Hundreds of explicit and implicit definitions of communication have been published in the communication and related literatures for use by scholars and practitioners trying to describe, predict, and understand communicative phenomena. These definitions vary around the common language definitions, with variations depending on individual scholarly interests and general scholarly trends. The diverse definitions of communication offered by Hauser [Hau96, p. 7] serve as a representative, albeit small, sample of ideas about communication from a wide range of disciplines. Of seven definitions provided by Hauser, three definitions of communication place communication in the context of humans or organisms, while a majority mention the effect of a message on its recipient.
Edward Wilson notes that ``the ongoing fragmentation of knowledge and resulting chaos in philosophy are not reflections of the real world but artifacts of scholarship" [Wil98, p. 8]. The definition of communication developed below is both rigorous and general in capturing all and only the communication phenomena in the ``real world." We are not trying to build on traditional definitions of communication and our definition isn't ``an artifact of scholarship"; instead, we build a model of communications on both a precise definition of information and on a list of required characteristics for a definition of communication. Thus, there is no explicit preference for which side should win in long-running intra-disciplinary debates, such as whether there is intrapersonal communication, for example. We assume that communication has the following characteristics (derived in part from questions posed by Motley [Mot90]):
We choose not to place an emphasis on communication as symbolic in nature. Motley [Mot90] emphasizes this symbolic nature, mentioning that ``Cronkhite (1986) presents a very compelling case for symbolic behavior as the common denominator of all communication study." However, as Hauser [Hau96, p. 507] notes, the symbolic approach may not be the best in all situations, although it is certainly a reasonable assumption for some models of communication. We also choose not to focus on intentionality, motivation, or the behavior of the sender or receiver. Instead, we adopt an approach that, with the precise model of information developed above, provides us with a more physical, observable, and precise set of requirements for a definition of communication.
We may define communication in a precise and information-based manner using the characteristics above, allowing us to understand the mechanisms underlying communication.
Communication occurs if, and only if, information moves from the input to one process to the output from a second process, the latter process being the inverse of the first process.We refer to the information at the output of this inverse, receiving, process, as a communication. Communication is more complex than information; communication processes are composed of multiple complementary informative processes.
Here we have two informative processes, the second of which ``undoes" what the first process ``does." Viewed loosely, hearing, for example, undoes what speaking does. Telephones provide communication circuits, providing an input device at one end of a connection and an inverse, decoding process at the other end. Similarly, the language component of a person talking on the telephone may be said to communicate with the (inverse) language component of the listener. The knowledge components of the two are in communication.
Using this model of communication, we may define a communication receiver as the implementation of a function f-1() where f() is referred to as the communication transmitter.
If the first process merely copies the input to the output, and the inverse process copies its input to its output, communication is taking place under our model. This ``straight wire" system clearly neither encodes nor decodes, showing an obvious difference between this model and other models based on encoding and decoding of messages. None of these three concepts is essential to our model of communication. We use the communication jargon ``encoding" and ``decoding" to represent f() and f-1() but note that in cases such as the ``straight wire" system it becomes clear that this terminology captures the essence of this copying operation only weakly, if at all.
Not all information transmitted represents communication. Given communication defined in terms of inverse processes, the page you are reading isn't communicating with you. You are receiving the information that is on the page because of visual processes. The author is communicating with you through processes that first took ideas resulting in written text, and an inverse process within the reader is taking written text and transforming it back into thoughts. Similarly, if one person is talking to another and is nervous, the nervousness may be communicated to anyone who can translate observed perspiring or a quivering voice or shaking hands into an understanding that the first person is nervous.
Figure 1 shows a hierarchy of processes used for information transfer and communication. Consider each hierarchy representing an individual. One communicates her or his knowledge in language, which is further transformed or encoded as sounds. These are decoded on the right hand side, producing language from the sounds using an inverse function, and the language is converted back into knowledge, again through the function inverse to the function that initially transformed the information. The decoding takes place on the right hand side, with ascending function f-1(x) representing the inverse of the f(x) function; the functions on the right hand side of Figure 1 ``undo" what the earlier processes on the left ``did."
In addition to different physical phenomena being modeled by specific layers of a hierarchy, psychological and epistemological phenomena may be represented by layers in a communication hierarchy [Los97]. For example, a perception function may have as output a belief instilled in the individual, and a belief function in turn may produce knowledge.
Note that the choice of a hierarchy as a structure is arbitrary and that a set of communication processes may be also understood as placed in a straight line, with one set of processes producing information which is then presented to the inputs to the inverse processes. Using the notion of a hierarchy facilitates an understanding of the relationship between related individual information processes and the larger communication processes. We also do not mean to imply that encoding or decoding as Shannon viewed these operations are essential to this model of communication. While the hierarchical model emphasizes transformations which may be interpreted as the traditional encoding and decoding found in more traditional Shannesque models, remember that a straight wire may allow one end to communicate with the other end, with little processing present to suggest the encoding and decoding metaphor.
This model of communication differs from many more traditional approaches to communication. At the core of the difference is that the model used here is not human-centered. It doesn't have being human, or even organic, as a prerequisite for being an endpoint in a communication system. This expands communication systems to include processors such as computers, as well as less sophisticated reproducing devices such as photocopiers. By not arbitrarily limiting the system to humans, little is lost and much is gained, both by describing phenomena outside humans that seem very similar to what is traditionally referred to as communication and by greatly simplifying the definition and making it precise. Because the definition is not explicitly biological in nature, it is also not explicitly intentional in nature. Intentional characteristics can be incorporated into processes; modeling communication as complementary processes can be consistent with intentional processes and human communication, allowing models of communication such as that described by Sperber and Wilson [SW95] to be seen as describing communications that are special cases of this more general model of communication. Unlike many models of communication, however, our communication processes are not limited to intentional or human phenomena. Similarly, communication does not have to transmit meaning or be intrinsically symbolic, although these phenomena can be incorporated. This differs from Shannon's model of communication, which begins with a source which "produces a message or sequence of messages to be communicated to the receiving terminal" [Sha93b, p. 6]. Similarly, when Shannon says that we can "roughly classify communication systems," the categories are described as having a message component [Sha93b, p. 7-8].
Scholars such as Oliphant [Oli97] have accepted that many of the popular existing definitions of information are inadequate for their work and have developed their own definitions for the purposes of their research. We believe that the proposed definition of communication is a universal, domain independent definition that allows for many of the concerns motivating other definitions of information. We thus feel that this definition serves as a common definition of communication for the field, providing a common language. Additional constraints may be added to this, but they are not part of the fundamental nature of communication; they are additional constraints that may or may not be worthy of discussion based on their own merits.
As we defined an information channel, we may define a communication channel as the set of components in the universe that implement the functionality needed for the communication process to take place. Because communication requires two or more information processes, a communication channel always consists of two or more information channels in series. By using the word needed, we refer only to those components that directly and obviously contribute to the operation of the channel, ignoring the fact that a butterfly moving on the other side of the world does affect the performance of a communication system on the reader's side of the world.
Using this model of communication, we may measure both the amount of information that is present at different points in the model. For example, we may measure the amount of information entering the system as well as capture and describe the information at the output of the first and second processes and at intermediate points in a communication hierarchy.