Gender schema theory is a cognitive theory that uses an information processing approach to explain how gender development occurs.  A schema is mental representation that helps simplify and categorize new information. According to Martin and Halverson (1981), there are two types of gender-related schemas. Superordinate schemas help children categorize objects, characteristics, and traits into basic male and female categories.  Own-sex schema are used to identify and learn information consistent with a child's own sex. 

Schema play a role in self-socialization - that is, children actively seek information about what gender means and how it applies to them.  This understanding of gender then motivates their own behaviour. 

Gender schemas affect which kinds of information are noticed and remembered, with information inconsistent with the existing schema being overlooked and consistent information being processed.  Researchers have found that schema may lead to memory distortion. Information that does not match one's gender schema is overlooked or changed to fit one's schema.

Gender schema theory argues that children are active information processors and gender schemas influence children's behavior.

The experiment carried out by Martin and Halverson (1983) investigated the role of gender schema on a child's ability to recall information that was not consistent with their gender schema.

The sample was made up of 48 children (24 males and 24 females) aged 5 - 6 years old. All were enrolled in local kindergartens.

Children were given a test (SERLI) to assess their level of gender stereotyping prior to the experiment.

Then, children were presented with 16 pictures, one at a time. The researchers showed the children pictures of males and females in activities that were either in line with gender role schemas - for example, a girl playing with a doll - or inconsistent with gender role schemas - for example, a girl playing with a toy gun. The children were asked to identify the sex of the person in the picture (man, woman, boy, or girl). They were not told that they would have to remember the images.

A week later, the children were asked to remember what they had seen in the pictures. The researchers carried out a probed recall procedure. They were asked about 24 pictures  - the 16 they had seen and 8 that they had not.  The 8 "unseen images" were included to test for response bias. Children were asked, "Do you remember seeing a picture of something doing (activity) in the pictures I showed you last week?" If they were then asked if the person they remember seeing was a girl, a boy, a man, a woman, or "don't remember." They were also asked to rate their level of confidence from on a four-point scale.

For pictures with female actors, those activities consistent with gender stereotypes were more often remembered than inconsistent activities, whereas for pictures with male actors, those activities inconsistent with the stereotype were remembered better. This may indicate that male stereotyping as to what is "allowed" for males is more defined and rigid than for females in this population.

Regardless of their level of stereotyping, children had distorted memories of pictures that were not consistent with gender role schemas - they remembered the picture of a girl holding a hammer as a picture of a boy holding a hammer. Children were more confident and demonstrated less distortion of memory when the stories were consistent with gender schema.

This supports the theory that stereotypes affect both the encoding and retrieval of information. When they were asked to identify the sex of the person in the picture in the first phase of the study, children sometimes made errors that made the sex consistent with gender stereotypes.  This is evidence of distortion on encoding.  They also made errors in the recall or retrieval of the information.

The study is highly standardized and can be replicated to determine its level of reliability.

The study controlled for response bias. First, they asked them about images that were not seen in the initial part of the study. In addition, they asked them their level of confidence.  Those that claimed to remember images that were not seen tended to have very low levels of confidence.

The researchers avoided a "forced choice" response of "boy" or "girl" and gave the children five choices to choose from: man, woman, boy, girl, "I don't know."

The task has low ecological validity as the task is very artificial and the study is highly controlled.  The study may not reflect how children process information about gender in the real world.

The scores on the SERLI were not correlated with memory distortion. This then leads us to question how schema were operationalized in this study.

Finally, there is an assumption made that information is actively pursued by children and that their behaviour develops as a result of schema development.  But as this study was cross-sectional and not a longitudinal study of the children's behaviour, this study does not provide empirical support for that assumption.