Self Reference

(Editor:   Jean Mandernach, Park University)


The self-reference effect is the tendency for individuals to have better memory for information that relates to oneself in comparison to material that has less personal relevance. The self-reference effect was first proposed by Rogers, Kuiper and Kirker (1977) in a study expanding the classic depth-of-processing work by Craik and Tulving (1975). In order to understand the mechanisms underlying self-referent encoding, it is important to first review the depth-of-processing model.

Craik and Tulving (1975) proposed the depth-of-processing model to explain variations in memory as a function of the extent to which information is actively processed. In this classic study, the researchers manipulated depth of processing by presenting participants with a series of words; each word was followed by a simple yes or no question. The yes/no questions varied in order to manipulate the level of processing (shallow, moderate, or deep). In shallow processing tasks, participants completed a structural analysis by identifying if the target word was capitalized. At the moderate processing level, participants analyzed phonemic properties to determine if target words rhymed. At the deep level of processing, participants used semantic analysis to determine if words fit within the context of a sentence. Following the task, participants received a surprise recognition test. In this test, participants had to identify which words they had seen in the previous activity (target words) and which words were new (distracter words). Results of the study indicated that participants' memories were more accurate for words that they had processed at a deeper level.

Theorizing that personally-relevant information would be encoded at a deeper level than semantic analysis, Rogers, Kuiper and Kirker (1977) expanded the depth-of-processing study to examine the impact of self-referent encoding. Their experimental design replicated Craik and Tulving's (1975) original study with the addition of a self-referent encoding manipulation that asked participants to judge whether or not target words were self-descriptive. As such, following the presentation of randomly selected words, participants responded to the question, "Does this describe you?" At the end of the manipulation phase, participants recalled as many words as possible. Results indicated that self-referent encoding produced better recall than structural or semantic encoding of the same information. They explained the advantages of self-referent encoding as a function of the more elaborative processes individuals used to relate target information to their own self-concept. The result is that information relating to one's self receives preference in memory and that people organize such information hierarchically above less personally relevant information, thus making self-referential information more salient in memory.


The current experiment is essentially a replication of Rogers, Kuiper, and Kirker's (1977) study on self-referent encoding. Participants view a common set of 20 words presented in a serial fashion; presentation order is random for each participant. Following the presentation of each word, participants answer one of two questions: "Does this word have an "e"?" (structural encoding) or "Does this word describe you?" (self-referent encoding). Half of the words appear, on a random basis, with each question. Note that half the words have an "e" in them and half the words are less socially desirable; as such, regardless of which word is presented with which question, a relatively equal number of yes and no responses is likely. After a short delay period, participants are given a surprise recognition test in which they are asked "Was this one of the words you saw earlier?" The recognition test list includes the 20 target words along with 20 distracter words; recognition items are serially presented in a random order.

The study involves a within-subjects design. The independent variable is the level of encoding (structural or self-referent); the dependent variable is the accuracy in identifying the target words. In this study, accuracy is determined by looking at the hit rates and false alarm rates for self-referent encoded and structural encoded words. To provide a composite measure of discrimination, the hit rate and false alarm rate are combined to create a single discrimination index known as A-prime (A').

Example Data Analysis

The most direct comparison of the accuracy of memory for self-referentially encoded words compared to structurally encoded words is a dependent measures t-test that compares the discrimination index for the two encoding conditions. The anticipated result is that the discrimination index will be significantly greater for words that were encoded via self-referent processes compared to the more shallow structural analysis.

In addition to an examination of hit rates and false alarm rates, one may want to analyze the time participants spend in the study. Study times should be examined as a possible confound as longer processing and answer times could be solely responsible for increased accuracy in discrimination.

Data Download and Format

Performance is measured via the discrimination indexes for self-referent and structural encoded words. Sample data appears below:

Sample data image from the Self Reference Data experiment
Figure 1

The first five columns provide demographic and classification data (participant ID number, class ID number, gender, age, and study completion date). The remaining columns indicate the hit rate, false alarm rate and discrimination index for self-referent encoded words and structural coded words as well as the total time spent to complete the study. The variable codes are as follows:

  • TIME - total study time
  • Self HR = Self-Reference Encoding Hit Rate
  • Self FAR = Self-Reference Encoding False Alarm Rate
  • Self DI = Self-Reference Encoding Discrimination Index
  • EWord HR = Structural Encoding Hit Rate
  • EWord = Structural Encoding False Alarm Rate
  • EWord DI = Structural Encoding Discrimination Index

Applications and Extensions

Although research has consistently demonstrated the self-reference effect in memory, there is not universal agreement on the underlying mechanisms that cause memory advantages for material related to oneself. Symons and Johnson (1997) conducted a meta-analysis on the self-reference effect in memory and concluded that the self-reference effect "appears to result primarily because the self is a well-developed and often under-used construct that promotes elaboration and organization of encoded information" (pg. 1).

Other research (Bower & Gilligan, 1979) challenges the importance of self-schema in the self-reference effect. Findings from their research indicated that relating information to any well-differentiated person (such as one's mother) will produce memory gains equivalent to those found in self-reference situations. As such, it may not be the reference to the self, but the relationship of new information to an existing, highly developed conceptual network that produces enhanced memory.


Bower, G. H., & Gilligan, S. G. (1979). Remembering information related to one's self. 
	Journal of Research in Personality, 13, 420-432.
Craik, F. I. M., & Tulving, E. (1975). Depth of processing and the retention of words 
	in episodic memory. Journal of Experimental Psychology: General, 104, 268-294.
Rogers, T. B., Kuiper, N. A., & Kirker, W. S. (1977). Self-reference and the encoding 
	of personal information. Journal of Personality and Social Psychology, 35, 677-688.
Symons, C. S., & Johnson, B. T. (1997). The self-reference effect in memory: A 
	meta-analysis. Psychological Bulletin, 121, 371-394.