Integrating Computation into Science Education

INTRODUCTION

Increasingly, facility with computation is being positioned as foundational for “all students” learning science, not just for future computer scientists. At the K-12 level, computational practices are now explicitly listed in national science standards (National Research Council, 2012; NGSS Lead States, 2013). In higher education, experts in scientific disciplines, including biology, are calling for college-level instruction to keep pace with the use of computation in scientific research (e.g., Rubinstein and Chor, 2014). These conversations are about much more than students learning how to write code. Advocates of integrating computation into science classrooms use terms such as computational thinking, computational practices, and computational literacies to emphasize how computational tools have the potential to enrich students’ scientific learning as well as to engage them in questions about the role of computation in science and in society.

The articles in this Current Insights begin to bring clarity to these new terms while offering different perspectives on how and why educators might integrate computation into science education. The first two articles start from the premise that computational tools have transformed and will continue to shape how scientists think, work, and communicate. Therefore, there is a pressing need to integrate computation into scientific curricula and study its impact. The third article argues that with this surging interest should come critical considerations of the purpose and impact of integrating computation into science learning environments. These authors call on education researchers to broaden conceptualizations of computational learning beyond basic knowledge and skills. As a set, these articles can help frame conversations about the purpose and place of computation in science curricula that will be important to consider as educators begin translating these calls to action into classrooms and as researchers begin to study such efforts.

COMPUTATION AS A CORE DISCIPLINARY LITERACY

Odden, T. O. B., Lockwood, E., & Caballero, M. D. (2019). Physics computational literacy: An exploratory case study using computational essays. Physical Review Physics Education Research, 15(2), 20152. https://doi.org/10.1103/PhysRevPhysEducRes.15.020152

Like facility with language or mathematics, Odden and colleagues argue that computational literacy is foundational to science. Building on a framework presented by diSessa (2000), Odden and coauthors decompose computational literacy into three pillars that are intended to mirror core aspects of linguistic or mathematical literacies. The material pillar concerns basic fluency with the representational systems of computing—the ability to write, read, and otherwise manipulate code. The cognitive pillar describes the abilities to think with and about computational tools. And the social pillar entails creating and using computational artifacts to communicate with others in the community.

Odden and colleagues used this framework to identify and illustrate components of computational literacy in the context of undergraduate physics education. Specifically, they studied a physics program at the University of Oslo where computation has long been integrated into the curriculum. Students in this program begin learning programming in their first semester and routinely use computational methods to solve problems in their physics coursework throughout the program.

In 2018, the authors were part of a pilot project to introduce computational essays into the curriculum. In writing a computational essay, the students are asked to present an argument supported by blocks of code that are integrated into the text. Students’ essays provided the researchers with artifacts that could make components of their developing computational literacy practices visible.

In this exploratory study, Odden and colleagues recruited a group of 17 students (14 men and 3 women) who agreed to write computational essays to document results from a 4- to 6-week open-ended project that was part of their physics course. Researchers conducted interviews with 15 students about their projects and essays and then conducted a thematic analysis guided by the three pillars framework.

In the results, the authors further decompose each pillar of computational literacy into component practices (skills), knowledge, and beliefs. For example, the material pillar includes basic coding practices, knowledge about coding (e.g., when to use loops), and beliefs about what constitutes good coding practices. They illustrate each component with examples from interviews. Odden and coauthors clarify that while their analysis dissects computational literacy into component parts, in practice they are intertwined.

In the end, the authors make two main claims. First, they argue that their ability to identify examples of each pillar supports the use of computational essays as scaffolds that can help students develop and integrate the components of physics computational literacy (especially when coupled with an open-ended project). Second, they suggest that conceptualizing computational literacy as composed of pillars and components can function as a starting point for conversations about what aspects of computational literacy should be supported in physics and other scientific disciplines. For example, the authors note that this framework helps reveal that less attention has been paid to the social pillar in most curricula—a gap that the computational essay begins to address.

Thus, this article introduces a potential starting definition of computational literacy in the sciences, one that should be considered open to revision, expansion, and critique as different communities consider how to translate the broad goal of computational literacy into specific learning outcomes and supports for students.

THINKING WITH AN AGENT-BASED COMPUTATIONAL TOOL

Arastoopour Irgens, G., Dabholkar, S., Bain, C., Woods, P., Hall, K., Swanson, H., Horn, M., & Wilensky, U. (2020). Modeling and measuring high school students’ computational thinking practices in science. Journal of Science Education and Technology, 29(1), 137–161. https://doi.org/10.1007/s10956-020-09811-1

Arastoopour Irgens and coauthors organize their work around “computational thinking practices (CT practices),” a group of related practices including data practices, modeling and simulation practices, computational problem-solving practices, and systems thinking practices (Weintrop et al., 2016). The CT practices framework, much like the computational literacy framework, is intended to describe how computation functions within scientific disciplines to support new ways of thinking, solving problems, and expressing ideas.

In this study, high school students participated in a 10-day biology unit called From Ecosystems to Speciation that integrated agent-based computational tools into a series of lessons about ecology and evolution. One of the specific affordances of agent-based models is that they can help students think about the relationship between programmed rules at the level of individual agents (e.g., moose or wolves) and emergent patterns at the population level. In this unit, the students proposed programmable rules for biological agents and examined relationships between agent-level interactions and ecosystem-level patterns. Throughout the unit, the students answered embedded assessment questions designed to provoke thinking with and about the computational tool.

To measure the impact of the unit on students’ computational thinking practices, the researchers designed a pre-/postperformance assessment that asked students to interact with a novel agent–based model, interpret output, identify simplifications, and reflect on uses of the model. Responses to the pre-/posttasks were scored for the number of CT competencies present, such as explaining causal links between input and output or identifying missing elements in the model. On average, scores for the 41 students enrolled in the study increased from pre to post, but changes in scores ranged from −4 to +4 for individuals.

To examine this variation, the researchers split students into two groups: one comprised students who made positive gains (n = 20) and one comprised students who made negative or zero gains (n = 21) on the performance assessment. To better understand differences between the groups, the researchers compared how each group responded to embedded assessments. Using a combination of top-down and bottom-up coding, they first identified discourse elements related to aspects of computational thinking the unit was intended to support, for example, referencing agent-level actions or providing a justification to explain observed or predicted population-level patterns. They then used epistemic network analysis (ENA) to measure and visualize the strength of co-occurrences among discourse elements as network graphs for individuals and groups of students.

Overall, the results of the ENA analysis suggest that differences in how students engaged with the agent-based modeling tools helped explain differences in pre-/postscores. Students with positive gains used the computational tools to test out their ideas and wrote explanations that connected population-level patterns and agent-level actions. Students whose scores did not change or decreased tended to focus more on identifying agent-level behaviors and provided fewer explanations for changes at the population level. Thus, one contribution of this work is the use of ENA to provide an explanation for variation in scores that is connected to how students used the tools.

A larger question left open by this study is why student engagement was so variable. Why did some students seem to embrace the tool as a way to explore their ideas about ecosystems, while others seemed less engaged? The authors suggest that more instructor intervention or additional curricular supports could help students focus on the unit goals, for example, by directing them to attend to connections across levels. Although such supports might encourage students to focus on the intended goals of the unit, it is not clear that they would impact students’ motivation or interest in computation. Both this study and the previous study presume that students will find the uses of computation presented to them in their science courses meaningful and worthwhile. The next article argues that questions about how students understand the value of computation for themselves and for society are important for education researchers and curriculum designers to consider.

COMPUTATION FOR WHAT AND FOR WHOM?

Kafai, Y. B., & Proctor, C. (2021). A revaluation of computational thinking in K-12 education: Moving toward computational literacies. Educational Researcher, 51(2), 146–151. https://doi.org/10.3102/0013189×211057904

Kafai and Proctor argue that definitions of computational thinking, practices, and literacies need to be broadened. In addition to learning computational skills and knowledge, students should have opportunities to critically consider “inequities caused or exacerbated by the societal impact of computing” as well as to imagine its “socially responsible uses.” In this essay, the authors build on Sfard (1998) to articulate three theoretical perspectives that are often used to frame arguments about the purpose and goals of integrating computation into K-12 learning environments.

Cognitive framings foreground the importance of individual students acquiring the necessary knowledge and skills to do computing. Such framings tend to emphasize the importance of computation for future careers, particularly in the Science, Technology, Engineering, and Mathematics (STEM) disciplines.

Situated framings emphasize that learning about computation entails becoming part of a community. These approaches ask questions about the ways in which students develop identities and affinities for computation while acknowledging that opportunities to develop such interests have not been equitable. From a situated perspective, computational literacy extends beyond learning the rules of how to communicate computationally to include how students use computational tools to express their own ideas or generate artifacts for the communities to which they belong.

Critical framings expand attention to interrogating the role of computation in shaping society, including how computation functions to reinforce systems of oppression and how it can be used as a tool for social good. Critical computational literacy would include, for example, understanding how biases become embedded into computational tools, interrogating whether corporate interests align with promoting computing in schools, and raising questions about who ultimately benefits when computation is positioned as foundational to STEM learning.

Given these multiple possible framings, Kafai and Proctor suggest the need to be aware of multiple computational literacies, which they define as “a set of practices situated in a sociocultural context which utilize external computational media to support cognition and communication.” They suggest that this view leads to three considerations that should shape future priorities for integrating computation into educational environments.

1. For whom is computation in classrooms currently designed? Do the current calls to integrate computation into education serve all students? What needs to change to achieve this goal?

2. What are currently valued as examples of computational thinking or practices? How narrowly or broadly is computation defined? Is there room for these definitions to change, and if so, which communities will have a voice in shaping those changes?

3. How should computation be taught? Should it be integrated into disciplinary learning? What motivations of students can be tapped to do so? How much say will students have in how they use computation and to what ends?

Kafai and Proctor end by acknowledging that there is much work to be done, particularly to develop “transformative pedagogies” and professional development that can support instructors to address the political and ethical dimensions of computation. They offer these three framings to help orient researchers to their own and others’ motivations and to encourage conversations as the purposes of integrating computation into education continue to evolve.

IMPLICATIONS FOR INTEGRATING COMPUTATION INTO COLLEGE-LEVEL SCIENCE

For those interested in bringing computation into college-level science classrooms, this set of articles provides some practical examples. As Odden and colleagues describe, the computational essay may be a useful tool for introducing college-level students to computational tools and practices. The work of Arastoopour and colleagues highlights the potential for agent-based modeling tools, which have been primarily used in K-12 contexts, to support thinking about and investigating complex biological systems.

At the same time, Kafai and Proctor raise important questions about what students might (or might not) learn in environments that feature computational practices and tools. What should college students learn about the kinds of problems that can benefit from computational approaches? How will they learn to think about the meaning of computational output and the assumptions that underlie it? When will they have opportunities to discuss the ethical dilemmas raised by these methods? What will they learn about the kinds of people who do computational work and the purposes to which computational methods are put? Because computational approaches are not yet routinely embedded in science curricula, the community has an opportunity to consider a range of possibilities for what this integration can and should look like, perhaps even leading to conversations about the purpose of science education more broadly.

REFERENCES