Artificial Life in Game Mods for Intuitive Evolution Education
AArtificial Life in Game Mods for Intuitive Evolution Education
Anya E. Vostinar , Barbara Z. Johnson , and Kevin Connors Carleton College, Northfield, MN 55057 Grinnell College, Grinnell, IA [email protected]
Abstract
The understanding and acceptance of evolution by naturalselection has become a difficult issue in many parts of theworld, particularly the United States of America. The useof games to improve intuition about evolution via natural se-lection is promising but can be challenging. We propose theuse of modifications to commercial games using artificial lifetechniques to ‘stealth teach’ about evolution via natural se-lection, provide a proof-of-concept mod of the game StardewValley, and report on its initial reception.
Introduction
Teaching the theory of evolution via natural selection hasbecome a controversial issue in the United States educa-tional system in recent decades (Sinatra et al., 2008). Evenafter setting aside the political and religious motivationsfor resistance, there remains a central challenge that evo-lution via natural selection is counter-intuitive partially be-cause it occurs at spatial or temporal scales that are difficultor impossible for humans to observe, let alone understand.There are many efforts underway to improve this problemof observable evolutionary dynamics, and one of them isto incorporate evolution into games. This approach enablespeople to observe evolution in a context that can be non-confrontational and on temporal and spatial scales that theycan grasp.There have been a number of games that attempt to incor-porate evolution as a central or secondary game mechanic,however they run into several recurring issues. One suchissue is that evolution via natural selection is an inherentlydirectionless process and there is a great temptation by gamedevelopers to add direction to it to make a game more fun.This change, however, reinforces a common misconceptionof evolution via natural selection, namely that of some be-ing (either a deity or the organisms themselves) directingthe process towards a goal (e.g. Spore (Bean et al., 2010)).Some educational games have a primary aim of demonstrat-ing evolution by natural selection correctly, but end up be-ing games that are not particularly fun to play. Even whena developer succeeds at creating a game that is both scien-tifically accurate and fun, they face an uphill battle to gain a large fanbase and achieve the goal of demonstrating evo-lution via natural selection to a large audience. Finally, if agame markets itself as being a game about evolution via nat-ural selection, those who are resistant to accepting the ideaof evolution are unlikely to even try the game. Therefore, ifthe goal is to use games to improve evolution understandingand acceptance, a game would have to 1) correctly imple-ment evolution via natural selection, 2) be fun, 3) amass alarge fanbase, and 4) not be overtly about evolution via nat-ural selection.Creating a new game that will achieve those four crite-ria is a difficult problem, however we suggest an alternative.Some games support community modifications of the gamecode, called mods, that allow programmers to tack on ex-tra features to an existing commercial-off-the-shelf (COTS)game. By modding existing commercial games, researchersensure evolution is correctly implemented, are able to use afun game with a large fanbase (assuming the game is alreadycommercially successful), and could even utilize ‘stealth-learning’ by pitching the mod as making the game more re-alistic (Sharp, 2012). Some studies have begun to exploreusing this approach in both formal school settings whereteachers create educational mods of popular games (Moshir-nia, 2007) and during out-of-school learning opportunitiessuch as after school coding clubs in which learners modifyCOTS games to develop or demonstrate their understandingof academic content (Bertozzi, 2014; Hayes and King, 2009;Steinkuehler and Johnson, 2009).Here we discuss how research in artificial life could beharnessed for educational goals involving evolution via nat-ural selection through games and present a proof-of-conceptmod to demonstrate the possibility of this approach. We de-veloped this mod for the popular indie game Stardew Valley.This game, released on February 26th, 2016, is a “country-life RPG” that allows interactions with plants and animalsthrough farming, breeding, and fishing (Barone, 2020). Thistheme, combined with the fact that over 5 million peopleown the game (Orland, 2018) and that there already exists anactive modding community (of both programmers producingmods and players downloading mods) makes Stardew Valley a r X i v : . [ c s . N E ] J u l he perfect avenue for this proof-of-concept mod. The devel-oped mod targets the fishing feature of the game, adds func-tionality and possibly fun to that feature, and demonstratesa way that evolution via natural selection can be integratedinto existing games. Background
Games and virtual worlds have been used and tested in for-mal and informal educational settings to teach a variety ofsubjects, including STEM, as long as the technology has ex-isted. For many educators, particularly in K-12 classrooms,games (either physical or digital) are seen as a way to en-gage and motivate the new generation of learners who pre-fer and are increasingly adept at learning concepts and com-plex tasks through a variety of media (Sharp, 2012; Dahleenet al., 2014). Since Net Gen students (also called “Digi-tal Natives”) have grown up surrounded by digital mediaand routinely play computer (or smartphone) games evenas young children, the traditional methods of direct instruc-tion are ineffective tools for 21st century learners who thriveby learning through multimedia-supported exploration andknowledge construction (Prensky, 2010).It is important to note that, in this context, the term gamesincludes a wide variety of ludic tools. Games may includethose played with physical manipulatives, such as board andcard games, as well as digital games played on a computeror tablet device - with or without Internet connection. Thegames used may also be either those designed for instruc-tion (called serious games or serious educational games akaSEGs) or commercial games designed for entertainment butrepurposed as educational tools. A number of studies havedemonstrated success in using serious or commercial games,either physical or digital, in both the classroom and in in-formal situations such as afterschool programs. However,attention needs to be paid to both the questions of whetheror not the game is fun enough to hold the attention of thestudents (Baker, 2008) and also to what games are appropri-ate to particular pedagogical approaches and levels of learn-ing. Studies have shown that games (physical and digital)can be used to support traditional, instructionalist paradigmsas well as a variety of active learning strategies designed toengage students in higher order thinking, reaching into thehighest level of Blooms taxonomy: creating (Krathwohl andAnderson, 2009).
Direct instruction and practice games
Early educational games were often designed within the tra-ditional, direct instruction framework in which the com-puter substituted for the teacher in providing information ordemonstrating a skill. These edutainment “games” were of-ten digital, interactive versions of quizzes, flashcards, andautomatically scored worksheets and were not particularlyfun or game-like in themselves (Bertozzi, 2014). As anearly attempt to mimic the stereotypical classroom, these games generally targeted lower-order learning objectives inwhich students were expected to remember specific factswithout necessarily being able to apply them to novel sit-uations (Krathwohl and Anderson, 2009).
Experiential or inquiry-based learning
While direct instruction games focused on aiding studentsin remembering specific facts, active learning games wereearly experiments in creating a learning environment inwhich students would gain virtual experiences and collectinformation about a system in order to form and test hy-potheses about what is happening in the system. Studentscould be guided by an instructor or the in-game instruc-tions to run a simulation of a natural system and record re-sults (Christensen-Dalsgaard and Kanneworff, 2009; Gibsonet al., 2015; J¨ordens et al., 2018) in order to understand howthat system works and propose hypotheses that explain re-sults or predict future events as they continue to play thegame.In other science learning games such as Quest Atlantis(Barab et al., 2005), River City (Clarke and Dede, 2009),and Citizen Science (Gaydos and Squire, 2012), studentstake on the role of scientists to investigate a virtual worldproblem such as lake pollution. As students would maneu-ver their game avatar through a virtual world they wouldmake decisions about where and how to gain informationin order to form hypotheses about problems being experi-enced and to propose solutions. Often these learning ob-jectives were posed as quests, a popular fantasy game trope,that provides players with a defined, achievable goal that canbe reached by a variety of paths. This increases a sense ofagency and empowerment in the player/learner, and the in-trinsic reward of solving the puzzle or fixing the game-worldproblem keeps learners motivated to overcome challengesand increase learning or skills along the way.
Creative platforms for learning and assessment
Often, educators carry the burden of finding or creating agame to use in teaching their course content, however, it ispossible to pass over that responsibility to the learners andchallenge them to design a game while they increase mas-tery of academic content. Game creation or modding hasbeen used as a creative platform with a specified rule/tool setthat guides or inspires their development of understanding ofeducational content - particularly of large systems (Dahleenet al., 2014). Such constructivist learning approaches weredescribed by Lave et al. (1991) and are being tested as ameans for learning academic content such as second lan-guages (Monterrat et al., 2012).Games and game development processes can also be usedby students as creative platforms to demonstrate their un-derstanding of concepts that they developed by creating newgames or modifying existing ones. For instance, studentswho mastered fractions in an elementary math program weresked to create a game to teach those concepts to youngerstudents (Shreve, 2005), demonstrating proficiency as wellas creating an instructional tool that the teacher could lateruse with new students.
Stealth learning or persuasive games
Finally, students can learn by using non-traditional activi-ties, such as playing a game without overt learning con-tent that nevertheless teaches desired skills and academiccontent. Called stealth learning, this teaching techniqueis a way to engage students in a learning activity withouttheir awareness that their fun has a serious purpose. Forinstance, entertainment-focused board and computer/videogames have been used in the classroom to teach a wide va-riety of academic skills and content (Shreve, 2005; Sharp,2012; Squire and Barab, 2004).Stealth learning and persuasive games pose a unique chal-lenge since their purpose is to teach without seeming toteach. Hence, they cannot be overtly serious or educationalin nature. They also need to be fun and have sufficiently highquality graphics and challenging game play to entice play-ers to pick up and continue to play the game for enjoyment,hopefully recruiting friends to also play the game so thatthe education or persuasive messages will spread without ateacher imposing extrinsic motivation in the form of grades.This requirement to make an educational game look and playlike a commercial, entertainment-oriented game is a seriouschallenge since few educators have the skills or resources tocreate and distribute them. Hence, there is increasing inter-est in modding games, either through modding commercialgames or by using serious games that have proved to be suc-cessful in engaging students to keep playing outside of theclassroom.Studies have primarily concentrated on these approachesusing games/virtual worlds specifically designed for edu-cational purposes, but far fewer studies have concentratedon using a widely-used commercial game, designed for en-tertainment, as the basis for science education, particularlyin biology and evolution. Squire did a study with studentswhere they successfully learned about history through play-ing Civilization III (Squire and Barab, 2004), but the videogame Spore has been problematic when used to teach aboutevolution. In particular, Spore highlights how entertainmentfeatures of a game may lead to students developing misun-derstandings rather than acquiring desired competence withacademic content, unless the teacher explicitly identifies theflaws of the games rules (Schrader et al., 2016).Part of the challenge in using COTS (commercial, off-the-shelf) games is that few of them are modifiable from atechnical or legal standpoint. Even in games that allow mod-ding, the modding activity is often limited by the game pub-lisher to creating new maps for players to explore, addingcustomized appearances of game objects, or changing theappearance of the user interface (Thiel and Lyle, 2019). Figure 1: Screenshot from Stardew Valley of a farm withcrops growing, provided by StardewValley.net.Fortunately, some commercial game developers have begunto encourage modders by the release of software develop-ment kits (SDKs) or by creating sandbox games, such asMinecraft, that are designed for active user participation andcontribution of modified game components. While these area small portion of the gaming universe, the existence of toolsthat allow players to design new levels and even use pro-vided game engines to add new features and functionality tofavorite games opens up a world of opportunity for educa-tors to leverage a popular games reputation and its moddingcommunity to develop educational content and distribute itto players outside of a formal learning environment (Thieland Lyle, 2019).However, many games that support or encourage mod-ding, such as World of Warcraft, Call of Duty, or Half Life,are frequently focused on some sort of combat mechanic,which is problematic when considering a game for educa-tional use (Johnson et al., 2008). Further, most of thesegames that can be modified do not have a world or narrativethat can be appropriated to teach about the natural world andevolution via natural selection. The emergence of the ‘farm-ing simulation’ genre of game resolves these issues becauseit has little if any combat and instead focuses on the playergrowing crops, caring for animals, and possibly befriend-ing non-player characters in the world. This genre is there-fore friendly for educational goals and allows the incorpo-ration of evolution via natural selection into the story narra-tive. A ‘farming simulation’ game that also allows moddingis Stardew Valley and therefore is the game we use for ourproof-of-concept mod.
Methods
Stardew Valley is a semi-open world game where the playeris a new arrival to a small town and has inherited farmlandthat they can develop, as seen in Fig 1. The main activi-ties of the game are farming crops, raising animals, fishing,mining for minerals, and interacting with the townspeople.There is arguably no ultimate goal, though earning enoughoney to purchase items that the player wants is a centralfocus. In the game there are wild and cultivated plants, live-stock, wild animals, and a fishing minigame (a small gamewithin the larger game). All of these game mechanics aresuitable places to inject artificial life, as they are modeledoff of life forms already. We focused on one of them, thefishing minigame, which allows you to catch and sell fish.Fishing is an aspect of the game that is relatively sim-ple and therefore developing a mod for fishing was feasible.Fishing in the base game is a dividing feature: many play-ers enjoy fishing, so improving fishing would be desirable tothem; however, many players do not enjoy fishing becausethe minigame is too challenging; so giving them a reasonto fish would also increase the number of people using ourmod. Finally, the effects from fishing for profit is an exam-ple of rapid evolution in our world and therefore a relevantissue to demonstrate (Sharpe and Hendry, 2009).
Fishing Basics
In the base game, there are a finite number of species of fishthat are available to be caught, and certain fish can only becaught in certain circumstances (location, season, weather,or time of day), but an infinite population of fish that is notinfluenced by player actions. Which fish you catch is deter-mined by chance, as some fish are rare and some are com-mon. Each fish is of a different length, in inches, but thatlength is used for “fishing achievements” when you catch thefish, and then is later discarded. Our mod for fishing, namedR
EALISTIC F ISHING , makes use of the built-in length prop-erty, but modifies it to become a part of the resulting I
TEM obtained from fishing.Two complaints often voiced about fishing in StardewValley are that fishing is not very profitable after the firstyear, and that fishing is boring. Our mod increases playerengagement with fishing by fixing both problems. Fish aremade much more profitable by adjusting the sell price to bebased off of the length of the fish, rather than a fixed number.Fishing is made more fun by pressuring the player to makeinformed decisions about which fish to keep, and which fishto throw back into the water. These decisions form an intu-ition pattern that can be used to teach concepts about popu-lation effects and fishery collapse.All of this is made possible by expanding the way fish-ing works. There are four core changes that R
EALISTIC -F ISHING makes: 1) the addition of a persistent and evolvingpopulation of fish that is affected by the player’s actions, 2)the fish I
TEM has been expanded to have a
LENGTH prop-erty and that affects the sale price, 3) the player can onlychoose to keep 10 fish per day, and 4) the player can throw acaught fish back into the water. These four changes producean immediately obvious economic incentive to release smallfish and keep longer fish, because they sell for more money,which then affects the persistent fish population. We wantthe player to think critically about which fish to keep and which fish to throw back into the water, since in the natu-ral world, what you catch directly influences the population.The same is true with our simulated population: catch (andkeep) too many of the large fish, and your fish population isreduced to small, unprofitable fish.We used SMAPI, the Stardew Valley Modding API,to create our mod. SMAPI provides procedures to cap-ture game events and interact with them. All of the codefor R
EALISTIC F ISHING can be found on the GitHub repohere:https://github.com/anyaevostinar/RealisticFishing.
General Implementation Details
The base game doesnot have a persistent fish population. Instead, when theplayer catches a fish, there is a random chance of it beingany of the species that can be found in that area. We cre-ated an actual (virtual) population of fish so that when theplayer catches a fish, it is randomly selected from the listof existing fish in the population. When the fish is caught(and the player chooses not to throw it back), it is removedfrom the population. At the beginning of each game day, themod checks if the fish population is full and if it is not, arandom fish remaining in the population reproduces (asexu-ally) with a chance of its offspring having a mutated lengthtrait. The amount of mutation is pulled from a normal dis-tribution with a mean of 0 and a standard deviation of 2,which is then added to the parents length trait value to deter-mine the offsprings length trait value. There is a maximumand minimum length allowed for each type of fish, and soif the offspring exceeds these limits, its length is set to theexceeded bound.While a length trait was pre-existing in the game, it did notactually factor into the price that the player could sell the fishfor, nor was it tracked by the game after the player caughtthe fish. To enable there to be some reason for the playerto care about the length of the fish they catch, we addedthe length trait to the fish I
TEM . The amount of money theplayer receives when they sell a fish is now whatever the fishwould have sold for in the base game plus the length of thefish divided by 8. We decided on the divisor of 8 throughplaytesting and determining what seemed balanced.With the previous two changes, the player still wouldlikely not actually see evolution in the fish population be-cause the base game allows the player to keep all the fishthat they catch. To force the player to make decisions aboutwhich fish to keep, and to improve the realism of the mod,we introduced a daily limit of 10 fish. This addition is simi-lar to many fishing restrictions in the real world (Heino andGodø, 2002) and causes there to be an incentive to the playeronly keeping fish that are the most valuable, which are thefish that are the longest. To enable the player to keep onlythe longer fish, we also added the ability for the player tothrow back fish that they do not want to keep.These changes mean that a player that aims to make asmuch money as possible will keep only the ten largest fishigure 2: Screenshot from Stardew Valley of the dialoguewhen Demetrius, the resident scientist, introduces himself.that they catch each day. This decision will remove thosefish from the population, leaving fish that are smaller on av-erage to reproduce. If this decision is repeated, the fish pop-ulation will evolve to be smaller on average, as has occurredin some real world fisheries (Sharpe and Hendry, 2009).While the player may notice this change in their fish pop-ulation on their own, we decided to make it especially clearto them by expanding another existing game feature. Thegame includes a scientist in the town already, shown in Fig 2,(an inclusion that the authors find wonderful) and the playercan receive mail from the townspeople through normal gameplay. We added the feature that if the average length of afish type in the population is below a threshold, the scientistsends the player a letter stating:“Dear
PLAYER , I was conducting a field study on
FISH N AME the other day, and I discovered that the pop-ulation is in decline. To prevent a fishery collapse,please release any large
FISH N AME you catch until thepopulation is stable again. -Demetrius”
Results and Discussion
Conclusion
We have shown that it is possible to implement evolutionarydynamics in an existing popular game and deploy that mod-ification to the general public. In future work, we will con-duct experiments to determine if and how much such a mod-ification increases players understanding and acceptance ofevolution. In addition, we will test the feasibility and recep-tion of similar modifications to other popular and modifiablegames.By modifying pre-existing games, we are able to placeaccurate evolutionary dynamics into games that are alreadyfun and widely used. Further, the way that the evolution-ary dynamics are incorporated can be chosen to be subtleenough that players do not realize that they are witnessingevolution in action, possibly making some more receptive tounderstanding the dynamics of evolution, as has been donewith other forms of stealth gaming. While this strategy ofmodification is only one of many ways to increase evolu-tion understanding and acceptance, it is one that may reachfurther and therefore is worth considering.
Acknowledgements
This project was funded by the Mentored Advanced Projectprogram at Grinnell College. The authors would also like tothank Eric Barone for creating Stardew Valley and the mod-der Pathoschild who created SMAPI and has always beenilling to answer questions as we developed RealisticFish-ing.
References
Baker, C. (2008). Trying to design a truly entertaining game candefeat even a certified genius.
Wired Magazine , 16(04).Barab, S., Thomas, M., Dodge, T., Carteaux, R., and Tuzun, H.(2005). Making learning fun: Quest atlantis, a game with-out guns.
Educational technology research and development ,53(1):86–107.Barone, E. (2020). Stardew Valley Homepage.Bean, T. E., Sinatra, G. M., and Schrader, P. (2010). Spore: Spawn-ing evolutionary misconceptions?
Journal of Science Educa-tion and Technology , 19(5):409–414.Bertozzi, E. (2014). Using games to teach, practice, and encourageinterest in stem subjects.
Learning, Education and Games ,page 23.Christensen-Dalsgaard, J. and Kanneworff, M. (2009). Evolutionin lego R (cid:13) : a physical simulation of adaptation by natural se-lection. Evolution: Education and Outreach , 2(3):518–526.Clarke, J. and Dede, C. (2009). Design for scalability: A case studyof the river city curriculum.
Journal of Science Education andTechnology , 18(4):353–365.Dahleen, J., Hunsberger, A., Weber, R., Brylow, D., Longstreet,C. S., and Cooper, K. M. (2014). Towards a lightweightapproach for modding serious educational games: Assistingnovice designers. In
Proceedings of the 20th InternationalConference on Distributed Multimedia Systems . KnowledgeSystems Institute.Gaydos, M. J. and Squire, K. D. (2012). Role playing games forscientific citizenship.
Cultural Studies of Science Education ,7(4):821–844.Gibson, A. K., Drown, D. M., and Lively, C. M. (2015). The redqueens race: An experimental card game to teach coevolu-tion.
Evolution: Education and Outreach , 8(1):10.Hayes, E. R. and King, E. M. (2009). Not just a dollhouse: whatthe sims2 can teach us about women’s it learning.
On TheHorizon-The Strategic Planning Resource for Education Pro-fessionals , 17(1):60–69.Heino, M. and Godø, O. R. (2002). Fisheries-induced selectionpressures in the context of sustainable fisheries.
Bulletin ofmarine science , 70(2):639–656.Johnson, B. Z., King, E., and Yucel, I. (2008). User-created contentand program-modification in video games and virtual worlds.Panel at Meaningful Play.J¨ordens, J., Asshoff, R., Kullmann, H., and Hammann, M. (2018).Interrelating concepts from genetics and evolution: why arecod shrinking?
The American Biology Teacher , 80(2):132–138. Krathwohl, D. R. and Anderson, L. W. (2009).
A taxonomy forlearning, teaching, and assessing: A revision of Bloom’s tax-onomy of educational objectives . Longman.Lave, J., Wenger, E., et al. (1991).
Situated learning: Legitimateperipheral participation . Cambridge university press.Monterrat, B., Lavou´e, E., and George, S. (2012). Learning game2.0: Support for game modding as a learning activity. In
Pro-ceedings of the 6th European Conference on Games BasedLearning , pages 340–347.Moshirnia, A. (2007). The educational potential of modified videogames.
Issues in Informing Science & Information Technol-ogy , 4.OolongBoolong (2020). Realistic fishing mod page comment. https://forums.nexusmods.com/index.php?/topic/6885712-realisticfishing/page-2 .Orland, K. (2018). Valve leaks Steam game player counts; we havethe numbers.Prensky, M. R. (2010).
Teaching digital natives: Partnering forreal learning . Corwin press.Schrader, P., Deniz, H., and Keilty, J. (2016). Breaking spore:Building instructional value in science education using acommercial, off-the shelf game.
Journal of Learning andTeaching in Digital Age , 1(1):63–73.Sharp, L. A. (2012). Stealth learning: Unexpected learning oppor-tunities through games.
Journal of Instructional Research ,1:42–48.Sharpe, D. M. and Hendry, A. P. (2009). Synthesis: life historychange in commercially exploited fish stocks: an analysis oftrends across studies.
Evolutionary applications , 2(3):260–275.Shreve, J. (2005). Let the games begin. video games, once con-fiscated in class, are now a key teaching tool. if they’re doneright.
George Lucas Educational Foundation .Sinatra, G. M., Brem, S. K., and Evans, E. M. (2008). Changingminds? implications of conceptual change for teaching andlearning about biological evolution.
Evolution: Educationand outreach , 1(2):189–195.Squire, K. and Barab, S. (2004). Replaying history: Engaging ur-ban underserved students in learning world history throughcomputer simulation games. In
Proceedings of the 6th inter-national conference on Learning sciences , pages 505–512.International Society of the Learning Sciences.Steinkuehler, C. C. and Johnson, B. Z. (2009). Computationalliteracy in online games: The social life of mods.
Interna-tional Journal of Gaming and Computer-Mediated Simula-tions (IJGCMS) , 1(1):53–65.Thiel, S.-K. and Lyle, P. (2019). Malleable games-a literaturereview on communities of game modders. In