Prior Knowledge: Highly Organized, Intricately Interconnected, Constantly Evolving

Introduction

Our brain is like a computing device which has its own data structures and processing methods. It is able to identify and categorize objects efficiently, if not exactly, even when the input signal is weak or fuzzy. While bottom-up processing is a data-driven approach that helps our brain detect such signals, it happens only at an early stage and is therefore inadequate to explain the above. It is complemented by the conceptually-driven top-down processing, where the brain is able to utilize prior knowledge to examine, interpret, and provide meaning to the input signal (Lindsay and Norman, 2013). Such recall and its application are possible because of prior knowledge, stored in our long-term memory in a structured manner, which is efficiently interconnected and dynamically updated.

This paper examines three broad themes related to organization, interconnectedness, and the dynamic nature of prior knowledge in long-term memory. In the first part of the paper, these topics are discussed through the lens of various relevant theories and models, while in the second part, building on the sub-theme of affordances, SoloHealth - Walmart’s health assessment kiosk has been reviewed.

Long-term Memory and Prior Knowledge

Long-term memory is a repository of knowledge and prior events that can store a huge amount of information (Cowan, 2008). It encodes data semantically and can be used to compare or recall information. Prior knowledge, like all knowledge, is systematically structured. It is stored in our long-term memory and refers to both formal as well as experiential knowledge that a person has acquired over a lifetime.

The importance of prior knowledge for memory was proposed in the works of many scholars. Bartlett (1932) conducted studies that showed that human beings, when remembering a specific event, often create meanings based on their prior understanding of the world. Piaget (1964) argued that children very quickly begin to develop cognitive structures to make sense of everything around them. In an experiment, Bransford and Johnson (1972) observed that participants with prior knowledge had superior comprehension skills and their ability to recall past events from memory was also better. Thus, it can be inferred that prior knowledge acts as a structure or framework to process, categorize, and understand new information (Brod, Werkle-Bergner, & Shing, 2013). To understand how prior knowledge can be efficiently utilized to design better interactions, it is critical to understand its three main characteristics. Prior knowledge is: (1) highly organized; (2) intricately interconnected; and (3) constantly evolving.

Highly Organized

We live in an information-dense world. In order for the brain to process so much information, it should be able to organize the incoming data systematically and rapidly retrieve it from memory. This happens in a matter of milliseconds and is made possible by the fact that all knowledge is structured in the brain in a very efficient manner. Various scholars have provided their perspective on how such an organization happens, and Bartlett’s (1932) schema theory is among the most prominent. This theory describes how people make sense of the world around them (Axelrod, 1973). It introduces the term schema, which is the data structure or “basic building block” (Piaget & Cook, 1952) used to represent “generic concepts stored in memory” (Ortony and Rumelhart, 2017). It is a product of our experiences from a very early age that gets updated and refreshed throughout our lives. A particular schema associated with work process and task analysis is the notion of a script. It talks about how human action in a particular context can be seen as following a certain sequence of events (Schank and Abelson, 2013). Scripts are predictable, highly codified, and follow a step-by-step hierarchical structure.

In 1970s the idea of schema theory got reinforced when it was used by Minsky to propose the idea of frames. He stated that when humans encounter a new development, their minds select a structure called “frame” from memory. According to Minsky (1974), the frame is basically a “data structure” or “remembered framework to be adapted to fit reality by changing details as necessary.” It is the tiniest node of knowledge stored in the mind.

Mental Models

Human beings do not automatically depend on a particular set of rules of inference, but instead draw upon mental models that are based on their understanding and prior knowledge (Johnson-Laird, 1983). According to Olson, Arvai, and Thorp (2011), “Mental models are psychological representations of real or hypothetical situations” that humans use to understand and interact with the world. Although the idea of mental models was first introduced by Craik in 1943, the term gained theoretical popularity with the work of Johnson-Laird (1980). In the field of Human-Computer Interaction (HCI), Don Norman’s (1988) seminal book The Psychology of Everyday Things helped popularize the term.

According to Norman (2014), “mental models are naturally evolving models”. That is, as a person progressively interacts with a system, they keep modifying their mental models. He not only draws distinction between the mental models of a designer and a user, but also compares conceptual and mental models. Norman (2014) states that while conceptual models are “tools for the understanding or teaching of physical systems”, mental models are “what people really have in their heads and what guides their use of things.” Therefore, as a designer, our goal should be to create a conceptual model that complements a user’s mental models. 

Intricately Interconnected

Knowledge is not just highly organized, but it is also intricately interconnected. The human mind is adept at categorizing and chunking related attributes in way that makes recalling and updating information easier. To understand this interconnectedness, scholars have used various network models like propositional and semantic networks. A semantic network is a knowledge representation system that is characterized by relationships involving links or logical connections between nodes. Just like frames (Minsky, 1974), nodes signify ideas or knowledge. These nodes are connected to other nodes and other progressively larger networks. Priming of semantic networks improves the brain’s ability to effectively recall and classify objects.

The Hierarchical Network Model was one of the first semantic network models that was postulated by Collins & Quillian (1969). It suggests that knowledge should be organized hierarchically into two sets. They used the example of a canary bird to explain how information stored in sematic memory is connected by links in a huge network. Another important semantic network model is the Feature-Comparison model postulated by Smith, Shoben, and Rips (1973). It states that there are two types of features: (1) defining and (2) characteristic. These two features are compared to find out how similar they are.

A few years later, the spreading-activation theory was proposed by Collins and Loftus (1975) who stated that “the theory viewed memory search as activation spreading from two or more concept nodes in a semantic network until an intersection was found.” Spreading activation is a waterfall approach to meaning formation as more context is gained. The theory models fan out effect and proposes that as more connections are established between related concepts, the interconnected sematic networks expand and can result in very broad and complex associations (Cohen and Kieldsen, 1987). All these theories propose the intricate interconnectedness of knowledge which enables efficient recall and implementation.

Constantly Evolving

Our mind is constantly processing information as well as adapting to new changes. Inquiries into the dynamic and evolving nature of knowledge has been another important area of focus for scholars when studying the characteristics of prior knowledge. One of the first well-known efforts to examine this question came from the works of Swiss psychologist Jean Piaget. 

Piaget (1964) proposed what can be termed as his Theory of Cognitive Development. He disagreed with the idea of tabula rasa, which points to a child’s mind being like a blank slate; and argued that children very quickly begin to develop cognitive structures to make sense of the world (Piaget, 1964). He proposed two ways in which children organize data or knowledge: (1) assimilation and (2) accommodation. Assimilation is the process that occurs when a person encounters new information and has to fit it into his exiting memory schema or mental model, while accommodation happens when the new information cannot neatly fit into the existing framework and thus some modifications need to be made to assimilate the object (Piaget & Cook, 1952). Rumelhart and Norman (1976) classify acquisition of knowledge into three categories: (1) accretion or encoding new information in terms of existing schema, (2) tuning or slow refinement/ modification a schema, and (3) restructuring or creating new schemata. Since accretion and assimilation both deal with new information, they involve less cognitive effort. However, accommodation and restructuring involve greater degree of modification and hence require more cognitive effort.

Donald Norman has also pointed to the evolving nature of cognitive knowledge structures when he used the idea of mental models to point out that prior knowledge is constantly modified (Norman, 1983). As pointed out earlier, he states that mental models are updated as people encounter new situations. These models need not be exact or accurate so long they are functional.

Affordances

Since restructuring and accommodating can be somewhat difficult for users, it makes sense to keep the concept of affordances in mind when designing an interface. Norman (1988) borrowed the term affordances from Gibson (1979) and presented it in the field of HCI. An affordance is the aspect of an object which offers suggestions or clues to the user’s mind about how to interact with it. When users come across an object, they don’t just take into consideration its features, but they also use prior knowledge to understand how to use it (McGrenere, Ho, 2000). Affordance is offered by either the product itself or its context and it describes perceived possibilities of interaction with the object. These perceived properties, however, may or may not actually exist and can be in the form of unseen clues on usability. However, affordances can be useful in making interfaces easy to use and learn and can suggest ways to improve the usability of new products (Gaver, 1991).

SoloHealth - Health Assessment Kiosk at Walmart: A Design Review

SoloHealth is a free, self-service health assessment kiosk that allows people to quickly check their blood pressure, vision, body mass index, and overall health. An unmanned kiosk, it can be found across various stores like Walmart and Sam’s Club. For the purpose of this study, a SoloHealth kiosk at a Walmart Superstore in Hartford, CT, was reviewed.

The typical user of this kiosk would be a shopper at Walmart, mostly an adult living in the US, who wants to check his/her health for free. He/she would probably be a new user from low-income background who is neither very healthy nor highly educated. 

The kiosk is conveniently located at the end of an aisle in the Pharmacy section of Walmart. A first-time user looking to assess his health would want to find this facility close to the place where medicines are sold. The placement of the kiosk is therefore in accordance with a user’s mental model. Since health is personal for most people, they would expect some privacy while they view the assessment. The kiosk’s placement at the end of an aisle, ensure a user’s privacy and it would make them feel comfortable using it.  

If we look at the overall design and conceptual model of SoloHealth, its shape and look indicate that that it is a kiosk meant for interaction. Similar to an ATM, photo booth, or ticket-vending machine, it has a touch screen digital interface. Since the blue-and-white seat and kiosk are attached to the same base, a user’s mental model would help him/her perceive it as a single unit/system. 

Although interacting with a glass display does not provide an apparent affordance, it matches a user’s mental model of interacting with touch screens on mobile phones, ATMs, and ticket vending machines. 

Most buttons on the touchscreen are clearly visible and of appropriate size. These buttons provide the affordance of pressing/clicking them. The start button on the screen is green and explicitly reads “start”. Additionally, “Touch Start to Begin” communicates the exact action desired from a user. Such affordance can be useful but only for those users who have never used a touchscreen kiosk. The affordance to “touch start” is somewhat too explicit and redundant. This can overwhelm a user and therefore it would be useful to do away with the text “Touch Start to Begin”.

The white chair/seat in the kiosk is similar to the one a person would find at a physician’s clinic. It matches a user’s mental model and provides the affordance of sitting. Seat handles provide support and would be useful for older users and provide assistance while they are trying to stand up or sit down. The shape of handles gives the affordance of being held for support. The smooth edges also ensure better grip and their design is similar to handrails used in staircases. This would match with a user’s mental models.

A user can calculate his/her weight while being comfortably seated on the white seat/chair. While this might conceptually seem like a good idea, it would not really match with a user’s mental model. People measure weight while standing on weighing scales. Sitting on a seat does not provide the affordance of measuring weight. Introducing such a new concept before a novice user would definitely confuse them and only visual or verbal cues would be useful in helping them understand how their weight can be measured even without standing.

The blood pressure sleeve is a good feature and it looks like similar sleeves used in hospitals and clinics. Its conceptual model is in accordance with a regular user’s mental model. However, a user who has never used an unmanned kiosk before, would expect some sort of assistance while calculating his/her blood pressure. Although the digital interface on the touchscreen does provide clear visual and verbal cues on how to use it correctly, for a new user the affordance is not that clear or obvious.

Conclusion

The above exploration and discussion about prior knowledge, its characteristics, and related models holds important implications in the field of HCI. Awareness about how pre-existing ideas and memory structures influence user choice and usability can help create a dynamic knowledge-based model of design. It is crucial to align the mental map and conceptual model of a designer with the mental models of user. In essence, good design should focus on creating interfaces that lower cognitive load and improve user experience.

References

Axelrod, R. (1973). Schema theory: An information processing model of perception and cognition. American political science review, 67(4), 1248-1266.
Bartlett, F. C. (1932). Remembering: A Study in Experimental and Social Psychology. Cambridge: Cambridge University Press.
Bransford, J. D., and Johnson, M. K. (1972). Contextual prerequisites for understanding: some investigations of comprehension and recall. J. Verbal Learn. Verbal Behav. 11, 717–726.
Brod, G. Werkle-Bergner, M., & Shing, Y.L. (2013). The influence of prior knowledge on memory: A developmental cognitive neuroscience perspective. Frontiers in Behavioral Neuroscience. 7, 139.
Cohen, P. R., & Kjeldsen, R. (1987). Information retrieval by constrained spreading activation in semantic networks. Information processing & management, 23(4), 255-268.
Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological review, 82(6), 407.
Collins, A. M., & Quillian, M. R. (1969). Retrieval time from semantic memory. Journal of verbal learning and verbal behavior, 8(2), 240-247.
Cowan, N. (2008). What are the differences between long-term, short-term, and working memory?. Progress in brain research, 169, 323-338.
Craik, K. J. (1943). W. The Nature of Explanation.
Gaver, W. W. (1991, April). Technology affordances. In Proceedings of the SIGCHI conference on Human factors in computing systems (pp. 79-84). ACM.
Gibson, J.J. (1979). The Ecological Approach to Visual Perception. Boston: Houghton Mifflin.
Johnson-Laird, P. N. (1980). Mental models in cognitive science. Cognitive science, 4(1), 71-115.
Lindsay, P. H., & Norman, D. A. (2013). Human information processing: An introduction to psychology. Academic press.
McGrenere, J., & Ho, W. (2000, May). Affordances: Clarifying and evolving a concept. In Graphics interface (Vol. 2000, pp. 179-186).
Minsky, M. (1974). A Framework for Representing Knowledge. MIT-AI Laboratory Memo 306. Retrieved March 24, 2019, from http://hdl.handle.net/1721.1/6089
Norman, D.A. (1988). The Psychology of Everyday Things. New York: Basic Books.
Norman, D. A. (2014). Some observations on mental models. In Mental models (pp. 15-22). Psychology Press.
Ortony, A., & Rumelhart, D. E. (2017). The Representation of Knowledge in Memory 1. In Schooling and the acquisition of knowledge (pp. 99-135). Routledge.
Olson, L., Arvai, J., & Thorp, L. (2011). Mental models research to inform community outreach for a campus recycling program. International Journal of Sustainability in Higher Education, 12(4), 322-337.
Piaget, J., & Cook, M. T. (1952). The origins of intelligence in children. New York, NY: International University Press.
Piaget, J. (1964). Part I: Cognitive development in children: Piaget development and learning. Journal of research in science teaching, 2(3), 176-186.
Rips, L. J., Shoben, E. J., & Smith, E. E. (1973). Semantic distance and the verification of semantic relations. Journal of verbal learning and verbal behavior, 12(1), 1-20.
Rumelhart, D. E., & Norman, D. A. (1976). Accretion, tuning and restructuring: Three modes of learning (No. 7602). California University San Diego La Jolla Center for Human Information Processing.
Schank, R. C., & Abelson, R. P. (2013). Scripts, plans, goals, and understanding: An inquiry into human knowledge structures. Psychology Press.