Eyewitness testimony has been a significant discussion in forensic psychology for the past decades. Being used to instantiate ideals of justice, eyewitness testimony forms the bedrock of numerous judicial processes all over the world. This shows a need to be explicit about exactly what is meant by the term eyewitness testimony. It refers to the verbal account that a person gives about what he or she witnessed during an event that is being investigated (Baddeley 1999). Juries identify the accounts given by witnesses as authoritative and exceptional and its application has consecutively determined the conviction of criminals. However, over the years researchers have provided a great deal of empirical evidence to suggest that there are a number of limitations in the testimony of witnesses that must be taken into consideration. This paper will present evidence to highlight the strengths and the weaknesses of employing eyewitness testimony in courts. .
Reconstructive memory is a fundamental theory that pursues to demonstrate how memory may be distorted by prior experiences of an individual. Barlett (1932 cited in Gross 2010) introduced this approach to explain that people interpret and reconstruct events by using schemas - ''the generalized mental representations of everything that people understand by a given type of object or event based on past experience'' (Gross 2010). People use schemas in order to make a story, with unfamiliar words and ideas, more coherent for them. Schemas make the life of people more predictable without a need to store similar information for many times. After carrying out an experiment to illustrate the influence of schemas, Barlett established that reconstructive errors and schemas are psychological factors that might affect the accuracy of eyewitness testimony in courts. However, the theory of Barlett has been challenged by a number of analysts. Gauld and Stephenson (1967) for example, point out that if the participants were aware of the importance of their recall, then the number of their errors would be considerably reduced.