The aim

1. The aim was to find out whether verbal labels change how an ambiguous figure is later reproduced.

What the participants did

1. 95 participants took part, split into List I (48), List II (38) and a control group (9).
2. There were 12 ambiguous figures, and each one had two possible names.
3. Each figure was introduced by a name before it was shown, with a different name for List I and List II, while the control group saw no name at all.
4. Everyone viewed all twelve figures and later reproduced them from memory.
5. Independent raters scored all 3051 reproductions on a five point scale running from almost perfect to almost completely changed.

What the most changed reproductions showed

1. Within the 905 reproductions at the most changed end of that scale, 74% of List I drawings and 73% of List II drawings resembled the name that group had been given.
2. Only 45% of control reproductions resembled either of the two names.

The conclusion

1. Naming a form before it is seen can determine the form that is later reproduced.
2. Recall is therefore reconstructed rather than replayed, which supports the top-down account of perception.

How strong is the evidence

1. A strength is the tight control, since the unnamed control group isolates the label as the cause of the difference.
2. A second strength is that 3051 reproductions is a very large number of observations, so the pattern is unlikely to rest on odd drawings.
3. A weakness is that the stimuli were artificial, since you are rarely asked to redraw a meaningless shape from memory.
4. A second weakness is that judging whether a drawing resembles a name is subjective, so two raters could score the same drawing differently.

Turning this into marks

1. Quote 74%, 73% and 45% with the restriction that they describe only the 905 most changed reproductions, not all 3051.
2. The mark most often lost is describing the procedure without ever judging the study.