9.2.1, 9.2.2, 9.2.3 - Comparing classic and unseen studies

Manipulating social and cognitive conditions

Sherif and Baddeley both changed conditions to test an explanation, but they studied very different kinds of behaviour.

Sherif et al. (1954/1961) investigated how relations between groups develop. At the Robbers Cave camp, 22 selected boys aged about eleven were divided into two groups. The participants were from broadly similar backgrounds and were initially strangers. Staff acted as researchers within a functioning summer camp. First the separate groups developed their own activities and identities. Next, competitions with prizes created incompatible group goals: one group's success meant the other's failure. Hostility included insults and destructive actions. Finally, the researchers introduced superordinate goals, such as restoring a disrupted water supply, which required both groups' cooperation. Repeated shared goals reduced friction; mere pleasant contact had not been sufficient. This supported realistic conflict theory's claim that competition and interdependence shape group relations. Original report, Chapters 2–3 and 5–8.

The camp gave the boys sustained, meaningful relationships rather than a brief questionnaire about imaginary groups. That helps explain why the behaviour is relevant to real group life. Nevertheless, staff arranged events, groups passed through stages in a fixed order and the sample was narrow. Later cooperation could partly reflect time together or staff influence, so the evidence does not isolate every cause as neatly as a randomly assigned comparison of competition and cooperation would. It also does not show that all social prejudice is caused by material competition.

Baddeley (1966b) investigated coding in memory for word sequences. In the visually presented third experiment described in Pearson's study support, 72 adults from a participant panel were allocated to four list conditions: acoustically similar, acoustically dissimilar, semantically similar or semantically dissimilar words. Each list contained ten words. Participants learned the order across four trials, with digit interference intended to reduce short-term rehearsal, followed by a delayed retest. Semantic similarity impaired later performance relative to semantic dissimilarity; the acoustic pattern differed. The evidence supports an important role for meaning in long-term coding, not exclusive semantic coding or a direct test of the entire working memory model. Pearson study support, printed pp4–6.

Here the changed list characteristic is the independent variable and sequence-recall performance is the dependent variable. Standard presentation and scoring make the comparison precise. Yet remembering the order of laboratory words is narrower than remembering a personally meaningful event. The interference procedure also matters: a delayed test alone does not guarantee that only long-term memory is involved. Compared with Sherif, Baddeley gained control over the task while sacrificing much of the social and everyday context.

Conditioning evidence and brain differences

Watson and Rayner changed an infant's experience, whereas Raine compared people whose relevant group membership already existed.

Watson and Rayner (1920) asked whether fear could be conditioned, generalised and retained. Their single infant participant, Albert, initially showed no fear of the white rat. At just over eleven months, they paired the rat with the noise of a struck steel bar. Across seven pairings he came to withdraw and cry when the rat appeared alone. Early transfer tests showed fear towards a rabbit and other furry stimuli, but reactions varied; some later animals were also directly paired with noise. Some reactions remained at the final test about a month later. Planned removal of the fear did not occur because Albert left the hospital. Original report.

The change from reaching towards the rat to avoiding it is compatible with classical conditioning: a previously neutral stimulus became a signal for an aversive event. A detailed record reveals change within one person. However, without an equivalent unconditioned comparison participant, sensitisation to frightening experiences and other changes cannot be fully separated from the learned association. Variable procedures complicate generalisation claims. One infant also cannot establish how common the effect is. Causing distress raises a further ethical objection, independently of whether the account is informative.

Raine et al. (1997) investigated brain functioning in 41 people charged with murder or manslaughter who pleaded not guilty by reason of insanity, compared with 41 controls matched on age and sex, with relevant clinical matching. Participants performed a continuous performance task while a glucose tracer allowed PET measurement of metabolic activity. The NGRI group showed reduced prefrontal activity and differences in several other regions, including the corpus callosum and subcortical asymmetries. The authors interpreted the pattern as consistent with biological vulnerability to violence, requiring other influences as well. Original abstract, Pearson support, printed pp12–13.

Raine's numerical, standardised scan measures avoid asking people to judge their own brain activity. Matching reduces some alternative group explanations. It does not randomly allocate a brain difference or a history of violence: life experience, illness or other factors may influence both. PET measures activity during a task, not a person's inevitable future behaviour. The selected NGRI group cannot represent all aggression. Thus the scan study supplies broader group evidence than Albert's case, but weaker evidence for the direction of a cause than an ethically controlled manipulation could supply.

Diagnosis and institutional context

Rosenhan studied how a setting and an initial label could affect interpretation of later behaviour.