These orientations have been similarly described and critiqued by Terry Winograd and Fernando Flores (Understanding Computers and Cognition , 1986) under the label of a rationalistic tradition . Winograd and Flores aim their critique more at epistemology than ontology (cf. 1986, pp. 14-20), but their delineation of these problematic tenets generally overlaps with Lakoff's objectivism. The ontological component of their rationalistic tradition specifies '...a 'real world' made up of objects bearing properties' and ' 'objective facts' about that world that do not depend on the interpretation (or even the presence) of any person.' (pp. 30-31)
Most Western philosophers have taken such objectivity for granted, spending their energy debating the manifestation or comprehension of the 'real world' and 'objective facts'. Hilary Putnam notes that it's impossible to find any philosopher between the Pre-Socratics and Kant who didn't subscribe to objectivism as outlined here (Reason, Truth and History ), 1981). The only significant development so far was that in the wake of Descartes, Locke, and Kant the presumption of objectivity was submerged in the debates over how a subject comes to know it (i.e., comes to cognitively reflect the world's objective order) (cf. Richard Rorty's Philosophy and the Mirror of Nature , 1979). Rorty notes (1979, pp. 5 ff.) that three of this century's major philosophers (Heidegger, Wittgenstein, and Dewey) all constructed (and then recanted) a comprehensive vision of how we engage the world as it (presumably) is. To use the postmodern jargon, these (and other) philosophers have run into the impossibility of a 'totalizing narrative'.
Nowhere is this limitation more apparent than in studies of actual human (inter-)activity. People are continually formulating and changing their 'rules of engagement', including the delineation of which knowledge elements and which relationships among those elements are 'in play' at a given moment. This is most relevant where the purpose at hand is the design of prospective artifacts and procedures. Design (in this sense) is effective only to the extent that it recognizes and addresses the situated contextuality of the target work activity. (cf. Lucy Suchman, Plans and Situated Actions , 1987; Pelle Ehn, Work-Oriented Design of Computer Artifacts , 1988). In addressing IT interventions, it is necessary to set aside grandiose allusions to 'objective reality' and concentrate on the 'objectives realized' in actual or prospective praxis.
There are two distinguishable senses of reductionism, and they have both proven problematical in addressing human (inter-)activity and IT interventions. In terms of methodology, reductionism is the position that a phenomenon of interest is best addressed by breaking it down into its atomic constituents. This (in theory) permits the study of the phenomenon at its most basic level of manifestation, and (again in theory) allows the prediction of the phenomenon's 'macro'-level behavior from its 'micro'-level character.
In terms of epistemological criteria, reductionism refers to the mapping of all phenomena of interest onto a set of (presumably well-understood) elements or principles. Where these are atomic, this sense of the term intersects the previous one. Radnitzky (Contemporary Schools of Metascience , 1970) attributes this second sense of reductionism to the prevailing tradition in Western analytical science -- logical empiricism -- and delineates this primarily Anglo-American tradition in terms of three key features:
In the strictly methodological sense of reductionism, this has led to a futile search for the atomic elements of rational cognition. In the strictly epistemological-criterial sense, it has led to the construction of brittle models treating human cognition as a series of deterministic processes. Both senses of reductionism came into play when the metaphor of the digital computer was applied to the mind via cognitivism -- the subject of the next section.
Cognitivism -- '...the general doctrine that behavior can be explained by reference to cognitive or mental states' (Woolgar, 1987, p. 313) -- views humans as symbol processing systems whose essential behaviors can be accurately modeled by abstract formalisms realized via computer-based operational models (simulations), logical models, quantitative models and formal grammars. As Dreyfus (1988, p. 100) puts it: 'Cognitivism is rationalism plus the computer as a model of how this rationalist account of mind actually works.' While the roots of cognitivism can be traced back centuries, its rise and influence on IT are recent. Taxonomically, its descendants include cognitive psychology's human information processing (HIP) paradigm and artificial intelligence (AI). Methodologically, applied cognitivism (1) explicitly prioritizes functional simulation over structural analysis and (2) perpetuates Cartesian mind / body dualism by segregating cognitive phenomena from their physiological bases in enquiry.
The cognitivist version of the human mind is an input / output model of information processing. Exclusive of the 'inputs' and 'outputs', Winograd and Flores (1986, p. 25) summarize the tenets as follows:
- All cognitive systems are symbol systems. They achieve their intelligence by symbolizing external and internal situations and events, and by manipulating those symbols.
- All cognitive systems share a basic underlying set of symbol manipulation processes.
- A theory of cognition can be couched as a program in an appropriate symbolic formalism such that the program when run in the appropriate environment will produce the observed behavior.
After some 20-30 years of dominance, cognitivism has shown its limitations, and this approach is now being challenged. In its guise of 'cognitive science', cognitivism is critiqued by Cherniak (1988) on the basis of unreasonable demands entailed for the cognitive capacities it presumes. As a component of the 'rationalistic tradition', cognitivism is critiqued by Winograd and Flores (1986) on the basis of its conceit that cognitive 'quanta' are dissociable from the biology, history, and phenomenology of the individual. Finally, as a 'functionalist approach' to the ontological problems of mind and body, cognitivism is qualified by Churchland (1990) as a simulation exercise rather than an analytical framework per se.
These three critiques share a prescription for greater attention to the individual as an embodiment of experience rather than a symbol processing automaton. Cherniak points to neurophysiology and related research as a direction for cognitive science, insofar as '...the field seems to have recalled lately that cognition is, after all, accomplished with a brain.' (1988, p. 412) Winograd and Flores, in emphasizing personal experience (individual phenomenology and situated action) over rationalist cognition, have served as a central reference point for alternative IT perspectives. Churchland suggests resolution of the mind / body problem must now be informed by insights into the physical basis of cognition (e.g., neural networks) rather than simulation of cognitivist models projected onto behavior. This means approaching directed cognitive activities (e.g., decision making) as personally-embodied, experience-derived action rather than abstract, algorithmic processing.
In the mid-1980's, I was a conventionally-trained researcher in artificial intelligence (AI). This meant that objectivism and cognitivism were integral components of my field's approach to IT interventions. During a wide-ranging survey of knowledge based systems implementations (cf. Östberg, Whitaker and Amick, 1988), it became apparent that the problems bedeviling the capture and replication of experts' 'knowledge' derived from mistaken presumptions that:
This led me to reformulate the proper role of knowledge based systems in terms of the pattern(s) of interaction between consultant and consultee (cf. Whitaker and Östberg, 1992). This focus on interactivity was reinforced by my exploration of the Scandinavian approach to workplace IT interventions -- Participatory Design (PD). It was further reinforced by the necessity of addressing interpersonal interaction in my later CSCW work. I turned away from objectivism, reductionism, and cognitivism because they could not constructively explain the patterns of interactivity which were increasingly important in my work. More generally, I abandoned these orientations because the fact that they were systematic obscured the fact that they were not suited to addressing the systemic character of cognition and interactivity.
