Cognitive Approach studies

Aim (Cox and Griggs)

To see if matching bias was less commonly used to solve the Wason Selection Task when the task was personally relevant

Procedure (Cox and Griggs)

Participants were given a workbook with three problems, and the order was different across six randomly allocated groups to counterbalance order effects. For each task they were asked what cards would need to be turned over to prove the statement true. The abstract statement was “if a card has an A on one side, then it has a 3 on the other side”, with one card showing A, one showing B, one showing 2, and one showing 3. The intermediate statement was “if a person is wearing blue, then the person must be over 19 years old”, with the cards showing wearing blue, wearing green, 22 years of age, and 16 years of age. The memory cueing statement was “if a person is drinking beer, then that person must be over 18 years old”, with the cards saying drinking beer, drinking coke, 22 years of age, and 16 years of age.

Results (Cox and Griggs)

5/144 (3%) solved the abstract task correctly, 62/144 (43%) solved the intermediate task correctly, and 87/144 (60%) solved the memory cueing task correctly. Researchers also found that if the abstract task was first, participants were more likely to show matching bias in the following tasks. Therefore, when the task cued memories of a past experience, a more logical approach was taken, but if the task was abstract, heuristics tended to be used.

Evaluation (Cox and Griggs)

Strengths:

• Study has been replicated (results are reliable)

• Repeated measures design controls for participant variability

• The study is counterbalanced to control for order effects

Limitations:

• Highly artificial task (low ecological validity)

• Some factors are not accounted for, such as the importance of the decision or the role of others in decision making

• Sampling bias (study was done only on undergraduate psychology students; not representative of the entire population)

Aim (Englich and Mussweiler)

To determine whether the request for a certain length of a prison sentence would influence the decision made by a judge

Procedure (Englich and Mussweiler)

To control level of experience, the sample consisted of 19 young trial judges with an average of 9 months of experience. The participants were presented with an alleged case of rape, with the prosecutor in the high anchor condition recommending a 34 month sentence, and the prosecutor in the low anchor condition recommending a 2 month sentence. The case materials were developed with advice from highly experienced trial judges, and a pilot study was conducted on senior law students to determine a prison term of 17 months as a basis for the anchors. The participants were given the case materials and the penal code, and were given 15 minutes to read the materials and form their opinion. They were then told what the recommended sentence for their condition was and given a questionnaire. The questionnaire asked “do you think the sentence was too low, adequate, or too high?”, “what sentence would you recommend?”, “how certain are you about your sentencing decision?” (a scale of 1 – 9), and “how realistic do you think this case is?” (a scale of 1 – 9).

Results (Englich and Mussweiler)

The low anchor group suggested an average sentence of 19 months, with a standard deviation of 9, and the high anchor group suggested an average sentence of 29 months, with a standard deviation of 6.5. The average certainty of the participants was 4.5, while the average certainty of the pilot group was 7. It appears as though anchoring bias did influence the decisions of the judges.

Evaluation (Englich and Mussweiler)

Strengths:

• Cause and effect between anchor and sentence can be established

• The pilot group allowed reasonable anchors to be established and demonstrated system 2 thinking, as opposed to the system 1 thinking used in the experimental group

Limitations:

• Independent samples means participant variability could have impacted the results

• Small sample size (19) makes it difficult to generalize findings

• Sample was only young, inexperienced judges, so the findings are best generalized to this small group

Aim (Loftus and Palmer)

To investigate whether the use of leading questions would affect estimations of speed

Procedure (Loftus and Palmer)

45 students were divided into five groups of nine. Seven films of traffic accidents taken from driver’s ed films, ranging from 5 to 30 seconds in length, were shown. After watching a film, participants were asked to give an account of the accident and fill out a questionnaire, with the most important questions being to estimate the speed of the cars in the accident. This question was the same for all groups except for the critical word; one group got the question “about how fast were the cars going when they hit each other?”, and the others had the same question but hit was replaced by collided, bumped, smashed, or contacted.

Results (Loftus and Palmer)

The mean estimates of speed were highest for the smashed group (40.8 mph) and lowest for the contacted group (31.8 mph). A statistical test showed that the results were significant (p ≤ 0.005). The results indicate that the critical word impacted the participants’ responses. Researchers argued that response bias could have caused this; if a participant was unsure of the speed, a strong verb like smashed would bias their estimate. It is also possible that the verb changed the participants’ perception of the accident by activating schemas. This supports the theory of reconstructive memory, in that it is not the actual details that are remembered, but rather what fits with the schema.

* In a follow-up experiment using just the verbs hit and smashed after being shown a video of an accident, those who received the question with smashed were more likely to report seeing broken glass in the video (16/50 vs 7/50; 6/50 for the control that didn’t estimate speed) when filling out a questionnaire a week later, even though there was none. This demonstrates that schema processing was used to inform their perception of the accident.

Evaluation (Loftus and Palmer)

Strengths:

• Confounding variables are controlled

• Cause and effect can be established

Limitations:

• Low ecological validity (films were for teaching purposes and don’t reflect the emotional state of witnessing an accident)

• Students are not representative of the entire population (sampling bias; students tend to be inexperienced drivers)

• It is difficult to estimate the speed of a car

Aim (Yuille and Cutshall)

To determine whether leading questions would affect the memory of eyewitnesses at a real crime scene (the same as Loftus and Palmer but with a real event)

Procedure (Yuille and Cutshall)

21 eyewitnesses were interviewed by the police for their account of an armed robbery in which the store owner was shot (this incident was chosen because there were many eyewitnesses and forensic evidence could confirm the eyewitness accounts). 13 of the eyewitnesses agreed to be in the study four months afterwards, in which they gave their account of the incident and were asked questions. Two leading questions were asked; half the participants were asked if they saw “a” broken headlight on the getaway car while the other half was asked if they saw “the” broken headlight, and the same thing was done when asking about a yellow panel on the car (there was no broken headlight and the panel was blue). Participants were also asked to rate the level of stress during the incident from 1-7.

Results (Yuille and Cutshall)

The accuracy of the eyewitness compared to police reports was 79-84%, and they did not make errors because of the leading questions (10/13 said no to the leading questions). This contradicts the results of Loftus and Palmer*, and could be explained by the presence of an emotional response, since, although the witnesses didn’t report feeling afraid, they reported an adrenaline rush.

*A higher percentage of people remembered false information in this study than Loftus and Palmer, which means that leading questions may still impact reconstructive memory

Evaluation (Yuille and Cutshall)

Strengths:

• The accuracy of memories could be verified

• Consent was given by all participants, so recalling the incident would not cause undue stress or harm

• Naturalistic setting (high ecological validity)

Limitations:

• The study is not replicable or generalizable (it is a specific instance)

• Variables were not controlled (difficult to know the level of rehearsal; the participants could have agreed to be in the study because they had spent the most time talking about the case)

• This could be a case of flashbulb memory because the participants’ safety was threatened, which would mean it can’t be compared to Loftus and Palmer

• Attempted deception (ethical concerns)

• Qualitative responses could be open to researcher bias

• Very small sample size

Aim (Loftus and Pickrell)

To determine if false memories of autobiographical events can be created through the power of suggestion

Procedure (Loftus and Pickrell)

Before the study, a parent or sibling of each participant was asked for three childhood memories of the participant and if they were ever lost in a mall. The participants then received a questionnaire by mail asking them to describe four memories: the three real ones and getting lost in the mall. They were told that if they didn’t remember the event they should write “I do not remember this”. The participants were interviewed twice over four weeks, in which they were asked to recall as much information about the events as they could and to rate their confidence in their memories from 1-10. They were debriefed after the second interview and asked to identify what the false memory was.

Results (Loftus and Pickrell)

About 25% of participants “remembered” getting lost in the mall, but they also ranked this memory as lower confidence than the others and wrote less about it in the questionnaire.

Evaluation (Loftus and Pickrell)

Strengths:

• Memories could be verified through parents or siblings

• High ecological validity (talking about memories is a realistic situation)

• Findings have been applied to eyewitness testimony and therapy

Limitations:

• Doesn’t explain why some people were more susceptible to suggestion than others

• You can’t determine whether the new memory is truly a false memory or if it is just a distortion of a real memory of getting lost

• Ethical concerns around deception and the potential stress of having a false memory implanted

• The participants could have spoken with someone when filling out the questionnaire at home (contamination)

• Potential demand characteristics (social desirability)

Aim (Tversky and Kahneman)

To test the influence of positive and negative frames on decision-making

Procedure (Tversky and Kahneman)

Participants were asked to make a decision between two hypothetical responses to the outbreak of a contagious disease. For participants in the first condition, the choices were framed positively: if program A is chosen then 200 people will be saved, if program B is chosen then there is a ⅓ probability that 600 are saved and a ⅔ probability that no people will be saved. The options were presented with a negative frame in condition 2: If program C is adopted 400 people will die, and if program D is chosen there is a ⅓ probability that nobody will die and a ⅔ probability that 600 people will die. Note that program A is the same as program C and program B is the same as program D, just framed differently.

Results (Tversky and Kahneman)

In the positive framing condition, 72% of participants chose program A, and in the negative framing condition, 78% of participants chose program D. When the options were framed positively, more people chose the certain outcome, but when the options were framed negatively, more people chose the risky option. This demonstrates the influence of the framing effect.

Evaluation (Tversky and Kahneman)

Strengths:

• High internal validity

• Cause and effect can be established

• Highly standardized (replicable)

• Reliable results

• The framing effect has many applications such as marketing and health campaigns

Limitations:

• Low mundane realism (hypothetical situation; in a real situation like this there would be more emotion involved and many people would be making the decision together)

• Sampling bias (Western university students)