Long Term Memory

​

Our abstract concepts are based on our interactions with the physical world. As we experience new things information gets stored in long-term memory adding to our existing set of knowledge. Memory can be divided into episodic (Tulving, 1993), involving the collection of personal experiences related to the past (Clayton & Dickinson, 1998) and semantic, allowing the ability to recall general facts, associated with meaning (Schacter et al, 2007). To understand how prior knowledge can be applied to the domain of interaction design, the three main characteristics of prior knowledge must be understood. It is structured in the following three ways, (1) Highly organized, (2) Highly interconnected, and (3) Constantly evolving.

​

1. Highly Organized: Scholars provided various theories and models on how knowledge is organized in the mind, among them schema theory is one of the most prominent. The term schema was first coined by Bartlett (Bartlett, 1932), based on his research evolving around the narrations of folklores, he observed that participants only remember the gist of stories and they tend to recall it by adding their own details.

 

Based on this feature of memory, he proposed the schema theory, defining schemas as ‘building blocks of cognition’ and the smallest unit of knowledge is a concept referred to as schemata’(Rumelhart, 1984). In an event, experienced by the sensory system, the low-level schemata get activated which in turn activates its interconnected higher-level schemata. The corresponding high-level schemata then activate the ‘conceptually driven ‘sub schemata that were not already activated to check ‘goodness of fit’(Norman & Rumelhart, 1975). Thus, a top-down network activates and flows down from its sub schemata to sensory schemata. “The configuration of this schemata forms the basis of our understanding” (Rumelhart, 1984).

 

The major functions of schemata are perception, comprehension, encoding, and remembering (Tulving & Thomson, 1973). Thus, a schema represents our knowledge at all levels of abstractions and is responsible for our ideologies & cultural truths. Schemas for routine activities can be defined as scripts (Bower et al, 1979), represented in the memory as ‘hierarchical and organized information packets’ (Schank & Abelson, 1977) (Abbott et al., 1985). Scripts are knowledge structures that contain information about the actions in an event as well as the sequence of these actions evolved over the repeated experience of a situation. In 1974, Minksy extended the schema theory by proposing the idea of frames, “a frame is a data-structure for representing a stereotyped situation”(Minksy, 1974), having some default values dictating its use and expectations. In the visual world, continuity in frames depends on the confirmation of expectations based on prior knowledge (Minksy,1974). For designers, the applicability of frame theory includes developing interactions as a continuity of experiences, confirming the user’s expectations.

​

Mental Models: The term mental model was coined by Johnson-Laird in the theoretical works of cognitive science (Laird, 1980) however, the concept was first introduced by Craik in 1943. Mental models are broad conceptualizations of a person’s thought process about how things work in the world, he/she ‘translate external events into internal models (Craik, 1943).' They are created at the moment, as people form psychological “internal representations of themselves and the objects with which they interact” (Norman, 1983). The term was popularized in the file of HCI through the works of Don Norman. Study on mental models (Laird, 2010) suggests that higher the number of models a person needs to consider while making an inference, the harder it is, leading to error. Drawing a comparison between an expert and a novice user, an expert has a coherent model, works by applying trail-and error strategies whereas a novice is unable to rapidly access concepts at issue and is ‘paralyzed by his/her inability to solve’ problems (Norcio, 1993), leading to abandoning the situation.

 

Therefore, as designers, our goal should be to understand the user’s mental model and align the conceptual models of a system using ‘analogies or metaphors to function as tools of thought to structure unfamiliar domains’ (Gentner & Gentner 1983) and to create ‘intuitive’ schemas & scripts, connecting related concepts. Thus, defining a predictable experience for the user that minimizes error and the need to recall.

​

​

2. Highly Interconnected: In order to understand the interconnectedness capability of the mind, scholars proposed various models such as the propositional and semantic networks.  A semantic network can be expressed as a ‘web of interconnected elements of meaning’ (Collins & Quillian, 1969). Different concepts are stored as nodes and are connected based on their relationships with each other. The ‘Hierarchical Network Model’ was introduced by Collins & Quillian, 1969, based on a seminal study where participants were given parts of a sentence and asked to verify statements with related concepts, as the conceptual category of predicate of the sentence became more hierarchical remote from the subject, people took longer and found it difficult to verify a statement.

 

Thus, the human mind is able to organize knowledge, like a tree diagram, and the information known about the higher elements in the hierarchy is applied to all the lower elements, defined as the concept of ‘inheritance’ (Sternberg, 2009). Thus, recalling an element can be related to the frequency and the recency of association. Later in 1975, Collins and Loftus, proposed a modified version of the hierarchical network. It expressed that all the ideas in the mind are connected, it also signified that in the attempt to remember one concept all the linked concepts are ‘activated’, this is referred as ‘Spreading-activation theory’(Collins & Loftus, 1975). Thus in memory search, as activation spreads from the ‘primed’ node in the semantic network travels till an intersection is found.

 

Models of ‘storing semantic information were further proposed to store information in a computer memory’(Quillian, 1962). Researchers proposed various models of information processing, namely ACT(Anderson, 1976) and PDP(Rumelhart & McClelland, 1986), they suggest that the mind can process information as serial information processing channel in semantic networks or as parallel channels of words, sounds, and images, explaining the speed and accuracy of human information processing (Pinker, 1988). The effects of parallel processing are explored by the ‘Stroop effect’ (MacLeod, 1991). Additionally, the categorization theory (Barsalou, 1992) defined the process of categorization in semantics. It states that firstly when a search is conducted, first, the mind forms a description of the entity, then deploys searches for categories such as structural features, then selects the most similar category, draws an inference about the entity, and finally stores this information about the categorization (Barsalou, 1992). Thus, designers must use words similar to user’s domain knowledge utilizing mind’s ability to activate a chain of related concepts.

​

3. Constantly Evolving: As we get introduced to new information day by day, the mind constantly processes and adapts it. Various researchers attempted to study these dynamics and the evolving nature of knowledge, particularly relevant is the work of Jean Piaget (Piaget,1964). In his ‘Theory of Cognitive Development,’ he hypothesized that children are born with schemas ‘reflexes’ operating at birth with which they adapt to the environment (Piaget, 1971).

 

Piaget suggested that children sort their knowledge by groupings referred to as schemas and organize data in two ways,

(1) Assimilation and

(2) Accommodation (Huitt & Hummel, 2003).

‘Assimilation’ is the process of acquiring new information by placing it into the existing schema whereas ‘Accommodation’ is the process of modifying an existing schema, when a new set of information does not match the existing framework (Huitt, W., & Hummel, J. 2003).

 

Rumelhart & Norman, 1976 (Spiro et al, 2017) also explained the model of learning and acquisition of knowledge, it follows three ways:

(1) ‘Accretion’, is the mode of learning in which new data structures are added to the existing schemata

(2) ‘Tuning’, occurs when new information has minor changes but is consistent with the basic organization of the schemata

(3) ‘Restructuring’, occurs when new information does not relate to an existing schema, (Rumelhart & Norman, 1976) a brand new schema is created.

 

Don Norman supported the notion of evolving nature of knowledge through the idea of mental models that are constantly being modified (Norman,1983)(Norcio,1993). Thus, for an interaction designer, designs that require the user to learn new information by restructuring, accommodation will cause cognitive load, as compared to an intuitive experience based on information utilizing the prior knowledge of the user, through accretion, tuning, and assimilation.

​

Metaphors

As accretion and assimilation allow users to quickly adapt to a situation, it is logical to use metaphors, that are an integral part of our language and thought, in interface design (Erickson, 1995). Interface metaphors use visuals to explore the familiar knowledge of uses and equip the user to interact with the system by drawing parallel to physical entities. (Richards et al., 1994). Metaphors are used to increase the effectiveness of use, learnability, and to improve memorability (Hollan et al., 2000). Metaphors can provide guidance and act as a starting point to use a system (Carroll et al., 1988). They provide motivation for users, promote learning the domain and can also be used as tools to articulate mental models.

 

Metaphors should have the following properties (Wang & Huang, 2000):

(1) They must be created from the user’s knowledge domains

(2) The operational of the metaphorical components should be the same as their source

(3) The metaphors used in system design should be the same as the conceptual domain.

 

Thus, for interface design, the use of metaphors as icons can be leveraged to allow the user to step beyond the limits of semantic constraints and create an intuitive experience.

​

Design Review: Bhuvan- An Earth Observation Visualization Utility

Bhuvan is an Indian Earth Observation Visualization utility, an open-access database, that allows users to access government data on land planning and geography. The user profile considered is a novice, a student seeking information on urban planning. He/ she would be having some domain knowledge about the subject matter but is not an expert.

​