Critical evaluation of DC

The reviewed DC studies highlight how medical practitioners manage cognitive overload by attending to material structures (including team members) that co-organise and represent information no single individual could handle by individual cognition. The studies emphasise how cognition is distributed in local interaction between a practitioner and his environment. However, explanations of how a distributed cognitive system manages cognitive processes are not fully elaborated. Below, it is argued that to come up with comprehensive explanations of the enabling conditions for local situated cognition, the trans-situational and non-representational aspects of cognition should be discussed explicitly. Such discussion will bring us closer to an understanding of how the use of external structures (verbal, physical etc.) is enabled by tying temporal distribution to sociocultural norms and experience. The main critique points relate to the following headings: (a) artefacts and local cognitive scaffolding and (b) representations, design and generalisations.  

2.3.3.1 Artefacts and local cognitive scaffolding

Within the reviewed articles, there is a tendency to adapt DC as a framework that describes how healthcare professionals rely on technological artefacts in decision-making and on information sharing. As such, a prevalent argument for applying a DC framework is that the approach is ideal for investigating these types of cognitive processes. Yet, as Hutchins particularly underlines, DC is not a special type of cognition that is identified in certain circumstances; rather, cognition can always be seen as distributed (Hutchins, 2014). Thus, when Rajkomar and Blandford (2012a:588) emphasise: “One question this study sought to address was whether DC is particularly well suited as a theoretical framework to study healthcare socio-technical systems […] by looking at whether cognition was distributed in the ICU”, they miscommunicate the underlying assumptions of DC. In a similar vein, DC is regularly reduced to a simplified framework for looking at how multiple aspects of a situation affect outcomes in a stressed environment. The overwhelming focus on artefacts is crucial to understanding why DC is widely acknowledged in healthcare: “Distributed Cognition has been proposed as being a framework of choice for studying healthcare work.

It is particularly well suited for studying interruption resumption since it explicitly considers the design and use of artifacts, which play a crucial role in supporting interruption resumption” (Rajkomar and Blandford, 2012b:108).

Thus, in the studies, focus falls on how external structures materialise in artefacts and how such artefacts scaffold cognition within a spatially distributed cognitive system.

However, the perspective is more than a context sensitive model of cognition that takes the role of external artefacts into consideration. While Hollan et al, (2000) identify three ways in which a cognitive system can be distributed (in time, in and between people and in environmental structures), the reviewed studies almost exclusively emphasise how information processes are facilitated by technological solutions in various ways. For instance, such solutions reduce cognitive overload (Bang and Timpka, 2006), enhance medical overview (e.g. Bossen and Jensen, 2014) and imply a coordinating role for action (Rajkomar and Blandford, 2011). Although this is useful knowledge, a narrow focus on functional properties of artefacts reduces the explanations of decision-making to how physical materiality impacts action in relation to a specific task. It appears as if activities that do not contribute to a smooth and immediate accomplishment of an explicit goal are evaluated as either dysfunctional or unimportant. Local detours related to interpersonal issues (for instance, concerns about the well-being of team members, calming down patients and chatting with relatives) and other situational features are not addressed explicitly in the DC studies. However, rather than being noise in problem-solving, such detours often serve as anticipatory activities that, by a circuitous route, enable an overall more effective accomplishment of goals. This will be demonstrated in the analyses.

The immense focus on problem-solving in the DC studies entails a single-minded reading of activities in the cases investigated. By defining a problem space, a task and a pre-defined goal, much is left out in the process of achieving that goal: “Problem solving research typically focuses on how subjects find their way through an abstract problem space. However, placed in a concrete and imaginable setting, one can scrutinise what

happens“ (Cowley 2014b:243). Cognitive science applies models of cognition and test hypotheses, providing clear, demarcated and comparative results but leaving out the relevance of dynamics in human interactivity that cannot be classified as logical task performance. In the search for explanations of information processing and functional properties in task accomplishment, the concept of representational states in particular is explored. According to this perspective, it becomes relevant to investigate how effective representations are embedded in the ubiquitous artefacts that are used in healthcare decision-making (Kaufman et al., 2009).

2.3.3.2 Representations and artefacts

With the ethnographic approach, analyses based on interviews and video-observations, generally highlight how information flows with the aid of artefacts. Informed by rich data material, the conclusions are based on a rather outworn understanding of communication and language processes. For instance, Kaufman et al., (2009) indulge in a Turing-like model of information transmission in their otherwise convincing investigation of how technological issues affect workflow in telemedicine: “The information was transmitted through eight communication modalities […] Three of them are synchronous and the remaining five are asynchronous. […] The medium of communication provides different resources for establishing common ground or mutual understanding" (Kaufman et al., 2009:584). Ironically, DC represents third wave developments in cognitive science, where the computational metaphor of cognition is exchanged with a more ecological and contextual understanding of the matter. Nonetheless, several DC studies blend a third wave systemic approach with old school conceptualisations that reduce cognitive processes to computational calculations of representations: “Sensemaking as a process of reciprocal, ongoing interaction between the search for information, meaning ascription and action fits nicely with our empirical data. Also, it makes good sense to consider the distributed socio-technical system the primary analytical unit, as suggested by distributed cognition, and the achievement of overview a computational process across actors and artefacts” (Bossen and Jensen, 2014:266). In their study, the classic metaphor of the cognitive domain – cognition as computation – is adopted and guides the analysis. Hansen and Lyytinen (2009) emphasise how Van Gelder (1995) has criticised this view on cognition in a critique of DC: “Thus, intelligent behavior (i.e., cognition) is understood as an information processing mechanism that results in appropriate action toward the achievement of specific goals: “the mind is a special kind of computer, and cognitive processes are the rule-governed manipulations of internal symbolic representations” (Van Gelder, 1995: 345). While the focus is on artefact-based tasks, the processes of information, memory and representation in relation to how a goal is achieved are sought at the expense of an understanding of the multi-scalarity of human interaction.

Finally, while the cognitive sciences have been preoccupied with statistical analysis of how output is generated, the ethnographic approach in DC-studies apparently contrasts with conventional, experimental approaches. With DC the context is crucial to how results

are produced. DC’s context sensitivity favours particulars and deals with the reciprocal and dynamical relations within the distributed cognitive system. System output is a result of how individual capabilities are used in interaction with what the environment offers. As interaction continuously affects how the system operates, standardisation and generalisation of the enabling conditions for decision-making are challenging:

Taking context seriously means finding oneself in the thick of the complexities of particular situations at particular times with particular individuals. Finding commonalities across situations is difficult because studies may go off in so many different directions, making it problematic to provide the comparative understanding across domains that Brooks (1991) advocates. How can we confront the blooming, buzzing confusion that is “context” and still produce generalizable research results? (Nardi, 1996:70)

However, as much depends on situations, particulars are often highlighted in DC-studies followed by a request to look deeper into similar cases. Critical qualitative research suffers from the authority of the ordinary conception of conventional rules for generalisations.

However, rather than being just ‘a cup of coffee theory of understanding’, as the philosopher Karl-Otto Apel (1973) describes qualitative studies, reliable qualitative research insists that general features are connected with particular situations:

there are not just particulars in concrete situations […] For example, anxiety holds distinctive general qualities and significances which have to do with losing one’s grip over one’s situation (Holzkamp, 1983). Knowing such general features is helpful in guiding the understanding of concrete anxieties. But concrete anxieties also hold particular personal and contextual qualities and significances. (Dreier, 2007:190)

To sum up, DC has been preoccupied with functional manipulations of external structures.

The immense focus on how artefacts are used as cognitive resources is tied to only one of Hollan et al.’s (2000) three dimensions of distribution. A few studies emphasise what the perspective originally sought to describe: how teams manage complex tasks with clear and distinctive operational criteria for accomplishment. For instance, Fioartou et al. (2010) investigate how distributed situation awareness helps avoid fixation errors. Still, at an empirical level, cognitive distribution in time is not emphasised in any notable way, and the way in which culture, experience and social phenomena influence situated cognition is left for general reflection and theoretical discussions.