next up previous
Next: A Process Up: A Discipline Independent Definition Information Previous: Definitions of Information

(pdf of full article)

Defining Information

A definition of information should capture the essential nature of the information phenomena in a precise description. While making explicit the similarities between information phenomena and other related concepts such as meaning, certainty, or knowledge, at the same time it should bring forward the differences between these concepts. For example, in what sense does information move through time or space? How is information similar to knowledge and how do they differ?

The nature of information may be brought forward through use of a metaphor or a model. For example, Shannon modeled information in communication processes as being transferred through a channel. Scholars and their students who have been trained in this model since the 1950s now consider this self-evident and an obvious base on which to build virtually any discussion about communication, although communication scholars have gone far beyond Shannon's initial model.

Information has been defined innumerable ways. As previously mentioned, Cookie Monster referred to it as ``news or facts about something," while others have defined information as the meaning of a signal and some understand it as the signal itself. It may prove useful here to examine the requirements of a satisfactory definition of information. Dretske suggests that we must ``preserve enough of our common understanding of information" if we are to maintain a link with the majority of the ideas about information present in our culture [Dre83]. Further, the definition should be as precise as possible, allowing non-informative phenomena to be clearly rejected. When the definition is applied in studies of information processing in humans, information may or may not have a person-specific aspect, and the quantity of information need not be the same for all sentient beings in a given situation; however, the amount of information should vary only as other factors, parameters of a definition, vary between the individuals. These external factors may be either personal or subjective and still allow for a precise definition and the useful measurement of information.

A good definition or theory of information both describes factually what occurs or what exists, as well as provides an explanation of events. In addition, it should bear some resemblance to the natural language notion of information but need not adhere to it when the natural language definition loses its generality and explanatory power. This happens when the common language definition of information, for example, becomes conflated with the notion of useful information, that is, information is understood to be in all cases useful. For those accepting this concept of information, if it isn't useful, it isn't information. Requiring that all information be useful limits the domain of discussions about information to cognitive processes that can "use" something; it excludes the information carried by a subatomic particle which is not sensed by a cognitive process. We try to avoid excluding information phenomena.

Information is often defined in terms of the human mind, although it is clear that very similar phenomena can be studied in lower level beings, such as communication and information transfer among ants [RR89]. Dretske notes that ``it is common among cognitive scientists to regard information as a creation of the mind, as something we conscious agents assign to, or impose on, otherwise meaningless events. Information, like beauty, is in the mind of the beholder" [Dre83]. This human centered view of information limits information to that perceived or produced by the human mind; it rules out information not perceived by a mind, such as physical events on a microscopic, non-discernible level, such as is studied by physicists. It also means that the concept ``information" is definitionally intertwined with the concept ``mind." There is a commonality to what physicists and epistemologists study, and this is captured by our broad use of the term information. Our purpose here is to capture this commonality--we do not wish to imply that particular disciplines or group of scholars, such as those interested in information collection, information organization and information use in humans, should not place limitations on what they choose to study, or to narrow the concept of information further, according to their specialized interests. We believe that it is helpful if this field-specific definition is consistent with the general definition except for limitations imposed by a field. The author believes that the general definition is as useful as the field specific definitions in all cases.

A clear statement of what is information and what is informative can lead to a strong qualitative understanding of the fundamental nature of information. A measure of one or more of the characteristics of informational events is inherently quantitative. Such a measurement either determines a characteristic value of the phenomena of interest or compares it to a second phenomena. Two different measurements taken of the same type of characteristic, for example, what might be obtained when measuring the amount of information transmitted by two different computer modems, may be compared to determine whether the information capacity of one channel is greater than, equal to, or less than the other. These relations compare data whose values can be ordered by using these relations. This allows statements such as

x is more informative than y
or
x is equally informative to y
to be made, based on the measured informativeness.

Given a number of possible definitions for a concept, the better definition is usually the simpler and often broader definition. This rule, referred to as Occam's razor [BEHW87], suggests that the simplest explanation has the highest ratio of signal to noise; that is, the highest ratio of helpful information to distracting or ``erroneous" information. Occam's law may be expanded to a range of environments, suggesting that the simplest law or definition that describes a phenomenon over the largest set of situations, the widest set of disciplines and theories, is the superior law. Consider a definition of heart disease that describes heart disease only in Caucasian males. A better definition would capture the essence of the disease in all homo sapiens. This more general definition might emphasize less those factors more likely to be experienced only by one subset of the population and is more likely to capture characteristics of homo sapiens in general. We do not want a definition of heart disease that is so broad as to be useless, such as a definition that covers all human processes and not just heart disease; a definition of heart disease is needed that describes the phenomena found in all species with various forms of hearts and circulatory systems, capturing the generalities that do exist in all heart disease. This develops theory as suggested by data. Obviously, medical doctors specializing in human cardiology need to understand the peculiarities of that species, while also understanding the broader processes taking place.

A similarly general but precise definition of information should exist at the core of information science. Imposing one's political and cultural interests on others in an academic discipline should be minimized--this imposition is clearly seen in those who define the field of information science as limited to the human use, organization, production, and retrieval of information, excluding other information phenomenon.

We note that individual and group biases and interests do affect what one knows and how one views the world. A strength of our definition of information is that it provides a place in a hierarchy of physical and mental processes for these biases to affect information in the form of biases acting as the input to mental processes. This acceptance of human biases as a cognitive reality is separate from our rejection of the imposition of individual biases on other scholars; we continue to believe that disciplines capture generalities and that information science should not reject any valid research on information and should attempt to embrace all forms of information.


next up previous
Next: A Process Up: A Discipline Independent Definition Information Previous: Definitions of Information
Bob Losee
1999-03-10