Index Last Week Contact DQ News Latest Issue
Creating False Memories
An article in the September issue of Scientific American explores how suggestion and imagination can create "memories" of events that did not actually occur. The article's author, Dr. Elizabeth Loftus, a professor of psychology at Washington University, describes the research she and colleagues at other universities have conducted into memory distortion. Beginning in the early 1970s, the author and her colleagues have conducted hundreds of experiments involving tens of thousands of individuals to document how exposure to misinformation induces memory distortion. According to the author, the studies show that when people who witness an event are later exposed to new and misleading information about it, their recollections often become distorted.
Moreover, the author and her colleagues in other laboratories have planted memories of traumatic childhood events that never happened (for example, being lost in a shopping mall, an overnight hospitalization, an embarrassing incident at a wedding reception) in adult research subjects. Research subjects may also be hypnotized, or be asked to participate in "imagination exercises" - imagining events that might have occurred without worrying about accuracy.
The author concludes that false memories are constructed by combining actual memories with the content of suggestions received from others. During the process, individuals may forget the source of information. The precise mechanisms by which false memories are implantred await further research.
Data Mining....Where's the Quality?
Three articles in the September/October issue of the magazine PC AI explore the field of "data mining." The first article, "Data Mining at Work," looks at the data mining products of five vendors of data mining systems. The products range from an automated flight control system validation to financial decsions. The unstated assumption in all the product descriptions is that the data being "mined" is of excellent quality.
The second article, "Data-Mining in the Financial Markets," reviews the author's experience with data mining in 30-year US Treasury Bond Futures. The author (Thomas Rathburn) describes how events and information affect price movements in the financial markets, how he is able to extract information from data, and how he uses the information to make profitable trades. Using neural networks and chaos theory as tools and never putting more than 5 % of capital at risk on a single trade, the author states his trading has made him "comfortably profitable." Like others who write such articles, the author is never specific about how he extracts relevant information from data.
In the third article "Data-Mining and Genetic Programming," author Andrew Colin describes a system for making intelligent real world decisions by combining inputs and outputs (data). The data items may algebraically combine into a rule to give an outcome. In order to produce candidate rules, a large number of sample inputs and outputs is necessary. The author also extensively discusses genetic algorithms, genetic programming, and program coding. There is no mention of data or information quality in the article. Information about the magazine PC AI may be obtained from the magazine's web site (http://www.pcai.com/pcai). The magazine is available at larger bookstores that sell magazines.