Cognitive load theory and learning
Since the 1980’s, cognitive load theory (CLT; Sweller & Chandler, 1991) has been established as one of the most applied theories for considering relationships between instructional design and mental problem-solving resources. The theory is concerned with defining the overall mental effort (cognitive load) attributed to working memory resources delegated to accomplishing a task (Kalyuga & Liu, 2015). Cognitive load (CL) comprises three sub-types (van Merriënboer & Sweller, 2005), namely intrinsic cognitive load (ICL), extraneous cognitive load (ECL), and germane cognitive load (GCL).
Intrinsic cognitive load (ICL) is a result of cognitive activities needed for understanding the information inherent in a task and depends largely on the complexity of this information. A high level of ICL is caused by a large number of “task elements” which must be processed simultaneously during learning and also determined by the level of the learner’s knowledge (Sweller, 2010). Extraneous cognitive load (ECL) is experienced by learners when they are forced to invest their cognitive resources in activities that are not immediately relevant to the learning task at hand. The main source of ECL is non-optimal or flawed instructional design, such as an unnecessary complex layout of a digital learning interface (Klepsch et al., 2017). The final sub-type, germane cognitive load (GCL), results from constructing schemas (Sweller et al., 1998) or mental models (Paas et al., 2004) during meaningful learning processes. An example of an activity that can increase GCL is integrating new information with knowledge the learner already has. Therefore, high levels of GCL can be interpreted as a sign of meaningful learning (Klepsch et al., 2017).
From a CLT perspective, learning is defined as constructing and automating schemas in long-term memory (Paas et al., 2004), which involves all three sub-types of cognitive load to some extent or other. To optimise learning, the sum of the load types should not exceed the learner’s limited working memory capacity. Hence, assuming that ICL is inherent in the nature of the task, optimising learning should focus on minimising ECL and increasing GCL (see Klepsch et al., 2017). Although the cognitive load construct provides insight into the usefulness and effectiveness of learning with new educational technologies (e.g., Kalyuga & Liu, 2015), it remains challenging to measure (e.g., Klepsch et al., 2017). The literature contains multiple cognitive load measures that have emerged over time, which range from (the most widely-used) self-rating techniques to recent physiological measures (e.g. pupillary responses). From an instructional design standpoint, it is crucial to deduce ways to measure CL differentially by exploring the relative impact of all three load sub-types during learning (Ibili & Billinghurst, 2019; Klepsch et al., 2017; Mutlu-Bayraktar et al., 2019).
Interactive educational technology and cognitive load
Technology enhanced learning is becoming more and more apparent in higher education bringing with it both hopes and challenges. Amongst the hopes, digital resources that integrate interventions such as AI may help support the learning of complex scientific knowledge such as biology (e.g., Corbett et al., 2010). Interactive technology can also help students to learn more efficiently by offering multimedia resources, interactive glossaries, prompts, answers to questions, help in constructing models and even personalised suggestions for further learning (Aleven et al., 2003; Koć-Januchta et al., 2020; Linn et al., 2014). Nevertheless, new technological opportunities for learning are also associated with multiple challenges. From a motivational perspective, students may experience decreased motivation when learning on their own from a digital learning environment (e.g., DeVore et al., 2017). Consequently, learning with digital tools often requires advanced skills in independent learning, self-regulation and learning strategies (Glover et al., 2016; Means et al., 2009). Additionally, learners may experience cognitive overload when learning with digital technology (Aleven et al., 2003).
As elucidated previously, cognitive overload is often caused by high levels of extraneous cognitive load (Klepsch et al., 2017). While ICL and GCL concern processing of learning elements and promoting meaningful learning, respectively, extraneous load arises mainly from the way information is conveyed. Poorly designed digital learning tools may increase ECL to such an extent that it impairs learning (Moreno & Mayer, 2007). Therefore, where possible, ECL should be reduced by optimising the design of the learning environment. Design elements such as the range and complexity of implemented digital features must be carefully considered, since a complicated interface can render cognitive overload (Scheiter & Gerjets, 2007; Sweller et al., 1998). Lowering extraneous cognitive load frees the availability of cognitive resources that can be directed to germane load and thus stimulate deeper learning (Klepsch et al., 2017). As part of our previous work (Koć-Januchta et al., 2020), we presented students’ opinions suggesting that ECL may increase over time when using an AI-enriched book. As part of that study, where we compared an AI-enriched book and a traditional E-book, students from a research university were interviewed about their experiences of using both types of digital book. They pointed out several advantages of the AI-enriched book over the traditional E-book (e.g., obtaining pop-up definitions to terms in real-time) but also reported growing dissatisfaction with the AI-enriched book as usage time progressed. We observed that a longer usage of the book revealed potential design-related disadvantages (e.g., AI-functionalities were sometimes confusing or contained too much information, or made one unsure of their learning). At the same time, the more difficulty students perceived learning with the AI-enriched book, the less positively they assessed its usability (Koć-Januchta et al., 2020).
Usability perception and cognitive load in educational technology
Usability is an essential measure when exploring user experiences of digital educational technologies (Diefenbach et al., 2014). The concept includes subjective and objective components, which consist of perceived usability or satisfaction (how comfortable it is to use a digital tool) and efficiency (the time and effort cost in using the digital tool), respectively (Lewis, 2018). One of the most popular measures of perceived usability is the System Usability Scale (SUS) questionnaire developed by Brooke (1996). The SUS is suitable for measuring satisfaction (e.g., meeting expectations) and ease of using the learning tool.
Many studies show that perceiving a learning system as useful is associated with a reduced cognitive load (e.g., Pantano et al., 2017), whereas feeling confused when using a system leads to increased cognitive load (Kılıç, 2007). Moreover, Costley and Lange (2017) found that an increase in users’ intention to use a tool is influenced by effective instructional design. Additionally, optimal instructional design correlates with increased germane load indicating deep learning (Costley & Lange, 2017). Notably, Ibili and Billinghurst (2019) have stated that perceived usefulness (perceiving a learning tool as improving learning) and perceived ease of use (perceiving a tool as easy to learn with) were strongly correlated with all three types of cognitive load (ICL, GCL and ECL). Specifically, usefulness was negatively correlated with ICL (for females) and with ECL (for males). At the same time, both usefulness and ease of use were strongly positively correlated with germane cognitive load (Ibili & Billinghurst, 2019). Lastly, cognitive load is strongly connected with self-regulated learning. For example, high cognitive load might originate from students’ insufficient self-control skills and low willingness to learn (de Bruin et al., 2020; Eitel et al., 2020).
Self-regulated learning and cognitive strategies
Acquiring self-regulation skills is important for learning and a research topic of high interest when it comes to individual learning with digital tools (Steffens, 2006). Zimmerman (2011) relates self-regulated learning to the degree to which learners participate actively in their own learning at the metacognitive, motivational, and behavioral level. In addition, Paris and Winograd (2003) describe self-regulated learning as a process in which learners approach problems, apply strategies, monitor their performance, and assess the results of their efforts. Self-regulated learners are more likely to improve their academic achievements by selecting and controlling cognitive processes involved in learning (Pintrich & De Groot, 1990). To learn deeply, one should be able to elaborate and organise information and monitor one’s learning process (Pintrich & De Groot, 1990; Soenens et al., 2012). In this regard, technology-enhanced learning environments offer an opportunity to support self-regulated learning by helping students to plan, monitor and evaluate the cognitive, motivational, and affective components of their own learning (Steffens, 2006).
Learning biology: conceptual knowledge and cognitive skills
Biology is a natural science concerned with studying structures and processes associated with living organisms (e.g., Sadava et al., 2017). Learning biology involves building a conceptual understanding of the structure of the (bio)molecules of life that include proteins, enzymes, carbohydrates, lipids, and nucleic acids. This learning also includes developing core knowledge about the “unit of life” (the cell) and the plethora of cellular processes such as DNA replication, mitosis, meiosis, and gene expression. In turn, such knowledge must be integrated with understanding physiological functions such as photosynthesis, muscle contraction, neural and endocrine control. Furthermore, all these aspects of biology contribute to understanding populations, ecosystems, and evolution. Moreover, constructing biological knowledge involves making links to other scientific disciplines and reasoning at various levels of spatial and temporal scale.
Cognitive skills associated with successful biology learning include: retaining biological knowledge, integrating knowledge with other concepts (while transitioning different levels of biological organisation), transferring learnt knowledge to novel tasks, as well as reasoning both “locally” and “globally” about a biological concept (Anderson & Schönborn, 2008).
Aims of the study
The objectives of this study are to investigate:
-
1.
Any differences in how students experience the three types of cognitive load (ICL, GCL and ECL) while learning with an AI-enriched biology book at the beginning and the end of the study.
-
2.
Relationships between the three types of cognitive load, usability, self-regulation, cognitive strategy use, and learning gain while interacting with the AI-enriched book.