[ BACK ] [ NEXT ]
[ Print View ] [ DEUTSCH ]

Information Processing by Humans and Machines

Computers have originally been developed to solve complex mathematical problems like differential equations or optimization tasks and (in a second step) to handle large amounts of data. Most humans have to go through a long and sometimes painful learning process to solve such problems. On the other side humans are rather good in "soft" tasks like recognizing a picture, a place, a smell, or a sound, finding "similar" situations or objects, summarizing a text or making a judgment (in particular with a weak factual basis). These tasks are difficult for computers and require a high amount of resources, delivering poor results if they are solved at all.

Strengths and Weaknesses of Humans and Computers.

A main reason for these differences is the way information is processed by humans and computers. While computers follow a sequential "bit by bit" type of handling data, human information processing seems to follow a more parallel type of processing, taking into account vague and redundant information. (The notion of redundancy is probably not even appropriate for biological processing, because there seems to be little pressure to "save resources" - in terms of brain capacity - for single "steps of calculation" and "bits to store".)

Thus computers are often very effective tools for humans if their specific strengths are required for example as calculators or text processing devices. In other cases the different ways to store and process information cause problems for the effective use of computers by humans: To retrieve a document with a specific content in an information system, a user has to specify her information need in such a way that the computer can "understand" it. In many cases this requires a lot of adaptation of the users to the system.

Models of Information Processing

In order to develop methods and programs, that are more adapted to the human type of information processing, it is often useful to study models of human knowledge representation and processing as well as alternative models of information processing with computers. Sometimes it is even useful to try to simulate single features or effects of human memory and cognition with computer programs.

In most cases this will not lead to a definitive explanation of human cognition, but it may add plausibility to some models and approaches and open the way to new ones. Such models can be used to support and simplify human-computer-interaction. During the last years I have investigated some aspects of these questions in a number of studies and projects.