Examples of serial reproductions of a drawing from memory showing that the schema of a face guides the construction of memory. In 1932, Frederick Bartlett [58] showed students a drawing of an African shield, and asked them to draw it from memory. He gave their drawings to others, asking them to reproduce the drawings from memory; these drawings were then reproduced from memory by a third group, and so on. The drawing bore the title ‘Portrait d’homme’, and reproductions of it gradually began to look more like a portrait of a man and less like an African shield, as shown above. The schema of a face suggested by the title guided people’s memories, illustrating Bartlett’s theory of constructive memory. This classic experiment has been recently repeated with the addition of mood [22]. Blind ratings of the drawings that resulted showed that those reproduced by individuals in happy moods were more face-like than those reproduced in sad moods. Positive affect thus seems to promote the use of accessible schemas, whereas negative affect inhibits their use, leading to more local, stimulus-bound processing. Reprinted with permission from Ref. [58].

Global–local perception experiment. (a) Respondents indicate, in each of 24 trials, the comparison picture (bottom) with which the target picture (top) goes. In this instance, if participants base their similarity judgment on global features, they would say that the comparison picture with the overall shape of a triangle is more similar to the target picture. If they base their similarity judgments on local features, they would choose the comparison picture in which the component elements are squares. Sample trial taken from Ref. [23]. (b) Results show a more global focus for happy and neutral than for sad mood groups. Data taken from Ref. [22].

False memory experiment. Thirty-six word lists, each comprising words highly associated with a non-presented lure, are presented. Relational processing of happy and control groups results in the false recall of many lures, whereas item-specific processing in sad moods was more accurate. Figure based on Ref. [44].