Computational Essays: An Avenue for Scientific Creativity in Physics
CComputational Essays: An Avenue for Scientific Creativity in Physics
Tor Ole B. Odden
Center for Computing in Science Education, Department of Physics, University of Oslo, 0316 Oslo, Norway
Marcos D. Caballero
Center for Computing in Science Education, Department of Physics, University of Oslo, 0316 Oslo, Norway andDepartment of Physics and Astronomy and Create for STEM Institute, Michigan State University, East Lansing, MI 48823
Computation holds great potential for introducing new opportunities for creativity and exploration into thephysics curriculum. At the University of Oslo we have begun development of a new class of assignment calledcomputational essays to help facilitate creative, open-ended computational physics projects. Computationalessays are a type of essay or narrative that combine text and code to express an idea or make an argument, usuallywritten in computational notebooks. During a pilot implementation of computational essays in an introductoryelectricity and magnetism course, students reported that computational essays facilitated creative investigation ata variety of levels within their physics course. They also reported finding this creativity as being both challengingand motivating. Based on these reflections, we argue that computational essays are a useful tool for leveragingthe creative affordances of programming in physics education. . INTRODUCTION
Traditionally, learning physics is not an especially creativeendeavor. In many physics courses, students spend most oftheir time solving problems that have already been solvedhundreds if not thousands of times, which have (at most) onecorrect answer, and which have only a small number of viablepaths to the solution. Laboratory work is often distilled intoa set of prescribed steps, with little room for deviation or ex-ploration. This educational model, however, is not authenticto the discipline of physics—outside of classroom contextsthere are seldom singular correct answers, and there is almostnever just one path to a solution. Often, "real" physics is aprocess of refinement, as expressed by the maxim "nothingworks the first time" [1].Certain reform-based teaching practices have started to addopportunities for creativity back into the curriculum. For ex-ample, design-based labs present students with challenging,open-ended tasks that have multiple possible solutions [2–4].Modeling curricula have also opened up room for alternativeapproaches to problem-solving [5–7], aided in large part bythe increased integration of computation into physics educa-tion. However, more remains to be done in this area.Computation offers numerous affordances for this type ofcreative learning. Computation allows students to accomplishmuch with a relatively small set of programming techniques.This can be empowering to some students, and opens upopportunities for exploration-based learning [8]. Addition-ally, computation allows students to easily explore and "playaround" with topics or concepts that might be analytically in-tractable or far above their educational level [9]. However,these creative affordances of programming have not been welldeveloped or studied within the context of physics education.In this study we examine how computation can create oppor-tunities for creativity in the physics curriculum, and how stu-dents experience that creative freedom. Specifically, we use anewly-developed class of assignment called a computationalessay to explore the ways in which programming can open upcreative opportunities for physics learning.
II. COMPUTATIONAL ESSAY DESIGN PRINCIPLES
Terms like creativity can be difficult to operationalize sincethey tend to have so many meanings across different contexts.In this study we are specifically interested in scientific creativ-ity in the context of task and curriculum design (rather thanas a property of individual students).
A. A framework for scientific creativity in education
Drawing on the limited work explicitly addressing creativ-ity in the science education and physics education literature[10, 11], we gauge the scientifically creative affordances oftasks and curricula by looking for the following qualities:1. Openness in task design2. Opportunities for original solutions3. Opportunities for exploration 4. Opportunities for collaboration and cooperation5. Disciplinary authenticityBased on this framework, we expect that tasks whichpresent opportunities for creativity are those which give stu-dents a challenge that is open-ended, in that there is no singleright answer nor an obvious singular path forward. Such chal-lenges should be related and relevant to the physics discipline,while also being accessible enough that students are able to"play around" and explore a variety of fruitful ideas related topossible solutions. They should also offer students the possi-bility to collaborate with one another, both when working onthe solution and when presenting their results.As previously stated, we see significant potential to lever-age computation and programming to create these types oftasks. Therefore, our research questions are as follows:
Howcan scientific creativity be used as a design principle to cre-ate open-ended computational projects? What effect does thisproject design have on student experiences?
B. Computational essay project context and design
To address these questions, we are developing a new typeof computational assignment in a computation-rich physicslearning environment, the physics department of the Univer-sity of Oslo, Norway. Since 2003 computation has been acornerstone of the University of Oslo physics major, with allstudents taking a Python programming course and a numer-ical methods course during their first semester. Subsequentcourses build on this programming foundation, having stu-dents write simulations as part of their weekly homework as-signments and exams. Building on this existing course struc-ture, in 2018 we began development of a teaching tool in-tended to capitalize on this programming focus. We calledthis tool a computational essay.Computational essays were originally proposed by diSessa[9] as a form of writing that uses text, along with small pro-grams, interactive diagrams, and computational tools to ex-press an idea. They are often written in so-called notebookenvironments such as Jupyter notebooks [12], programmingenvironments that allow users to combine code and text intoa single document. Notebooks consist of a series of blanks(called "inputs") in which users can type both code and text,from single lines to whole programs or paragraphs. They havelong been used by data scientists and professional physicists,both in exploratory analysis and presenting findings [13, 14].However, to our knowledge Jupyter notebooks and compu-tational essays have not yet been widely used in educationalenvironments.In the current project, we conceptualize computational es-says as a type of essay or report that explicitly incorporateslive code to support its thesis, usually written in a notebookenvironment. Computational essays include all of the ele-ments one would expect in an ordinary essay: for example,an introduction, thesis statement, body paragraphs, and con-clusion. They also have a similar purpose, to present a step-by-step argument or explanation. However, the argument in acomputational essay is driven by the output of various blocks
IG. 1. Example computational essay, showing the mix of text,code, and pictures. of code, with the text serving to both explicate the meaningof the code and to explain the output (see Fig.1).In the fall semester of 2018, we ran a small pilot projectto explore how computational essays could be used in theUniversity of Oslo’s introductory undergraduate E&M coursefor physics, engineering, and geoscience majors (roughly270 enrolled students). Students in the course were giventhe option to write a computational essay as an alternativeto a mandatory presentation-based assignment. Those whochose to participate were challenged to conduct an open-ended computational investigation of some phenomenon re-lated to the course content, then to write a computational es-say that summarized their investigation. Students were al-lowed to work individually or in pairs, and were given ap-proximately 4-6 weeks to conduct their investigations. At theend of the semester students presented their essays to theirpeers in mock research-group meetings, and the resulting es-says were graded pass/fail.To help scaffold the assignment, we provided students witha project description outlining the expectations, and an ex-ample computational essay on the topic "how much currentwould one need to use a railgun to resupply the InternationalSpace Station?" (Fig.1). We also provided the students with several "basic" simulations of electricity and magnetism phe-nomena written in Jupyter notebooks, including a simulationsof a cyclotron, storm cloud, lightning strike, and magnetictrap [15] for the students to build on if they chose, as well assome suggested investigation questions. The first author alsostaffed "help desk" hours where students could drop in to askquestions or work on projects.To address our first research question, this project was de-signed to explicitly incorporate the above-mentioned aspectsof scientific creativity. The project had an open-ended taskdesign, in that students were encouraged to define and pursueinvestigation questions based on their own interests. Becausestudents defined their own questions, their solutions (i.e., theinvestigative approaches, modified or re-written code, and ex-planatory text) were also original. This open-endedness wasalso tied to significant opportunities for exploration, in that itrequired students to spend time "playing around" with differ-ent types of simulations and code in order to both define andanswer their proposed questions. With regard to the collab-orative aspect of scientific creativity, direct collaboration onthe project was optional; however, the end-of-semester pre-sentations ensured that all students had a chance to share theirwork with their peers, including time for discussion and ques-tions about each project. The aspect of disciplinary authentic-ity was addressed in three ways: first, through the explicit useof the computational notebooks, which are a common tool inprofessional computational physics research; second, by hav-ing students engage in the practice of presenting their workto peers in a similar way to professional scientists; and third,in the requirement that students formulate their own researchquestion, rather than working off of a given prompt or ques-tion [16].
III. DATA COLLECTION AND ANALYSIS
In total, 17 students participated in the pilot implementa-tion of the project, working singly or in pairs to produce atotal of 11 completed computational essays. Three of the par-ticipating students were female and 14 were male; all femalestudents worked in a pair with a male partner. All essays werecollected, and all but one pair (15 of 17) students also con-sented to be interviewed shortly after completing their essays,resulting in a total of about 10 hours of interview data. Inter-view prompts asked students to walk the interviewer throughthe development process of their essay, reflect on the connec-tion between computation and learning physics, and reflecton the ways in which computation related to creativity in thecourses they were taking. To probe their views of creativityand computation, students were prompted with a statementsuch as "One of the goals with this project was to give youa little bit of creative freedom, to let you ’play around’ withthe physics and programming that you have learned so far.Could you talk a little bit about how that felt to you?" How-ever, some students also spontaneously brought up themes re-lated to creativity at various points in the interviews. Sincestudents were native Norwegian-speakers interviewees weregiven to option to speak English or Norwegian depending onheir preference, and three groups chose to conduct the inter-view partially or entirely in Norwegian.After all interviews were completed, the resulting record-ings were thematically analyzed and coded for themes aroundstudents’ perceptions of the relationship between computa-tion and physics learning, their general approaches to thecomputational essay project, and their views on creativityand computation. Essays were also analyzed, with an eyeto places where they overlapped with self-reported data fromthe interviews and places where they addressed things not dis-cussed (like specific coding practices, communication styles,and report structures). This analysis contributed to the emerg-ing themes.In what follows, we address our second research questionby presenting the subset of emergent themes specifically re-lated to computation and creativity.
IV. RESULTS
Based our analysis, we identified three primary themes:first, that students felt that this type of open-endedprogramming-based project presented numerous opportuni-ties for creative freedom on both a macro and micro-level;second, that they felt this freedom was a major source of chal-lenge in the project; and third, that they felt this freedom wasmotivating. Below, we elaborate on these themes with se-lected quotes from the interview data. All quotes are fromEnglish interviews, although these themes were also presentin Norwegian interviews, and all names are pseudonyms.
A. Computation allows for creative freedom in both approachand project
There was a consensus among the interviewed students thatprogramming provides significant creative possibilities, espe-cially in physics. For example, one student, Mel, had this tosay:
Mel:
So, overall I think programming is a very nice wayto be creating, because it’s very easy to get results from yourcreativity. You can do something for, like, not a long time,and then you get something! It’s like, it’s something specific,and you can change something, and then you can get some-thing else right away. So, it’s a, like a very op—a very easyplatform to be creative on, generally.
Here, Mel is explicitly reflecting on the creative affor-dances of programming, suggesting that it allows you to getinteresting results with a relatively small amount of effort.Furthermore, it is extremely easy to modify a written pro-gram, and any changes made will have an immediately per-ceivable effect.Beyond the general creative affordances of programming,students distinguished between two specific types of creativefreedom: freedom on the micro-level, in that programmingallows for variation in how individual students accomplishsimilar sub-tasks, and freedom on the macro-level, in thatcomputation allowed them to pursue a wide variety of top-ics and projects. On the topic of this micro-level of creativity,another student, Jeffry, added the following:
Jeffry:
There were always many different ways to solvea problem. And, in time, thinking about, "Okay, what is thecorrect syntax in—what is the preferred syntax in doing this?Like is it... Should you write it like that or like that?"
In other words, programming allows for multiple av-enues to accomplish similar tasks, giving students freedomto choose their preferred approach.In contrast to this micro-level of creativity, another student,Morten addressed the larger-scale creative uses of program-ming:
Morten:
I guess that’s the fun part of this, this project aswell. Like the computational essay. That you have a kind of ablank slate, you have kind of a situation you want to explore,and then you, like, make your own problem, so to speak. Andthen kind of just see what happens. And that’s a lot easier todo using programming.
Here, Morten echoes Mel’s sentiment that programmingis a useful tool for inquiry because it allows you to quicklymake modifications and see the effects of those on your sim-ulation. However, he also connects this affordance to theopen-endedness of the project, suggesting that it allows youto "make your own problem."These reflections are supported by the fact that, across all11 projects, no two groups chose exactly the same topic andapproach. There was a roughly even spread of projects basedon the various pre-built simulations provided, but in mostcases students who chose the same topic diverged signifi-cantly in their implementation and investigation questions.For example, of the two groups who investigated cyclotrons,one did a straightforward implementation of the effects ofspecial relativity on the provided simulation, while the otherre-wrote the simulation to investigate the effects of relativityon the LHC.This theme provides confirmation that our attempt to de-sign an open-ended project based on the principles of sci-entific creativity (research question 1) was at least partiallysuccessful.
B. Students found the openness of the project challenging andappreciated the scaffolds
Although students expressed appreciation for the opennessin the project, they also reported that this openness was a sig-nificant challenge in completing the project. For example,Jeffry added the following in his interview:
Jeffry:
We had the freedom to choose, of course. The prob-lem is that it makes it hard to choose a topic. So I went backand forth, looked at different topics. And even once I’d cho-sen that I wanted to do, the cyclotron, also trying to think of,’what is it I want to do with a cyclotron?’
Gerald echoed this sentiment in his interview:
Gerald:
I think, from where we were in the very begin-ning, [it] was a bit hard because it was almost a bit too open,’Cause I didn’t really know what to do. And so maybe wespent a bit too much time playing around with it and stuff.
In these two excerpts, both Jeffry and Gerald point to thedifficulty of identifying an interesting project topic. Thisifficulty is unsurprising, since the process of identifying a"good" research question is a challenging prospect, regardlessof field [16]. Thus, we argue that this challenge is authenticto the discipline of physics, likely more so than the standarddifficulties students face on problem-solving assignments.Despite these challenges, the students expressed apprecia-tion for the provided guidelines and scaffolds, suggesting thatthey helped to temper this difficulty. For example, Morton’spartner Kurtis added the following:
Kurtis:
It was good to have something to work around. Asa starter. When, like, it was like, ’okay, use electromagneticsto do something’ and then we were like ’okay, what does thatmean?’ It’s very good to have these examples to begin with.
Here, Kurtis refers to the fact that the original project de-scription left the topic very open, essentially asking studentsto "investigate a phenomenon related to electricity and mag-netism." Although the students found this openness challeng-ing, they appreciated being able to use the provided simula-tions as a starting point for their investigation.
C. Students were motivated by this creative freedom
Although they found it challenging, most of the studentsalso reported that they felt this freedom was motivating, fa-vorably comparing the computational essay project to morestandard programming-based assignments. For example, Lilymade the following comparison:
Lily:
Honestly I think this is better than the obligs [oblig-atory assignments] in a way because I think we pushed our-selves harder here than we would with those assignments. Be-cause then you have an endpoint like, okay, I know what theprogram or what the assignment asked me to do and here’sthe program. But now, when we finished something, it waslike ’this is really cool to actually see. What else can we do?’
Here, Lily explicitly compares this project to compulsoryweekly assignments, stating that she felt motivated by the factthat there was not a defined goal or end-point with this assign-ment. Mel echoed this sentiment in his interview:
Mel:
I feel like I had a lot of creative freedom, and it wasfun to, you know—normally, I’m not that easily motivatedwhen I have, like, one thing, and you have to do this spe-cific thing. But now I could choose my own thing and do thatthing. So it’s much easier to work and be effective when ac-tually working. And it was interesting trying to search on theinternet and find more out about this, things I didn’t know,and also try to implement parts of the course into the pro-gram.
Mel explicitly states that he is not usually motivated bystandard assignments that ask him to do "this specific thing."However, like Lily, with the computational essay project hefelt motivated to both work on his project and try to find newfeatures to build into it, drawing inspiration from course top-ics and online sources.These reflection are supported by the fact that about halfof the students used significantly more time on the compu-tational essay project than would have been expected for theproject it replaced. That is, the mandatory assignment sup-planted by the computational essay project was expected to take roughly 4-8 hours of work, with some students anecdo-tally putting in substantially less. However, when the inter-viewed students were asked how long they spent on compu-tational essays, all groups reported spending over 6 hours onthe project, and half (5 of the 10) reported spending upwardsof 14 hours per person. Most of the students also reported do-ing significant background reading and research to find pub-lished results with which to compare their simulations, in-cluding reading through published scientific literature.This motivational aspect, we propose, is one of the rea-sons it is especially important to build more of these typesof opportunities for scientific creativity into the physics cur-riculum. Several students explicitly stated that they felt un-motivated by standard physics assignments; these types ofstudents, we suggest, are being underserved by the standardmodel of physics instruction. Although activities like open-ended labs, inquiry-based teaching, and undergraduate re-search experiences may help, we also see open-ended pro-gramming projects like this one as an excellent way to bringin additional creative opportunities.
V. LIMITATIONS AND CONCLUSIONS