8.3 Natural and Cultural Dymanics of Information and Media Technologies

What distinguishes contemporary humans from our pre-Ice Age ancestors is that our adaptations are primarily cultural. Many of the processes of natural selection that shaped Homo sapiens have been superseded by much faster mechanisms of adaptation. The human evolutionary clock may have slowed for the moment in some respects, because selection pressure can be accommodated by technical and social means rather than natural selection.

As Donald (1992) argues, the information age extends previous trends in the evolution of human cognition. His reconstruction of the origins of the modem mind makes the credible claim that the unfolding drama of our distinctly human neurological capacity has been characterized primarily by externalization of information, first as gestures and rudimentary songs, later as high-speed articulate speech, and eventually as visual markings that enabled storage of information in stable nonbiological systems.

Norman (1993) has succinctly captured this theme of information externalization in the title of his trade book, Things That Make Us Smart. He argues that the hallmark of human cognition lies not so much in our ability to reason or remember but rather in our ability to construct external cognitive artifacts and to use these artifacts to compensate for the limitations of our working and long-term memories. Norman defines cognitive artifacts as artificial devices designed to maintain, display, or operate on information in order to serve representational functions.

As Greeno (1991) notes, "a significant part of what we call 'memory' involves information that is in situations ... rather than just in the mind of the behaving individual" (p. 265). Indeed, a sizable body of literature describes some profound limitations of internal representations, or in our terms, MIROS, that is, Mental-Internal Representations of Situations (see, for example, Carroll & Olson, 1988; Craik, 1943; di Sessa, 1983, 1988; D. Gentner & D. R. Gentner, 1983; D. Gentner & Stevens, 1983; Greeno, 1989; Johnson-Laird, 1983; Larkin & Simon, 1987; Lave, 1988; Payne, 1992; Rouse & Morris, 1986; Wood, Bruner & Ross, 1976; Young, 1983; see also 12.3.1.2). These works suggest that without the support of external devices or representations, MIROS are typically simplistic, incomplete, fragmentary, unstable, difficult to run or manipulate, lacking in firm boundaries, easily confused with one another, and generally unscientific.

8.3.1 Thermodynantic Efficiency of Externalization

There is reason to believe that the scope and complexity of MIROS are constrained by the thermodynamics of information storage and processing in biological systems (see 3.1.35). Seemingly lost in 3 decades of discussion on the problems of internal representation is Hawkins's (1964) insight that external representations can confer gains in thermodynamic efficiency.

The capacity to learn is an externalization of function, the creation outside the cell nucleus of a new way of acquiring and storing vital information. The nucleus has its limitations, of information capacity and rate of evolution.... The point of innovation is that the code description of a machine [cell] that learns, that acquires information from and about the environment, can be small compared with what the machine [cell] learns . . . When such a step occurred in the evolution of animal species, an essential limitation upon all previous evolution was removed: The self-reproducing molecule was no longer burdened with the organism's entire stock of information. The importance of such a step is comparable to that of the beginning of life itself... (pp.272-73).

This line of argument is based partly on the work of Shannon and Weaver (1949), the mathematicians who applied thermodynamic analysis to technical problems such as the coding of messages, transmission of messages over channels, the maximum rate of signal transmission over given channels, and the effects of noise. Hawkins (1964) reasoned from Shannon and Weaver's theoretical treatment of information that learning, whether the system that learns be machine or human, ultimately confers its benefits through increased thermodynamic efficiency.

In the conditioned reflex and in the switching mechanism that is the basis of the large digital computer, the essential thermodynamic condition is again the availability of free energy for the performance of entropy-reducing, order increasing work. The switching mechanisms transmit flows of energy larger than the incoming signals that direct their behavior. Through reinforcement and inhibition, relatively simple stimuli come to release complex responses adapted to the character and behavior of the environment. The patterning of such responses represents, vis-à-vis the environment, a lowered entropy of arrangement (p. 273).

The externalization of information beyond the limits of the cell nucleus referred to by Hawkins is only one of the first of many strategies that life has evolved for increasing thermodynamic efficiency. Even greater gains accrue if an organism can off-load the work of information storage and processing to the environment itself and thus reduce the biological costs associated with maintaining and processing in neural networks. Unfortunately, explanatory models in the cognitive sciences still emphasize relatively complete mental representations rather than models that account for representation as distributed between the environment and the brain. As Zhang and Norman (1994) argue, this traditional approach to cognition

... often assumes that representations are exclusively in the mind (e.g., as propositions, schemas, productions, mental images, connectionist networks, etc.). External objects, if they have anything to do with cognition at all, are at most peripheral aids. For instance, written digits are usually considered as mere memory aids for calculation. Thus, because the traditional approach lacks a means of accommodating external representation in its own right, it sometimes has to postulate complex internal representations to account for the complexity of behavior, much of which, however, is merely a reflection of the complexity of the environment (p. 88).

All things being equal, we might expect investment of organic resources in improved perceptive capabilities to be a more effective strategy for organisms than construction of elaborate MIROS. Regardless of whether improved perception is acquired through learning or natural selection, it allows organisms to more effectively exploit information reflected in the structure of the environment--information that is maintained at no direct biological cost to the organism.

Yet all things are not equal: A number of factors determine how biological resources are divided between perceptual capabilities and MIROS. These factors include the niche or occupation of the organism, the availability in the environment of information related to the niche, the biological costs of action requisite to information acquisition, the costs of developing and maintaining perceptual organs, and the costs of developing and maintaining the MIROS. In addition, when information acquisition involves exploration or investigation by the organism, there is a cost of opportunities forgone: Moving or adjusting sensory organs to favor selection of information from one sector of the environment may preclude, for some time, selection of information from other sectors.

Consider how these factors operate at the extremes to favor development of, respectively, perception and MIROS in two hypothetical groups of people concerned with navigation in a high-security office building. The first group are ordinary workers who move into a building and after a short time are able to navigate effectively using an environment rich in information such as signal, landmarks, changes in color schemes, and the like.

If the building is well designed, it is unlikely the workers will invest much mental effort in remembering the actual details of the spatial layout. "Why bother?" they might ask. "It's obvious; you just keep going until you find a familiar landmark or sign, and then you make your next move. We don't need a mental model because we can see where to go." Norman and Rumelhart (1975) have demonstrated that living in buildings for many months is no guarantee that inhabitants will be able to draw realistic floor plans. In fact, such residents often make gross errors in their representation of environmental layouts--incorrectly locating the position of doors, balconies, and furniture.

Returning to the high-security building, suppose a second group, more nefarious and temporary, are commandos hired to steal company secrets in the same building during the dead of night when visual information about the environment is not so easily obtained. Each use of flashlights would entail risk of discovery (a kind of cost) and each act of exploration or orientation would increase the possibility of being caught. In preparing for their raid, therefore, the commandos might be willing to spend a great deal of time familiarizing themselves with the layout of a building they may raid only once. "Sure," they might say, "we have to invest a lot of mental resources to memorize floor plans, but it's an investment that pays off in saved time and reduced risk."

8.3.2 Coupling and Information Transfer

Perception, in the view of ecological psychologists, cannot be separated from action: Perceiving involves selecting and attending to some sources of information at the expense of others. Human eyes, for instance, are constantly flicking across the visual field in rapid eye movements called saccades. Natural environments cannot be easily modeled in terms of communications channels, because such environments typically contain numerous independent sources of information. Organisms attend to these sources selectively, depending on the relevance of the information to their needs and intentions. To use inadequately the communications metaphor, organisms constantly switch channels. Moreover, most organisms employ networks of sensors in multiple sense modalities and actively manipulate their sensor arrays. It is unclear how we should think of such sensor networks in a way that would be consistent with Shannon and Weaver's rigorous technical meaning for channel in which they model information flow as a single stream of serial bits (see 4.4.2).

According to Gibson's paradigm (1979), the information contained in situations is "picked up" or selected rather than "filtered" as suggested by the metaphors associated with many popular models of memory and perception. In the context of thermodynamics, selective perception of the environment confers benefits similar to the switching mechanisms of learning referred to above by Hawkins: Organisms expend small amounts of energy attending to those aspects of the environment that might yield large returns.

Hawkins extended another Shannon and Weaver insight by noting that some kind of coupling is a necessary condition for duplication or transmission of patterns. He argued that the idea of coupling--widely misinterpreted by communications and media theorists to mean mechanical, deterministic coupling--was used by Shannon and Weaver to refer to thermodynamic (probabilistic, stochastic) coupling. Thermodynamic coupling is a many-to-many form of linkage, a concept of coupling that not only accounts for the possible gains in efficiency but also preserves the ancient Greek sense of information as transference of form:

Man's physical coupling with his environment is not that of an intrinsic source *of energy, but is weaker, more purely thermodynamic. He controls his environment by subtle changes in its order, so that the streams of natural process flow in new channels. But the control runs both ways. Competence is derived from acceptance of the de facto order of things. The potter who shapes the clay has long been the image of a godlike power, but this is not the perception the potter has of himself. He must be sensitive to the properties of the mix and to its responses to firing in shape and color and texture. The potter is as much transformed by his art as the clay is (Hawkins, 1964, p. 310).

As Maturana (1978) notes, information conceived as transfer of pattern or form implies that

... learning is not a process of accumulation of representations of the environment; it is a continuous process of transformation of behavior through continuous change in the capacity of the nervous system to synthesize it. Recall does not depend on the indefinite retention of a structural invariant that represents an entity (an idea, image, or symbol) but on the functional ability of the system to create, when certain recurrent conditions are given, a behavior that satisfies the recurrent demands or that the observer would class as a reenacting of a previous one (p. 45).

Behavior so informed by the environment represents a lowered entropy--that is, a greater orderliness of arrangement. Chaotic, disorganized, and arbitrary aspects of an organism's activity are ameliorated by attention and intention directed towards aspects of the environment that are related to the organism's ecological niche. The orderliness and organization of behavior that results from niche-related attention and intention can be characterized as intelligence. Such intelligence is thermodynamically efficient because it leverages the expenditure of small amounts of biological energy (Gibbs Free Energy) to guide much larger flows of energy in the external environment.

Media users benefit from this thermodynamic leverage when they expend modest attentional resources to acquire information about how to control large amounts of energy. A speculator who makes a quick killing on Wall Street after reading a stock quote is making thermodynamically efficient use of media technology.

To summarize the preceding discussion of coupling and information transfer, one should understand that the extension of human cognitive capacity through media technologies reflects broader evolutionary trends characterized by increasing externalization of information storage and processing. Such externalization increases thermodynamic efficiency, reducing the organic costs of cognition by distributing the "work" of representing situations between humans and their external environment. Indeed, one arguable way to define higher-order learning is by the degree to which it permits individuals to benefit from externalization of information storage and processing. This can be conceptualized as literacy or, more generally, we propose, as mediacy. Both literacy and mediacy are qualities of intelligence manifested by the facility with which an individual is capable of perceiving and acting on mediated information. Bruner and Olson (1977-78) invoke this concept of mediacy succinctly when they define intelligence as "skill in a medium."

8.3.3 Simplicity and Complexity

Ecology in general is concerned with predicting and explaining how matter and energy are transferred and organized by members of biological communities. Since transfer and organization of matter and energy are ultimately governed by thermodynamics rather than purely mechanical exchanges, ecological sciences eschew purely deterministic explanation (one-to-one, reversible couplings) in favor of stochastic, probabilistic explanation (many-to-many, nonreversible couplings). Stochastic description and analysis is based on information transfer and formalized by measures of entropy or organized complexity(3).

Information is thought of roughly as a measure of level of organization or relatedness. Entropy is a measure of degrees of freedom (von Bertalanffy, 1967; Gatlin, 1972) or opportunities for action. So viewed, complex systems can be said to offer more freedom of action than simple systems because complex systems (see 3.1.1.1.1) are more highly organized, with more and higher-level relations. Complex biosystems, for example, encompass more species and support longer food chains than simple biosystems; tropical rain forests afford more freedom of action, more opportunities to hunt and gather than does arctic tundra. To change the context, a city offers many more opportunities for human action--different types of work, recreation, and socializing--than does even the largest cattle ranch.

Extremely simple systems can be said to offer no opportunities for action because (a) there is no organization--all is chance and chaos or (b) organization is rigid--all relations are absolutely determined. A square mile of ocean surface is simple and chaotic, whereas a square mile of sheer granite cliff is simple and rigid. Rigid systems compel, yet they do not enable.