Abstract
When asked to generate any self-related information, subjects generally produce a higher recall score than when asked to recall experimenter-given information. This is known as the self-generation effect. To explain this effect, Itsukushima et al. (1992) proposed network theory of memory. In this theory, it is hypothesized that in the self-generation condition the subject can create necessay nodes and connect these nodes strongly to old and well established self-networks. The level of activation in the network is strong enough to make the spreading of activation. In the given-information condition, subjects might not fully create the necessary nodes, resulting in a weak connection to the self node. As a result, the level of activation itself might be weak.
To test this hypothesis two experiments were conducted. In experiment 1, subjects were asked to add adjectives to the objects in a different sentence completion tasks. In experiment 2, subjects were asked to add adjectives not only to the object but also to the subject (that is, the name). In both experiments the newly added modifiers in the self-generation condition showed higher recall results than those in the name-given condition. This suggests that new nodes which represent sentence elements such as subject modifiers and object modifiers are formed during their sentence completion. These results are interpreted by the sperading activation from the self node to the relevant nodes. In addition, the relationship between self and memory was discussed.
