With our limited attentional resource, humans have evolved to process the world efficiently and effectively. Our memories are great information processors that guide us through life and the decision-making process. Therefore, we should aim to generate designs to be natural and compatible to the information processing systems of all users, so that they may expend as little energy as possible in their interactions with the human-centered interfaces and products. Through the lenses of theories to be discussed—and from the top-down driven world by the prior learning of long-term memory—this review examines the infinite capacity system of the human mind that is highly organized, intricately interconnected, and constantly evolving. These three themes encompass sub-themes that include categorization theory to be explored in the design review of the mega menus on a 501(c)3 nonprofit website, Step Up for Students.

Long-Term Memory

The ability to retrieve information quickly from past experiences helps humans to be functional in their everyday lives. Long-term memory is stored knowledge that allows fast connection and has an infinite capacity to keep every piece of learned information and events acquired over a lifetime (Cowan, 2008). The theoretical cognitive psychologist Endel Tulving proposed three types of long-term memory: semantic memory which allows for the recall of symbols (e.g., words, numbers, colors), episodic consisting of series of time-based events, and procedural being associated with how we do things (Tulving, 1985). Long-term memory is modeled as a networked connection of schemas, frames, framesets, and categories. Newly acquired information can be integrated into the existing structure via assimilation or integration (Lee Wood, et al., 2018; Lefa, 2014; Minsky, 1974 & Johnson-Laird, 1980).

Highly Organized

The highly organized structure of long-term memory allows for the effective management of information stored over a person's lifetime. (Camina, et al., 2017). This highly organized concept of storage is described by three theoretical models: schema, frames, and categories—and the sub-categories of scripts, system conceptual model, and user’s mental model. Schemas and frames provide the structure in which base knowledge is stored and organized (Lefa, 2014; Minsky, 1974 & Anderson, 1983). Schemas, a central precept of French psychologist Jean Piaget's research on child education, are sensory mapping systems that store groups of related subjects and are built through learning processes (Lefa, 2014). They are the base storage and organizational units of long-term memory and help with the interpretation of signals coming in from the world. They can include both linked groupings of knowledge, as well as details on how the knowledge was obtained (Lee Wood, et al., 2018). Individual schemas contain information on a specific subject or event, and schemas related by topic are linked in groups known as schemata. Mapping a real-world event against existing schemas will allows for perception (Lee Wood, et al., 2018). Building on the idea of schemas and believing that perception requires a frame of reference in prior knowledge, computer scientist Marvin Minsky developed the concept of frames (i.e., data-structures), and frame-systems. A frame is a networked entity that includes nodes of previously stored information, and associated attributes (Minsky, 1974). To allow for quick traversal, individual frames are networked into frame-systems, and then into a data retrieval network. Data retrieved from a frame system need only provide a best fit mapping to a signal (Kelemen, 2007 & Minsky, 1974). A category contains similar elements that meet specific conditions for inclusion in that category. Categories allow for the creation of hierarchical relationships between entities (Grossi, et al., 1987). Categories can exist across a single schema, or an entire schemata (Lefa, 2014). A frame can represent either an object or a category, and frames can have relationships to other objects or categories (Minsky, 1975). Categories will be discussed further in the design review.

Scripts, a type of schema, are specialized organizational units that store and describe expected sequences of events around well-known situations (Schank, et al., 1975). Scripts are crucial to navigate certain situations in life as they provide us with the order in which events should proceed. An example of a script is the sequence of events one expects to experience at a restaurant that involves a menu and the processes of ordering, eating, and paying (Albarracin, 2021 & Schank, et al., 1977). As scripts are repeatedly invoked, they contribute the events to episodic memory. Scripts are important design considerations in critical applications.

Scottish psychologist Kenneth Craik laid the groundwork of mental model theory which proposed that as humans encounter new experiences, they store small models of these systems in long-term memory. These stored models provide a mechanism to allow humans to interact with the same, or similar, environments in the future. (Garnam, et al., 1996 & Johnson-Laird, 2010). As existing mental models are evoked, they are tuned and provide a refined interaction model (Hambrick, et al., 2020 & Anderson, 1983). Designers typically build products on the principal of conceptual models (i.e., system models) where they anticipate how users will interact with a system, and care should be taken to map these conceptual models as closely to possible to the expected user mental models, else errors or fatigue can occur (Norman, 1988). Both Minsky's and Piaget's models support the concept of being networked which allows for the schemas and frames in long-term memory to traverse for relevant information retrieval (Brady, et al., 2008).

Intricately Interconnected

Early work by cognitive scientists Allan Collins and Ross Quillan modeled the brains networking system to the semantic networks as described in computer science literature (Sowa, 1984 & Collins, et al., 1975). Computer scientist John Sowa, a pioneer in artificial intelligence and computational networks, depicted knowledge as an interconnected system of nodes and arcs. Nodes can be defined as representing a concept, and attributes of that concept point to other nodes in the network via arcs (Collins, et al., 1975). The strength of the links between nodes defines how critical a particular node is to another (Murphy, et al., 2012). When a particular signal is presented to long-term memory it will traverse the network until it finds a matching pattern in a node. From here it will traverse other strongly connected nodes in what is known as spreading activation and return the most comprehensively available match to the real-world information (Murphy, 2012, Collins, et al., 1975). This will not be an exact match to the system being interacted with, but it can provide a good enough match to allow humans to proceed. This good enough principle is compatible with Piaget's model of not requiring an exact pattern match (Lefa, 2014) on schemas, and also matches Craik's mini-model theory where our brain only stores limited data to match against real world scenes (Anderson, 1983). Efficiency is enhanced as humans scan scenes by the priming effect, where memory retrieval comes faster for an incoming signal that matches closely with the previous signal (Collins, et al., 1975, McNamara, 1992). If too much knowledge is stored in new nodes relative to a given concept, the schemata grows and the brain requires more effort to match a signal to that concept in what is known as the fanning effect, which can lead to slowness in signal matching and errors (Anderson, 1983). An alternative to semantic network theory is Pylyshyn's more abstract concept of propositional networks where the physical objects in semantic networks are replaced by propositions and symbols (Pylyshyn, 1973). Like semantic networks they have interconnected nodes and hierarchies, and new signals allow for the modification of existing schemas, or the addition of new ones. The constant influx of new or modified information requires that the networked schema system has mechanisms in place to allow it to evolve continuously.

Constantly Evolving

The two widely accepted theories of a constantly evolving long-term memory are Piaget's principles of assimilation and accommodation, along with Don Norman’s and David Rumelhart’s concepts on accretion, tuning, and restructuring (Lefa, 2014 & Rumelhart, et al., 1976). Piaget proposed that if newly detected sensory data has a good overlap with an already existing schema, then that information will be assimilated creating a more inclusive model. This corresponds to the accretion model which proposes that new facts pertinent to an existing schema are automatically added to it (Rumelhart, et al., 1976). Accommodation, which is like restructuring, occurs when existing schemas do not provide an appropriate match for new data. This is more time consuming and requires the creation of a new schema, and modification of the semantic network (Rumelhart, et al., 1976). Most theories contend that if incoming information has an acceptable level of match to an existing schema, then assimilation—which requires minimal cognitive effort—will occur. As humans continually interact with—or repeatedly visualize—an existing system, the underlying schemata structure will experience tuning where an existing schema is modified and refined until it aligns more closely with a real-world environment (Hambrick, et al., 2020). Tuning increases expertise in a particular subject which allows users to understand quickly and know how to interact with the same or similar systems (Ericsson, et al., 1995). Building around expected users’ expertise may allow designers to create systems that users will learn quicker. However, small modifications against that expected expertise can lead to confirmation bias, where humans will try to match systems against existing tuned networks, rather than making effort to integrate new design features (Stangor, et al., 2014).

Categorization Theory of the Mega Menus for Step Up for Students

Cognitive scientist and psychologist, Lawrence Barsalou, described the human conceptual system as a collection of categories—where each category relates to part of an experience (Barsalou, 2005). We acquire categories for places, events, products, relations, and other entities through learning. This conceptual and categorical model helps with perception by providing predictions on events and objects that are likely to be encountered as humans interact with the world—with anticipation of what to expect and filling in incomplete data from an overall categorization of related entities and events. As more inferences are gained across the categories, users gain expertise in particular world systems. (Barsalou, 2005 & Loken, et al., 2008). The three notable theories of how categories are described in memory are exemplar, prototype, and rule—all are attained based on personal experiences (Kruschke, 2008). Exemplars are a specific item, e.g., the categorization of mobile phones might be an iPhone. A prototype is a composite of the main features of a category, e.g., the prototype for a cat might include claws, fur, a tail, etc. Rule models define necessary and minimal attributes for membership in a category (Loken, et al., 2008 & Kruschke, 2008). Categories achieve meaning based on the context in which they are perceived (Grossi, et al., 1987). In information architecture of a website, content should be logically grouped into categories that match the users’ mental model and semantic network. All categories within a list should be mutually exclusive. This is a crucial design consideration as categories of objects in the wrong context create design errors that can lead to human errors and disorientations. Intuitive wayfinding is important in site structure.

Figure 1. Prior homepage design of Step Up for Students with the mega menu shown for 'Schools & Providers' tab.

Step Up for Students is a 501(c)3 Scholarship Funding Organization that helps administer the Personal Learning Scholarship Account-PLSA (Special Needs Scholarship) and the Florida Tax Credit Scholarship Program-FTC (Income-Based Scholarship) to Florida schoolchildren. Figure 1 shows an older design from 2022 for homepage, before a redesign, with the ‘Schools & Providers’ tab hovered over to expand the mega menu. The site’s mega menus from the primary navigations were confusing to scan, users would have to click on all categories (that were not completely exclusive with distinct sets of product content or groups of tasks) to understand what the organization was about. Several subcategories were repeated that were redundancies, e.g., multiple logins that went to the same page. This menu structure allowed scanning semantic networks, but it violated the expected mental models. The visual styling for the navigations and menus did not differentiate well between groups of categories, e.g., the text header for the ‘Schools & Providers’ tab were shown in the same black color, but a bit bolder, to emphasize being selected with expanded menu; users would not detect the small differences in bolding due to normalization. There were also issues of unrelated context, e.g., the link for ‘Parent Resource Page’ should be among the categories for ‘For Parents’ instead of being in this menu for ‘Schools & Providers’. This would not match existing schemata, or expected categorization, and the overall structure of the mega menu could cause a fanning effect that also may have caused fatigue as multiple schemata crossed to gain an appropriate description (Figure 1).

Figure 2.  2023 Homepage redesign of Step Up for Students with the mega menu shown for ‘Schools & Providers’ tab.

Figure 2 shows the improved information architecture of the global navigations for Step Up for Students. The mega menu for ‘Schools & Providers’ now distinguishes the thematic categories visually with images to help the users identify and understand key elements quickly; its text header also now shows a different color (red) as a signal for the users that they’re seeing the menu for ‘Schools & Providers’. However, there’s still an issue with scripting when the users click on ‘Login to Your Account’ that takes them to a shared page for all scholarship logins. The ‘Schools & Providers’ representatives will need to scroll down to the bottom of this page in order to find the correct login panel. A specific login page for ‘Schools and Providers’ would match better with the users’ mental model and expectation to skip the search process for quicker access.

Conclusion

From the perspective of categorization and information architecture, a website with high information density and complex organizational structure such as that of Step Up for Students would benefit from understanding how human long-term memory works. Website interaction can be controlled to predict responses and fulfill expectations by activating the users’ memories correctly with the right context. Along this knowledge, Step Up for Student should tailor the user experiences that correspond to the natural abilities of the company’s user groups.

References

Albarracin, M., Constant, A., Friston, K. J., & Ramstead M. J. D. (2021). A variational approach to scripts. Frontiers in Psychology, 7(20). https://www.frontiersin.org/articles/10.3389/fpsyg.2021.585493

Anderson, J. R. (1983). A spreading activation theory of memory. Journal of Verbal Learning and Verbal Behavior, 22(3), 261–295. https://doi.org/10.1016/s0022-5371(83)90201-3

Barsalou, L.W. (2005). Handbook of categorization in cognitive science. Elsevier. https://barsaloulab.org/Online_Articles/2005-Barsalou-chap-situated_conceptualization.pdf

Brady, T. F., Konkle, T., Alvarez, G., & Oliva, A. (2008). Visual long-term memory has a massive storage capacity for object details. PNAS, 105(38), 14325-14329. https://doi.org/10.1073/pnas.0803390105

Collins, A. M. & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological Review, 82(6), 407–428. https://psycnet.apa.org/record/1976-03421-001

Cowan, N. (2008). What are the differences between long-term, short-term, and working memory? Progress in Brain Research, 169, 323-38. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2657600/

Ericsson, K. A. & Kintsch, W. (1995). Long-term working memory. Psychological Review, 102(2), 211–245. https://doi.org/10.1037/0033-295X.102.2.211

Grossi, D., Dignum, F. P. M., Meyer, J. J., Serafini, L., & Bouquet, P. (2005). Context in categorization. Workshop on Context Representation and Reasoning, Paris, France. https://dspace.library.uu.nl/bitstream/handle/1874/19671/dignum_05_context.pdf?sequence=1

Hambrick, D. Z., Macnamara, B. N. & Oswald, F. L. (2020). Is the deliberate practice view defensible? A review of evidence and discussion of issues. Frontiers in Psychology, 8(11), 1134. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7461852

Johnson-Laird, P. N. (2010). Mental models and human reasoning. PNAS, 107(43), 18243-18250. https://doi.org/10.1073/pnas.1012933107

Kelemen, J. (2007). From artificial neural networks to emotion machines with Marvin Minsky. Acta Polytechnica Hungarica, 4(4), 1-12. http://acta.uni-obuda.hu/Kelemen_12.pdf

Kruschke, J. K. (2008). The Cambridge handbook of computational psychology. Cambridge University Press. https://doi.org/10.1017/CBO9780511816772

Lee Wood, M., Stoltz, D. S., Van Ness, J., & Taylor, M. A. (2018). Schemas and Frames. Sociological Theory, 36(3), 244–261. https://www.jstor.org/stable/26541789

Lefa, B. (2014). The Piaget theory of cognitive development: An educational implication. Educational Psychology, 1(9). http://moodle4.f.bg.ac.rs/file.php/326/Implikacije_Pijazeove_teorije_ENG_.pdf

Loken, B., Barsalou, L.W., & Joiner, C. (2008). Handbook of consumer psychology. Lawrence Erlbaum Associates. https://barsaloulab.org/Online_Articles/2008-Loken_et_al-chap-consumer_categorization.pdf

McNamara, T. P. (1992). Priming and constraints it places on theories of memory and retrieval. Psychological Review, 99(4), 650–662. https://doi.org/10.1037/0033-295X.99.4.650

Minsky, M. (1974). A framework for representing knowledge. MIT-AI Laboratory, Memo 306. https://dspace.mit.edu/handle/1721.1/6089

Murphy, G. L., Hampton, J. A., & Milovanovic, G. S. (2012). Semantic memory redux: An experimental test of hierarchical category representation. Journal of Memory Language, 67(4):521-539. doi: 10.1016/j.jml.2012.07.005.

Norman, D. A. (1988). The design of everyday things. The MIT Press. http://kowym.com/wp-content/uploads/2018/08/The-Design-of-Everyday-Things-Don-Norman.pdf

Pylyshyn, Z. W. (1973). What the mind's eye tells the mind's brain: A critique of mental imagery. Psychological Bulletin, 80(1), 1–24. https://doi.org/10.1037/h0034650

Rumelhart, D. E. & Norman, D. A. (1976). Accretion, tuning, and restructuring: Three modes of learning. La Jolla Center for Human Information Processing. doi: 10.21236/ADA030406

Schank, R.C. & Abelson, R.P. (1975). Scripts, plans and knowledge. International Joint Conference on Artificial Intelligence.

Schank, R. C. & Abelson, R. P. (1977). Scripts, plans, goals, and understanding: An inquiry into human knowledge structures. Lawrence Erlbaum.

Sowa, J. F. (1984) Conceptual structures: Information processing in mind and machine. Addison-Wesley, Reading, MA. https://re-dock.org/wp-content/uploads/2012/12/semantic-net-sowa.pdf

Stangor, C. & Walinga, J. (2014). Introduction to psychology (1st ed.). BCcampus Open Education. https://opentextbc.ca/introductiontopsychology

Tulving, E. (1985). Memory and consciousness. Canadian Psychology / Psychologie Canadienne, 26(1), 1–12. https://doi.org/10.1037/h0080017