explain system variables, estimator variables, the confidence–accuracy relationship, and the effects of confirming feedback. If the judge excludes the expert, the jury should still be warned about the contaminated procedure: the biased instruction, the non-blind administration, the feedback, and the cross-race issue. The defence application to exclude the identification has merit because the prejudice from a confidently stated but contaminated identification may outweigh its probative value. If the identification is admitted, the direction should make clear that the later confidence rating cannot be used as an independent marker of reliability.
3. If the police wanted to obtain a fresh identification, what should they do, and what limits remain?
Read the worked answer
A fresh procedure should use a double-blind administrator, a lineup of fillers matched to the witness's description, a clear instruction that the offender may or may not be present, immediate recording of confidence in the witness's own words, and no feedback. A sequential procedure may reduce relative-judgement comparisons, though the evidence is mixed. The original eyewitness should not be re-shown the same suspect and treated as a clean identification: memory is reconstructive, and the earlier feedback has already contaminated the trace. A fresh witness, if one exists, would be preferable. Limits remain: the original procedure cannot be undone, the initial weak confidence is on record, and a later confident identification after knowing the suspect may reflect contamination rather than improved accuracy. Police should document the original procedure fully and avoid presenting the second attempt as independent corroboration. The fair inference is that the first identification is unreliable, not that a second attempt can launder it.