A2 Psychology at Spalding Grammar School

Cognitive psychologists have done a lost of research into how we recognize faces. Elizabeth Loftus has shown again and again how people’s memories are reconstructive – we “invent” our memories from SCHEMAS (stereotyped ideas) and we are very influenced by leading questions. Even the difference between asking a witness to start by remembering the eyes or the ears could have a huge effect on the face they come up with.

One of the great breakthroughs in Forensic Psychology has been the FACIAL COMPOSITE technique. This involves helping witnesses put together the image of a face from the different components a face is made from – choosing the right nose from a selection of noses, the right eyebrows from a selection of eyebrows, etc. The first successful facial composite was IDENTIKIT, developed in the 1940s by the Los Angeles police. This was developed in England into PHOTOFIT, which used more realistic photographs rather than drawings. The latest computer packages, like E-FIT, help witnesses pick facial features from huge databases of faces.

Give it a go yourself. Download Ultimate Flash Face, a free software that lets you create facial composites. 

This approach comes from the Cognitive perspective. Psychologists from the GESTALT school of thought argue that this isn’t how the mind works at all – we recognize or recall things as a whole, which we then bring into focus. The Cognitive approach is rather reductionist and the Gestalt approach is more holistic. The Gestalt approach prefers the use of police artists to help witnesses and there is some evidence (eg. Davies et al, 1978) that artists’ impressions are more accurate than photofit composites. 

Prof. Vicki Bruce, now at Newcastle University, has done lots of research into how we recognize faces. She suggests (1999) that we have two different sets of cues: INTERNAL FEATURES (like the eyes, nose and mouth) and EXTERNAL FEATURES (like the shape of the head, ears and hair). She suggests we recognize familiar faces by their internal features, but external features are more important for recognizing unfamiliar faces (like a stranger who robbed you).

Prof. Bruce’s lab experiment used 30 staff and students from Stirling University (equal mix of sexes, average age 29) being paid £2 for a simple sorting exercise. The researchers had taken photographs of 10 celebrities and used these to create a set of 40 facial composites using E-FIT, PRO-Fit, EvoFIT and other facial composite software programs. The IV was the type of composite shown to the participants: some showed only external features, some only internal features and some were complete composites. Participants were asked to match the composite to the original photo and the DV was the number of successful matches. On average, participants had a 35% success rate using full- or external- composites, but only 19½% correct with composites using internal features.

In a second lab experiment, Prof. Bruce recruited 48 volunteer students (21 males, 27 females) to do another sorting exercise. As before, the participants were presented with a photo line-up of celebrities and had to match either external- or internal- composites to the right photograph. This time there was a second IV: the celebrities in the lineup either looked similar to each other (the difficult condition) or looked quite unlike each other (easy). The average success rate was 42% for the external features, but only 24% for the internal features. This was consistent in both the easy and difficult conditions.

Interestingly, participants performed only slightly above chance when using internal features. They performed much better with external features; in fact, external features turn out to be as helpful as whole faces. Given that the celebrity faces should have been familiar, this casts doubt on previous research.

A big debate in Forensic Psychology is whether we actually recall faces by recalling each component part then mentally putting them all together (the Cognitive approach which lies behind facial composite techniques), or whether we recall faces as a whole then bring them into focus (the Gesalt approach, closer to what police artists do). Prof. Bruce’s research suggests that something about internal features makes them very unhelpful when trying to put together a facial composite – we prefer the bigger features of head-shape, hair-style, etc. Some new computer programs take this into account, using FACIAL MORPHING [see right] to blend faces together to create a sort or bland, generic “anyface” that can be a great start for a witness to work from.

The ability to recognize faces is affected by the same sort of variables that affect recall generally, for example:

• Delay: how long ago did you see the face you are trying to remember?
• Exposure Mode: did you see the face in a very different context from the one you’re trying to recall it in?
• Stimulus Variables: how meaningful to you is the face you are trying to recall?
• Subject Variables: how good is your memory and how do you personally recall facial details?

An interesting Subject Variable is “CULTURE”.  A study by Rachael Jack et al. (2009) shows Asian people focus on the eyes, whereas Westerners scan the whole face. Look at the differences between how Westerners and Easterners show “smileys” and other emoticons;

A particular feature of Exposure Mode is “AROUSAL” – was the event you are trying to recall exciting or frightening? Research into arousal tells us that when experiences are very arousing, our ability to do things gets worse. Very arousing circumstances can be difficult to recall accurately – and being a victim of crime is often very arousing.

Elizabeth Loftus and her then-husband Geoffrey Loftus carried out a lab experiment (1987) to show how, during a crime, a witness’ attention might focus on the weapon being used at the expense of focusing on the offender’s face. This was a lab experiment involving 36 university students (aged 18-31). Half the students were recruited through advertising (self-selecting sample) and they were paid $3.50; the other half were classroom students who volunteered in exchange for extra course credits. The students knew this was a memory experiment, but thought it would be a study into leading questions.

The students watched a slide show of 18 slides showing a scene unfold in a fast-food restaurant. For the control group, the slides showed the second customer paying by cheque; for the experimental group, the same customer pulled a gun on the cashier. Apart from this, both slide shows were identical and every slide was shown for 1½ seconds. During the slide show, students’ eye movements were monitored (using a high-tech version of the EOG that Dement & Kleitman used to monitor eye movements of dreamers). Students completed a 20-item multiple choice questionnaire and then had to pick the second customer out of a photo line-up of 12 head-and-shoulder photographs. Finally, they were asked to rate how confident they were about their line-up selection, from 1 (guesswork) to 6 (total certainty).

There was no significant difference in answers to the questionnaire but the line-up produced different results. The chance of picking the correct photo out randomly was 8½%. In the control group, 38.9% chose the correct photo, but in the experimental condition only 11.1% got it right. Interestingly, there was no difference in the confidence levels. Eye fixation data showed more focus on the gun than the cheque (significant at p<0.025).

It’s clear that the Weapon Focus effect exists, but it’s not clear what causes it. Loftus suggests that the danger of the weapon is arousing, so we pay more attention to it and neglect to encode details about the bearer’s face. However, a study by Kerri Pickel (1999) suggests that it is the unusualness, rather than the danger, that is distracting. She showed students short videos about a man with a gun and they recalled details better if the setting was appropriate for guns (a rifle range, rather than a baseball game) or if the man was appropriate for guns (a cop, rather than a priest). It might be that schemas are at work, and we remember things better that fit in with out schemas.

Read this interesting article about a case of mistaken identity and the psychology behind it.

Evidence suggests standard police interview procedures are very ineffective: questions may be inappropriate, poorly worded or improperly sequenced, while witnesses are often interrupted. Edward Geiselman [right] & Ron Fisher developed the COGNITIVE INTERVIEW (CI) based on psychological theories of memory.  These theories are:

• Encoding Specificity Hypothesis: we recall better if we are in the same place, same emotional state or same context as when the memory was encoded (Tulving & Thomson, 1973)
• Hierarchical Network Model: memories form a network, so there may be several cues for retrieving a memory – if one doesn’t succeed, try a different cue (Collins & Quillian, 1969)
• Schema/Scripts: We have schemas for appearances and scripts for social behaviour and we tend to “fill in the gaps” of memories to match an appropriate schema; we also filter out (forget) information that is inconsistent with our schemas (Bartlett, 1932)

CI uses the following procedure:

1. Context Reinstatement: The witness is asked to recreate mentally the context of the incident, imagine themselves back at the scene along with any emotional responses. These feelings are as important for recall as remembering physical details about the place the event occurred in.
2. Focused retrieval: Since memories form a network, the witness must recall everything, even trivial details. Incomplete details from different witnesses may mesh together to provide important leads.
3. Extensive retrieval: Witnesses put themselves in the place of other victims/witnesses; this change of perspective reduces the effects of schemas in producing false recall. This is difficult for children, though, and may not be accepted as evidence in court.
4. Change Order: Recalling events in an unusual order (starting in the middle, working backwards, etc.) may help witnesses recall things that don’t fit in with schemas.  This technique can disrupt context reinstatement, however.

In 1991 two men murdered another person working in a Miami office building and the only witness was a woman who had passed the two attackers in the lobby. Her memory of the men was very sketchy, so the police called in Ron Fisher to use his CI techniques. Fisher helped the witness recall one man brushing his hair from his face and the recollection of a silver earring he was wearing. This proved to be a breakthrough in the case and CI is used today in many police forces across the world. It has been developed for use with children and in social work and psychiatry.

Fisher & Geiselman (1989) tested the CI in the field, with the help of the Miami police department. 16 experienced officers from the Robbery Division assisted. The researchers recorded these detectives carrying out their standard interviews for 4 months (88 interviews), then split them into two groups: a control group and a group who were trained in CI techniques. The training took 4 1-hour sessions and 7 detectives completed training. Over the next 7 months more interviews were recorded for both groups. The recordings were analysed by a team at the University of California who were “blind” to the conditions (whether the detective was CI-trained or not).

The CI-trained detectives helped witnesses recall 47% more information than before their training (repeated measures) and 63% more than the untrained detectives (independent measures). The interviews were tested for accuracy by checking them against another witness; of the 24 cases with this sort of corroborating evidence, 94% of the recalled information was corroborated. The CIs took longer to conduct than standard interviews, but not significantly so.

In fact, the CI used by Fisher with the Miami police was the ENHANCED COGNITIVE INTERVIEW. The enhanced-CI downplays the use of different perspectives and changing order in favour of listening skills and building rapport with the witness.

Some critics have suggested that the CI’s main benefit is just that it encourages detectives to approach interviews in a structured way, whereas traditional police interviewing was notoriously haphazard and unreliable. In other words, even if the CI does produce greater recall, that might not be due to the psychological techniques so much as the fact that witnesses are questioned in an organized and thorough way. Critics have attacked the enhanced-CI for ditching the real psychological strengths of the technique (changing order and perspective) in favour of common-sense techniques like listening skills and setting the witness at ease.