Investigations into the relationship between
memory and the
self originated in the field of
personality. Theorists held that an individual’s personality included something akin to a "directory" of
traits attached to the self, and that the way they interact with others uses this directory as a template for predicting their
behaviour. Traits can also be used to explain the past behaviour of oneself or others. Therefore, the "self" acts as an organisational agent for information in an individual’s world.
When asked to list traits describing themselves, most individuals will list positive ones first, such as intelligent, sensitive, friendly, etc. When reading a list of traits in another context, such as the diagnostic criteria for psychopathy, most readers will initially interpret these in relation to themselves, known as "medical student syndrome". Traits may be interpreted with the self as a type of superordinate schema.
The Self-Referential Encoding (SRE) effect holds that information relating to the self is preferentially encoded and organised above other types of information. In healthy individuals, this was first tested by Rogers et al (1977) who replicated Craik & Tulving’s (1975) classic depth-of-processing study. They asked participants to rate 40 descriptive adjectives on one of four tasks; Structural (Big font or small font?), Phonemic (Rhymes with xxx?), Semantic (Means same as xxx?), or Self-reference (Describes you?). This was then followed by an "incidental recall task". This is where participants are asked, without prior warning, to recall as many of the words they have seen as possible within a given time limit. Craik & Tulving’s original experiment showed that structural and phonemic tasks lead only to "shallow" encoding, while the semantic task lead to "deep" encoding and resulted in better recall. Rogers et al hypothesised that information with reference to the self would have even deeper encoding. They found a main effect for self-reference items to be recalled at least twice as well as semantic-encoded items. Thus, this phenomenon is sometimes called the "self-reference effect".
In 1982 one of the co-authors on the Rogers et al paper, Nicholas Kuiper, conducted a similar study comparing university students who were mildly depressed with those who were not (Kuiper et al 1982). A set of 60 adjectives were used, split into depressed words (e.g. bleak, dismal, guilty) and non-depressed words (e.g. amiable, curious, loyal) on the basis of a separate independent norming study. There were two tasks, the first being semantic (Does this word have a specific meaning or relate to a specific situation?) and the second self-referential (Describes you?). Four "buffer" items (two non-depressed, two depressed) were used at the start and end of each block but not analysed to avoid primacy and recency effects. Again, scores were transformed to control for biases towards items with "yes" responses. There was a significant main effect of the rating task (self-referential items more likely to be recalled than semantic items) as expected by the self-reference effect. Furthermore, non-depressed participants revealed enhanced recall of non-depressed words vs. depressed words, and mildly depressed participants had superior recall for both depressed and non-depressed words.
