More than two decades ago, when computational science (an interdisciplinary practice incorporating modeling, simulation, visualization, and problem solving) emerged as a new workforce strategy for institutions of higher education (IHEs) and as an innovative teaching pedagogy for K-12, many had hoped that it would revolutionize the STEM education. As such, in 1998, SUNY College at Brockport launched the nation’s first undergraduate degree program in computational science (Yaşar et al., 2000; Yaşar and Landau, 2003; Turner et al., 2011). In 2006, Jeannette Wing, an influential computer scientist and an assistant director at the US National Science Foundation (NSF), mobilized significant NSF resources, rebranded computational science as computational thinking (CT), and claimed in her 2006 essay that CT should be taught as a fundamental skill in public schools just like reading and writing (Wing, 2006). The notion of teaching computational thinking (CT) as a fundamental competency seems to have inspired many educators and researchers worldwide. However, teaching experts’ habits of practice to novices is inherently problematic because of prerequisite content knowledge and practice skills needed to engage in the same thinking processes (Kirschner et al., 2006), not to mention the cost of providing them a similar environment to conduct inquiry and design. A remedy has been suggested to link experts’ habits of practice to fundamental cognitive processes so we can narrow their skillsets down to more basic competencies that can be taught to young students.
Linking computation to cognition is not a new idea—in fact, it goes as far back as to the time of human computers during Babylonians (Denning and Tedre, 2019). Obviously, after the electronic computer age began in the 1940s, the term “computer” has often referred to electronic devices rather than human agents. What led to the design of electronic computing 80 years ago in the first place was that if thoughts (i.e., information) can be broken up into simple quantifiable constructs and algorithmic steps, then machines can add, subtract, or rearrange them as our brains do (Turing, 1936). The human brain employs a distributed network of neurons to rearrange information (Hebb, 1949). As such, information is stored into the memory via a specific pattern of neurons placed on a pathway and fired together. Arrival of new information lights up all related cues, neurons, and pathways in a distributive process that is similar to the top-down action in Figure 1, whereby a new concept is broken up into related pieces. The converse, retrieving information, involves reassembly of the original pattern of neurons and pathways in an associative process similar to the bottom-up action in Figure 1. Retrieval, in other words, is not an act of merely recalling facts and figures. It is a process of reassembly involving different pathways that are linked to one’s knowledge. What is retrieved is not a carbon copy of the original but a re-imagined copy of the original with some holes and/or extra bits. Neuroscientists see little or no distinction between the acts of information storage/retrieval and the act of (computational) thinking (Montague, 2006; Brown et al., 2014).
Our brain’s inclination to process information in an associative and distributive fashion, as well as to store and retrieve memories and concepts in a scatter and gather fashion by a distributed neural network, appears to be a manifestation of a basic duality engrained in the fabric of matter and information. Quantifiable things appear to behave in one of only two ways (as in Figure 1): they either unite associatively to form bigger constructs or break down distributively to smaller ones. Such a duality at the core of information processing by a computational mind carries itself up to higher-level cognitive processes, such as deductive reasoning in the form of distributive processing of information and inductive reasoning in the form of associative processing of information (Dunbar and Klahr, 2012; Yaşar, 2017, 2018).
We are all naturally inclined to employ inductive thinking and deductive thinking in everyday life. They are the two major cognitive competencies at the root of the CT skillset (Wing, 2006; Yaşar et al., 2016; Yaşar, 2018; Denning and Tedre, 2019; Mills et al., 2021), which are often cited as abstraction and decomposition skills. We all employ computational thinking by the virtue of having a computational mind. However, when used together in certain ways, the combination of inductive and deductive thinking becomes a much more powerful skill, as first described by Kant (1787) more than two centuries ago. For example, through iterative and cyclical use of inductive and deductive thinking, as depicted simplistically in Figure 1, does the conceptual change occur in our learning progression, all the way from childhood to the adulthood (Carey, 1985). Conceptual change is also at the heart of the scientific thinking (Vosniadou, 2013) both at the level of an individual scientist, or those who think like scientists, as well as that of the scientific progress by the scientific community (Kuhn, 1962; Thagard, 1999). Not surprisingly, imaging techniques have revealed that scientific thinking is not just thinking about the content (of sciences); it encompasses a set of cognitive processes, such as conceptual change, that transcend the field of science (Dunbar and Klahr, 2012). These processes include (a) problem solving, (b) design and modelling, (c) hypothesis testing, (d) concept formation, (e) conceptual change, and (f) reasoning (inductive, deductive, abductive, causal, and analogical thinking). According to Thagard (2012), these ST processes are no different from those employed in everyday living by non-scientists—the difference comes from how they are used. In a sense, what distinguishes ST from everyday thinking (i.e., computational thinking) is that while CT involves any use of inductive and deductive thinking, ST involves iterative and cyclical use of these two opposite reasoning skills to accomplish conceptual change and other ST skills listed above (Yaşar, 2021).
A great deal of efforts has gone into analyzing CT as a result of recent technological advancements which have affected our professional and personal lives. These efforts include definition of CT (Papert, 1980; Wing, 2006; Guzdial, 2008; Denning, 2009; Aho, 2012), its cognitive essence (Yaşar, 2017, 2018) and manifestations in different fields and ways to teach it at different levels of education (Denning, 2017a,b; Yadav et al., 2017; Denning and Tedre, 2019; Tedre and Denning, 2021). For a literature review, see Grover and Pea (2013); Angeli and Giannakos (2019); Denning and Tedre (2019); Kakavas and Ugolini (2019) and Saqr et al. (2021). In the 1990s, the focus was on literacy and fluency issues with a push to teach programming. The arrival of easy-to-use M&S tools, which hid the underlying mathematics and programming, allowed a new way of studying scientific phenomena and teaching CS principles in the 2000s. The present decade has seen even easier tools, such as mobile apps, to support children’s computational thinking and literacy skills (Papadakis, 2021).
Today, there are plenty of tools available for teaching various CT skills. However, the discourse on what it means to different stakeholders continues to this date. Some have suggested to categorize it as “CT for beginners” and “CT for professionals.” The same argument applies to teaching of ST skills. There is a need for innovative practices to provide continuity in CT and ST education all the way from elementary to post-secondary. We posit that an information processing approach to cognition, as briefly explained above, allows us to teach core CT/ST cognitive competencies with appropriate grade-level challenges and skills. If indeed the acts of information storage and retrieval strongly correlate to the act of computational and scientific thinking, then all we need to do is to strengthen those information processes. Whatever practices we come up with to strengthen them, one way to measuring their effectiveness could be through the act of information retrieval itself. We are lucky in that sense because long before such correlation was made, researchers in cognitive psychology had been studying the impact of memory retrieval practices on knowledge retention and other cognitive functions as explained in the next section. This article establishes ground that retrieval practices can be used as a way of strengthening CT and ST skills. We hope that the findings from our professional development program and related action research by participating teachers will shed a light on the discourse about CT and ST education. While the practitioners would benefit from reproducing similar results from a tested and scalable strategy, the researchers could expand their efficacy studies, via retrieval practices, to the teaching of more basic CT and scientific thinking (ST) concepts at a variety of grades. An approach such as the retrieval practice, which causes learning to stick and promotes core CT and ST skills, could have a broad impact in STEM education.