Distributed Cognition: A Useful Theory in Human-Computer Interaction and Interface Design

Submitted to IASDR - International Association of Societies of Design Research Conference, 2019
by Rachael Paine and Traci Rider

Designers aim to impact human activity through designed artifacts and systems. Effective inquiry in the field of design relies on researchers understanding the individuals, cultures, and societies for which they aim to produce new knowledge. By creating a broader view of cognition – one which includes the interactions amongst various users, technology, and artifacts – the theory of Distributed Cognition (DCOG) provides a systems perspective for designers. Useful when investigating collaborative working and learning, cognitive technologies, and more, DCOG is an approach to systematically researching all agents, both human and non-human, within cognitive systems. For designers of artifacts for human consumption, particularly in graphic and industrial design, DCOG is a useful framework when considering how technology can aid collaboration, reduce the cognitive load of the user, enhance memory of the user, and aid the process of internalizing external models of information.

Keywords: Distributed Cognition; human-computer interaction; user interface design; user experience design

2. Background and Historical Basis

As far back as the 4th century BC, Greek philosophers attempted to explain the nature of the human mind. Whereas it had previously been accepted that mental disorders were rooted in supernatural causes, in his writing “On the Sacred Disease”, the Greek physician Hippocrates began theorizing physical abnormalities were to blame (Thagard, 2018).

In addition to Greek theologians, works from Laozi and Confucius began to pave the way for Chinese psychology in the 2nd Century BC. Great thinkers of the time began to see the brain as the nexus of wisdom and sensation. An ancient text known as “The Yellow Emperor’s Classic of Internal Medicine” presented personality theories and models for analyzing mental disorders. Emerging theories stated that mental disorders resulted from instability within one’s physiological and social being (Raphals, 2017).

During the 19th century, researchers began applying experimental methods to psychological study. Termed experimental psychology, laboratory methods became the most prominent means of studying cognition (Thagard, 2018). Both human and animal participants were used in laboratory settings to study memory, perception, and learning. Wilhelm Wundt, known as the “father of experimental psychology,” developed the first methods for studying mental operations (Kim, 2016).

Over time, scientists began to focus on observable behavior rather than the invisible mind. By the early 1900s, this approach, termed behaviorism, dominated experimental psychology. Psychologists working under the behaviorist paradigm restricted themselves to examining the relationship between stimuli and observable behavioral responses (Ashcraft, 2002).

During the 1950s, many scientists began to reject the anti-mentalism approach of behaviorism in favor of theories of the mind based on complex representations and computational procedures (Thagard, 2018). Thus began the intellectual movement known as the Cognitive Revolution. Spanning the 1950s and 60s, cognitivism became the dominant approach to psychology resulting in the field that is now known as cognitive science. The field includes researchers from neuroscience, anthropology, psychology, linguistics, artificial intelligence, and philosophy.  

Founders of cognitive science including George Miller, John McCarthy, Marvin Minsky, Allen Newell, Herbert Simon, and Noam Chomsky, along with other researchers and thinkers of the time, were traditionally approaching research from a positivist perspective. Cognitivism practiced observing outward behavior as well as its relation to brain activity and building computationally-based models of the mind. The primary method of studying human cognition was through laboratory experimentation with human participants. Researchers studied thinking under controlled conditions (Thagard, 2018), surpassing behaviorism in popularity as a psychological paradigm and becoming the dominant approach to psychology. The Cognitive Revolution resulted in the accepted understanding that cognition was inside the head of the individual. Humans process information by perceiving stimuli with their senses, interpret the stimuli in their brain, and produce behavior in response (Ashcraft, 2002).

“When cognitive and symbolic anthropology split off from social anthropology in the mid-1950s, [scientists] left society and practice behind” (Hutchins, 1995a). Cognitivists turned away from explorations rooted in society and focused their studies on what the individual needed to know in order to function as a member of culture (Hutchins, 1995a). Many of these studies focused on language and what an individual could verbally communicate about what they knew (Hutchins, 1995a). By the mid-1980s, one cognitive scientist began to see a downside to this approach. Edwin Hutchins, a professor of cognitive science at the University of California, San Diego, was the first scientist to suggest that cognition is primarily socially distributed. He argued, “The emphasis on finding and describing ‘knowledge structures’ that are somewhere ‘inside’ the individual encourages us to overlook the fact that human cognition is always situated in a complex sociocultural world and cannot be unaffected by it” (Hutchins, 1995a).

This shift away from a focus on “mental content” reducible to an individual’s cognition to an approach proposing that cognition is not limited to an individual’s mind but involves interactions with other’s minds, with physical objects, and technology (Heylighen, Heath & Van Overwalle, 2004) represents a change at the level of inquiry within the field of cognitive science. When a researcher chooses what topic they are interested in and specifically what questions are being asked, they must decide what unit of analysis they are interested in investigating. Traditional cognitive science explored the individual mind. Hutchins’s Distributed Cognition changed the unit of analysis to the distributed socio-technical system made up of individuals and the artifacts they use (Mahamuni, Khambete & Punekar, 2017).

As a framework, Distributed Cognition is useful in analyzing the contributions and coordination of all individual agents within a system that facilitates the achievement of its goal. When discussing Distributed Cognition, the term “agent” is not always referring to a human. The English Oxford Dictionary defines agent as “a person or thing that takes an active role or produces a specified effect.” In “How a Cockpit Remembers its Speed” (1995b), Hutchins explored how a cockpit, with the actions of its pilots and instruments, and the coordination between them, complete a flight successfully. The pilot, copilot, and devices are working together to achieve a common goal. Here, each of these contributors – pilot, copilot, and devices – are all agents.  

The idea of humans offloading cognition into physical objects and technology is essential in the theory of Distributed Cognition. According to Heylighen et al. (2004), “Distributed Cognition is seen as the confluence of collective intelligence and ‘situatedness,’ or the extension of cognitive processes into the physical environment.” External actions can simplify mental computations. Activities such as adding, accounting, and navigation are more difficult if people rely on memory without aid from external supports.

The theory of Distributed Cognition is useful for human-computer interaction (HCI). HCI researches the design and use of technology, focusing on the interfaces between the user and computers (Carroll, 2011). Interfaces can aid users in thinking more fluidly by distributing cognition into artifacts in the world. “Certain cognitive and perceptual problems are more quickly, easily, and reliably solved by performing actions in the world rather than by performing computational actions in the head alone” (Kirsh & Maglio, 1992).

In their research (known as the “Tetris Case Study”), Kirsh and Maglio (1992) present data which suggests humans can improve performance by offloading cognitive tasks into objects. The researchers hypothesized that expert Tetris (a tile-matching puzzle video game) players would rely less on the interface to translate and rotate objects, performing the majority of these spatial cognitive tasks internally. To their surprise, they found the opposite to be true. Expert Tetris players relied far more on the interface for rotation and translation tasks. Kirsh and Maglio discovered the expert player was distributing the perceptual task to the interface, saving the mental effort of rotating the shapes. They determined that it is “cheaper” to do the cognitive task of performing all the “what/if” scenarios on the screen rather than in the user’s head.

Such research uncovers the enormous potential of relying on a distributed approach to cognition as an interface designer. How might user experience and user interface designers create technological artifacts that seamlessly allow users to offload ineffective cognitive tasks, such as rotating objects, managing speeds, or performing computations?

3. Ontological and Epistemological Assumptions

According to Guba and Lincoln (1998), ontological assumptions have to do with the nature of and what we can know about reality. More concisely, Crotty (1998) refers to ontology as “the study of being.”

Distributed Cognition assumes that the world consists of multiple, socially constructed realities (Groat & Wang, 2002). Gioia and Pitre (1990) use the term constructionism for research perspectives which assume something is being constructed socially, by groups of people in a social context. The term constructionism refers to how people make sense of the world around them, thus creating reality. Unlike a positivist stance which posits that there is a single reality “out there,” DCOG assumes the existence of multiple realities. “The study of any one part necessarily influences all other parts” (Guba, 1981). Positivism aims to single out variables to control where constructionism recognizes that all “parts” are interrelated. Constructionism views all knowledge and meaningful reality as being contingent upon human practices, being constructed in and out of interaction between humans and their world (Crotty, 1998).

Due to the overwhelmingly social aspect of Distributed Cognition, the theory falls under the more narrow window of social constructionism. According to Berger and Luckmann (as cited in Heylighen et al., 2004), “Sociologists have long noted that most of our knowledge is the result of social construction rather than of individual observation” (Berger & Luckman as cited in Heylighen et al., 2004).

According to Creswell (2014), social constructionists hold the assumption that individual persons seek understanding of the world, developing subjective meaning of experiences. Simply put, there are a variety of multiple realities. Subjective meanings are “negotiated socially and historically” (Creswell, 2014). These meanings are formed in individuals through interaction with others and adapting to historical and cultural norms. Neimeyer (as cited in Patton, 2002) states, “All of our understandings are contextually embedded, interpersonally forged, and necessarily limited.”

DCOG addresses interaction among individuals. The distributed approach of this theory proposes that cognition not be limited to the individual’s mind but instead involves interactions with other minds and with things. The primary goal of the Distributed Cognition approach is “to account for how the distributed structures, which make up the functional system, are coordinated by analyzing the various contributions of the environment in which the work activity takes place, the representational media (e.g. instruments, displays, manuals, navigation charts), the interactions of individuals with each other and their interactional use of artifacts” (Rogers and Ellis, 1994). The focus of interaction among agents in a system is a critical factor in constructionist research (Guba, 1981).

Epistemology is the study of the nature of knowledge and perspectives about how knowledge can be acquired. Guba and Lincoln (2005) describe the epistemological question as one in which we make assumptions about the nature of the relationship between “the knower or would-be knower and what can be known.” 

Unlike the positivist assumption that the inquirer can remain separate and objective from the objects of the inquiry (Guba, 1981), DCOG shares epistemological assumptions with constructionism. The constructionist paradigm asserts that “the inquirer and the respondent are interrelated, with each influencing the other” (Guba, 1981). According to Groat and Wang (2013), researchers working under the constructionist paradigm not only acknowledge the interactive link between the research and participants, but they also recognize this interconnectedness has value. 

4. Methodological Assumptions

Beliefs about ontology and epistemology lead to assumptions regarding data needed for the “would-be knower” to collect (Guba & Lincoln, 1998). Crotty (1998) defines methodology as “the strategy, plan of action, process or design lying behind the choice and use of particular methods and linking the choice and use of the methods to the desired outcomes.” Creswell (2014) explains how all assumptions work together to develop the process of research. These underlying assumptions of the researcher regarding ontology and epistemology, in conjunction with the nature of the research problem, the researcher’s personal experience, and the audience for the study make up the research design. 

Because the social constructionist believes in varied and multiple realities, they hold an assumption that research must look for “the complexity of views rather than narrowing meanings into a few categories or ideas” (Creswell, 2014). The researcher relies on the participants’ views of the situation under investigation. In this natural, real-life setting, the researcher is observing what people say and do, usually in the context in which participants live and work. 

When operating under the framework of Distributed Cognition, there is a methodological assumption that qualitative methods have superior relevance. “Qualitative research is a means for exploring and understanding the meaning individuals or groups ascribe to a social or human problem” (Creswell, 2014). The research Hutchins conducted while developing DCOG as a theory aligns with Creswell’s (2014) description of the process of qualitative research. While investigating, Hutchins’s questions and procedures emerged and changed as more information presented itself in the process. His DCOG approach collected data in the participants’ setting, whether it be the cockpit of a commercial airliner or abroad a US Navy Ship. Hutchins inductively analyzed the data, moving from particular instances to general laws. Hutchins’s process illustrates the appropriateness of qualitative methodology, rooted in social constructionism, for DCOG inquiries. 

5. DCOG Areas of Concern and Focus

Several unique characteristics of DCOG make it a useful theory in design research. 

5.1 A Systems Level Theory

Distributed Cognition is a systems level theory. Hutchins’s drive to see cognition as situated in the larger socio-technical system is a clear example of what Patton (1990) calls “viewing things as whole entities embedded in context and still larger wholes.” A systems level study asks how and why a system functions in totality, seeking understanding of a complex world (Patton, 1990). As a conceptual framework, DCOG moves cognition “out beyond the skin of the individual person” (Hutchins, 1995a) resulting in distributed cognitive systems comprised of self-organized agents adapted to the environment. 

According to Kirsch and Maglio (as cited in Heylighen et al., 2004), a cognitive system is continually evolving. Agents within the system adapt to changes in its environment and effect changes in its environment. This is an example of what Patton (1990) refers to as holistic thinking. “Changes in one part lead to changes among all parts and the system itself.”

Distributed Cognition is a powerful theoretical perspective for researchers aiming to explain system behavior rather than breaking large systems into smaller parts for analysis. Gharajedaghi and Ackoff (as cited in Patton, 1990) argue that “the essential properties of a system are lost when it is taken apart.” By switching from a localized perspective to a systems perspective, researchers have a wider set of affordances for capturing observations. Hutchins (1995b) states, “With this new unit of analysis, many of the representations can be observed directly, so in some respects, this may be a much easier task than trying to determine the processes internal to the individual that account for the individual’s behavior.”

5.2 A Collaborative Focus

Distributed Cognition is a useful theoretical perspective for analyzing larger societal systems and interactions among individuals and the tools they use. Rogers and Ellis (1994) express a limitation of other theories for analyzing computer-mediated collaborative activities. They claim many frameworks “do not present an adequate means of studying the dynamics of collaborative activity in situ.” Distributed Cognition, however, is “capable of capturing cognitive activities as embodied and situated within the context in which they occur: social and organizational.” 

5.3 A Metaphor in Learning Technology

Dillenbourg (1996) views the socio-cultural approach of Distributed Cognition as a useful metaphor for designers of learning environments in the field of educational technology. By adopting technological artifacts, software, and learners as one single cognitive system, cognition is seen as “distributed over a human and a machine.” Dillenbourg proposes Distributed Cognition as a theory useful in designing Intelligent Learning Environments (ILE).

“Design is a creative process during which one attempts to make sense of learning activities or system features within a theoretical perspective.” Dillenbourg (1996) suggests that the concept of Distributed Cognition is not factual, but instead a way to visualize socio-technical systems in order to ask compelling questions and produce all-encompassing designs.

6. DCOG Theory in Design and Design Research

Distributed Cognition is an approach to systematically researching all agents, both human and non-human, within cognitive systems. Design researchers can employ DCOG when investigating relationships between an individual and external artifacts, between various individuals’ minds, or between a more diverse set of humans and artifacts (Zhang & Patel, 2006). For designers of artifacts for human consumption, particularly in graphic and industrial design, DCOG is a useful framework when considering how technology can aid collaboration, reduce the cognitive load of the user, enhance memory of the user, and aid the process of internalizing external models of information. Distributed Cognition has potential uses for human-centered and participatory or co-design methods (Mahamuni et al., 2017). Additional potential applications of the DCOG theoretical perspective are below.

6.1 Information Design 

Research rooted in the theory of Distributed Cognition is concerned with “the distribution, transformation, and propagation of information across the components of the distributed cognitive system and how they affect the performance of the system as a whole” (Zhang and Patel, 2006). In simpler terms, design researchers are often interested in how information moves between agents and how design can facilitate effective performance. Zhang and Patel (2006) believe that DCOG is useful when information is being distributed across internal and external representations. Internal representation refers to the knowledge within an individual’s mind whereas external representations are the knowledge in the external environment (such as within cognitive artifacts or conversations between people).  

Information designers can use DCOG to guide the creation of cognitive artifacts. Zhang and Patel (2006) offer insight into how external representations might aid a user’s intelligent behavior. Designed artifacts can serve as memory aids, reduce cognitive load, reduce information processing effort, provide knowledge unavailable from internal representations, make invisible information more sustainable, limit abstraction, and aid in decision-making strategies. Each of these examples point to the benefits of filtering design processes through the theoretical lens of DCOG. Without the understanding that knowledge can simultaneously exist in both the user’s head and in the environment, designers are apt to miss opportunities to create objects which enhance the user’s ability to think, learn, remember, and experience.

6.2 Emerging Technologies and Ubiquitous Computing

In the paper “The Emergence of Distributed Cognition: A Conceptual Framework” (2004), Heylighen et al. discuss the technological applications offered by DCOG while positioning technological agents as equal to human counterparts. These applications include the semantic web, artificial intelligence, machine learning, ambient intelligence, and natural language processing. DCOG is particularly useful for researchers who envision a world where “everyday artifacts and devices such as mobile phones, coffee machines and fridges (exchange) information and (coordinate) with each other so as to provide the best possible service to the user, without needing any programming or prompting – thus effectively extending the user’s mind into his or her physical environment” (Gershenson and Heylighen as cited in Heylighen et al., 2004). Daily processes that once required the management of humans could be offloaded into designed “smart” artifacts.

6.3 Human-Computer Interaction Research

In the paper “Distributed Cognition: Toward a New Foundation for Human-Computer Interaction Research” (2000), Hollan, Hutchins, and Kirsh state that the theory of Distributed Cognition provides a new “whole environment” way to think about designing for human-computer interaction. “For human-computer interaction to advance in the new millennium we need to better understand the emerging dynamic of interaction in which the focus task is no longer confined to the desktop but reaches into a complex networked world of information and computer-mediated interactions.” DCOG aids researchers in understanding the interactions between people and technologies. Hollan, Hutchins, and Kirsh (2000) propose “an integrated framework for research that combines ethnographic observation and controlled experimentation as a basis for theoretically informed design of digital work materials and collaborative work places.” Such research aims to aid the development of digital materials and interfaces.

7. Research Studies Guided by DCOG Theory

7.1 Analyzing Collaborative Computer Science Learning

In a qualitative case study analysis of interaction and learning conducted by Deitrick, Shapiro, Ahrens, Fiebrink, Lehrman, and Farooq (2015), the researchers employed Distributed Cognition to investigate learning situations where “understanding may not be entirely locatable within the individuals’ minds.” The researchers chose Distributed Cognition as a means to explore learning within cognitive systems comprised of “culturally and historically situated groups of students, teachers, and tools.” The researchers’ analysis documents how a system comprised of students, teachers, and artifacts can “accomplish conceptually demanding computer music programming.”

This research effectively illustrates the affordances of DCOG over traditional theories of cognition. Deitrick et al. (2015) relied strongly on interviews, think-aloud methods, and discourse analysis, collecting data while observing small groups of students solving a problem. More specifically, each set of two peers used a computer music tool called a BlockyTalky and a guitar to accomplish the task of programming an original piece of music. The students interacted with the artifacts and with a teacher. The discourse analysis captured by the researchers demonstrated a “process of recall, representation, and data transformation.” As one student figured something out, they shared that information with their teammate. Tools were used to both collect and share information as well as serving as hands-on problem-solving tools. Cognitive streams flowed between all agents, human and non-human, to create one socio-technical cognitive system. 

According to Deitrick et al. (2015), “DCOG theory enables us to richly analyze how that group works and learns, including understanding the specific student knowledge and tool design weaknesses that cause a breakdown within the group work to occur.” Here DCOG is used to analyze both the effectiveness of classroom learning techniques and designed prototypes (the digital learning tool, BlockyTalky). 

7.2 As a Conceptual Framework for Service Design

Mahamuni, Khambete, and Punekar (2017) conducted an organizational recruitment service case study rooted in the theoretical principles of Distributed Cognition. The researchers argued that the DCOG approach could address many unique aspects of service design. Service designers must take into account whole systems including the environment, and allowing for multidisciplinary participation and effective collaboration. The process of service design relies on stakeholders externalizing thoughts through in-process artifacts.

Service design is a long-duration activity (Mahamuni et al., 2017). Mahamuni et al.’s (2017) case study consisted of observing participatory methods to improve employee perception about an incentive-based employee referral program for recruitment. The stages of the service design project included creating personas, identifying key scenarios, creating a customer journey map, and mapping design patterns to the customer journey map. Each of these tools used in service design serves as methods to externalize internal mental representations.

Once previously internalized cognitive structures are visualized, multidisciplinary teams can have more effective communication about key insights while transferring knowledge to all stakeholders (Mahamuni et al., 2017). In this research study, DCOG serves as a useful tool for externalizing mental representations for communication between collaborators in service design.

8. Conclusion

As Banthany (1996) suggests the goal of designers is to “make and create systems that do not yet exist and thus change the world,” the theoretical perspective of Distributed Cognition can aid researchers in transforming thinking and uncovering potential for non-existing systems. DCOG creates a wider view of human knowledge structures – moving cognition “out beyond the skin of the individual person” (Hutchins, 1995a). Cognition becomes a system comprised of interactions between humans, technology, and artifacts. Design investigations are enriched by the inclusion of the social, cultural, and technological aspects of the context under study. The DCOG approach to research suggests a human and interface can work together to create a distributed cognitive system, where interfaces have the potential to serve as cognitive artifacts, supporting the user’s information processing.