24. Learning and Constructivism
Teaching is an act of persuasion.
—DANIEL T. WILLINGHAM
Games are largely about learning. That is not to say that games are essentially edutainment, but that most games involve learning new concepts, mechanics, tactics, or strategies. A designer’s job is to make new games or update existing games with new parts. To understand how players will play a new game or new parts of a game, you need to understand how players will interact with that new information. Thus, a designer needs to know how people learn.
Edutainment is a marketing portmanteau for “educational entertainment.” Although games involve learning in general, edutainment games aim to teach the player something that is applicable outside the space of the game itself. Consider a game such as Oregon Trail, which teaches about frontier life, as an example.
Historic Approaches
The behaviorists discussed in Chapter 23 had their own applications of their theories to education. Skinner himself created a “teaching machine.”1 In it, a student would be presented with a question. If the student answered the question correctly, he would be rewarded. If the student answered incorrectly, the question would repeat until the student supplied the correct answer.
1 Skinner, B.F. (1958). “Teaching Machines.” Science, 128, 969–77.
Sure, this is boring, rote behavior, but you must remember that the behaviorists were not worried about things like motivation, retention, or fun, because those took place in the “black box.” It was only the behavior itself that concerned them. Thus, if they could get the student to answer correctly, it did not really matter if the student had full understanding or was satisfied with his performance. The behavior was enough. Reward success; punish failure.
The dissonance between the behaviorists’ experiments and how players learn in games should be clear. Designers do care if someone is having fun. Designers do care if their players can use knowledge they gain to solve problems. It’s not enough for players to know that Mario can bounce off a Koopa. They also need to be able to realize that they can bounce off multiple Koopas to cross a dangerous chasm.
One response to the behaviorist theories of learning is known as constructivism. Constructivism is a school of theories that treats the behaviorists’ “black box” as extremely important. The “construct” in constructivism means that each individual learner is constantly constructing new models of knowledge based on what they already know. One of its tenets is that how and what a person learns is influenced by that person’s prior experiences. A person learns by taking a concept she already knows and either assimilating her new experience to fit her current understanding of a thing or accommodating by adapting her understanding of the thing to deal with the new information.2 Each individual learner is a new challenge for the designer because each player’s past experiences are different.
2 Atherton J. S. (2013) “Learning and Teaching: Assimilation and Accommodation” [On-line: UK] retrieved July 7, 2019, from www.learningandteaching.info/learning/assimacc.htm.
Because of the influence of past experiences, constructivists highly value interaction with learning objects and learning-by-doing. If that sounds familiar, it’s because it’s largely the primary method that most games use to teach players about the game’s mechanics: the tutorial. The players start with some base level of prior knowledge and each step along the way guides the players so they gain experience in a safe way until they are proficient with the game. Most games assume a previous level of competency. For instance, most first-person shooters assume that the player is competent using an avatar in 3D space. To someone who has never played a 3D first-person shooter, moving the character around is a skill to be learned and a problem to be solved.
Novices and Experts
Understanding how people solve problems is important. First, there is an arbitrary distinction between classes of problem solvers. Players fall into one of two groups: novices and experts.
Experts are able to classify problems based on their structural parts. For instance, if I were to ask what 42 × 12 is, and then ask how much a book of 12 42-cent stamps cost, an expert would see these two problems as sharing the same structure. Novices would see these problems as an arithmetic problem and a problem about stamps.
Chess is a simple game of enormous possibilities. Chess has more valid, possible boards than there are atoms in the known universe.3 In fact, it’s likely that in any long game of chess, any two players will reach some board configuration that has never occurred in the history of the game.4 Psychologists attempted to study how chess masters were able to play such a massive game in terms of combinatorics at such a consistently high level. What do chess masters do that chess novices do not? Researchers found no difference in the memory abilities of highly skilled chess players; they cannot think much farther ahead than novices, nor is their level of intelligence vastly superior.5 What is different is their ability to break down chessboards into logical “chunks.” The positions of the pieces have meaning as relationships to one another that are organized in a way that only someone who has played thousands of games can understand, and they use this knowledge to identify common positions. Contrary to popular belief, they don’t think any farther ahead than the novice player, but they do understand a board at a glance by comparing what they see currently to other boards they have seen.
3 Shannon, C. E. (1950). “XXII. Programming a Computer for Playing Chess.” Philosophical Magazine, 41(314), 256–275.
4 Silver, N. (2012). The Signal and the Noise: Why So Many Predictions Fail—But Some Do Not. New York: Penguin Press.
5 de Groot, A.D. (1946). Het denken van de schaker. [The thought of the chess player.] Amsterdam: North- Holland. (Updated translation published as “Thought and Choice in Chess,” Mouton, The Hague, 1965; corrected second edition published in 1978.) Chase, W. G., & Simon, H. A. (1973). “Perception in Chess.” Cognitive Psychology, 4(1), 55–81.
Chess players remember and understand boards much like your average person remembers telephone numbers. You do not remember a telephone number as a disconnected string of ten random digits. Try to remember this number: 4122682323. Tough, right? Instead, you break phone numbers into an area code, an exchange, and the final digits. For instance, the phone number for the campus police at my undergraduate college was (412) 268-2323. I still remember this years later. The easiest part for me to remember is the 412. That area code is for all Pittsburgh numbers, and I grew up in Pittsburgh. Next, the 268 exchange is used for most of the university’s phone numbers. (The 268 uses the corresponding letters for the digits on the telephone keypad to create the letters of the university: CMU.) The last four digits, 2323, are also an easy to remember since they repeat. By “chunking” the phone number into easier-to-remember parts, you can increase the odds that you will remember it in an emergency.
Imagine explaining the game Braid (Figure 24.1) to someone who has never seen a video game. To expert game players, you explain that it is a platformer and that the triggers serve to rewind and advance time. Further explanation focuses on the time mechanics. But to novices, your explanation glosses over an incredible amount of detail by chunking a lot of discussion down to a single word: platformer.
BRAID © NUMBER NONE INC.
Figure 24.1 Braid is complicated, but less so if you are familiar with platformer mechanics.
Here’s what the explanation folds up by assumption into that one word:
• You control a single avatar.
• You use the control stick or D-pad to move left or right.
• Gravity pulls down toward the bottom of the screen. You have a side view.
• You have a jump button, which causes the avatar to jump.
• You can climb ladders with up or down movement on the control stick or D-pad.
• You cannot touch enemies from the side. Avoid them.
• You can touch enemies from above and that will kill them.
• Spikes will kill you.
• The screen can sometimes scroll left and right.
• Getting your avatar to some end point by jumping and movement is likely the goal.
Like the chess masters, designers fold huge amounts of experience into more compressed concepts. Giving control of Braid to a novice video game player is likely to result in the player asking “What do I do? What am I supposed to do?”
Genre is one way of condensing a lot of information into a particular schema. After mastering a first-person shooter, players do not need to be taught how to manipulate an avatar in a 3D space, reload, switch weapons, or any of the myriad mechanics associated with the genre. Players may have to be taught which buttons correspond to those actions, but they already have the why organized when they accept the genre. Novice players do not have those schemata to work from and so they need to be taught all these elements. Experts often forget that this knowledge is not innate. Unfortunately, most game designers are game experts and have difficulty thinking in terms of genre schema ignorance.
Cognitive Load
What happens if you present a player with too many problems to solve at once? How much “new” can someone handle? Psychologists refer to this as cognitive load. To discuss cognitive load, I need to define a couple of other terms central to the psychology of memory.
Long-term memory is generally what we think of when we talk about memories. These memories can last indefinitely. This is in contrast to what is called working memory.6 Working memory is the short-term buffer where we hold information that we have just received. If I pour myself a cup of soda in the kitchen, I use working memory to remember that I left it there so that I remember to pick it up when I return to the living room. Generally, we forget much of what we put into working memory. If I had used long-term memory, then I would remember weeks from now where I kept the cup on that day, which is not particularly useful.
6 Baddeley, A. (1987). Working Memory. Gloucestershire, UK: Clarendon Press.
When we have to learn and store new elements, we primarily tax our working memory. Unlike our long-term memory, which has a high limit on how much we can store, our working memory space is small and can handle only a few things at a time without forgetting other elements.7 For many years, researchers believed the amount of information we could keep in working memory was five to nine elements.8 More recent research has shown that the type of information stored (long words, short words, numbers, nonsense words) also affects how many elements our working memory can remember at a time.9
7 I contend that Darwin would have a hard time coming up with an evolutionary excuse for just how many movie quotes we can recall.
8 Miller, G. A. (1956). “The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information.” Psychological Review, 63(2), 81.
9 Hulme, C., Roodenrys, S., Brown, G., & Mercer, R. (1995). “The Role of Long-Term Memory Mechanisms in Memory Span.” British Journal of Psychology, 86(4), 527–536.
How psychologists define these limits is less important to the game designer than understanding that the limits exist. The five-to-nine estimate probably errs on the high side when you are considering how many concepts players can understand at a given time.
As in the Braid example, the number of concepts that have to be in working memory is simplified if you explain new concepts in terms of elements that are already in long-term memory. You don’t need to tell the player how to move if you can assume the player has already absorbed that knowledge from other games. This frees up space so you can talk about what mechanics are different in the game, and it is also a good reason for you to forgo complicated story elements early in a game. If the player is trying to remember the names and relationships between the members of the queen’s court, then this gives her less “space” to process how to play the game.
An apt analogy to use is the concept of juggling. Beginning jugglers need to start with only two items to juggle at a time. When they’ve mastered this, they can graduate to three. Then four. Then they can add tricks. If you try to teach someone by starting with how to juggle four balls at once, then the proto-juggler may never understand the concepts. The constructivist method of slowly building to a more advanced understanding is more likely to have success.
Portal is a game that expertly understands the role of cognitive load. The entire first half of the game is a tutorial, yet the player rarely feels like he is stuck learning instead of playing. This is because each test chamber adds only one or two mechanics or twists on those mechanics. The player slowly constructs his knowledge of the game’s systems, and by time the chambers get really complex, the player can juggle a dozen concepts in his head at once.
An opposing example is League of Legends. The game has a frustratingly high number of concepts to remember right off the bat.
• The goal is to defeat the opposing team’s base.
• There are common mechanics to master, such as moving, attacking, and recalling.
• There are four abilities unique to your character in addition to a passive ability.
• There are scores of champions, and each has unique abilities that can affect your character in many ways.
Taken all at once, these concepts are difficult to learn and reconcile. There just is not enough space in working memory to handle it all. It’s like starting to juggle by juggling eight balls at once. Once a player is able to commit some of these elements to long-term memory, then it becomes easier to learn and understand more and more new concepts.
Constructivist theorists call the process of providing just as much help as is needed at any given time scaffolding. The simile is that scaffolding in construction temporarily holds up a building as it’s being created. Learning scaffolding does the same; it supports the learner with effective help until that help is no longer needed. Portal has effective scaffolding because it tailors its level of help to the player’s estimated skill level based on the player’s location in the game. League of Legends provides little scaffolding. By comparison, it’s attempting to build a building by simultaneously constructing all the pieces in place.
A game that is well-scaffolded has an easier time accommodating new players. It first provides the player with all the help she needs at the time when she needs it. Then it slowly removes that help as the player’s skill increases. This is akin to taking off the training wheels when the player has learned to ride the bike. In digital games, the designer can poll game metrics to understand when the player is ready for decreased scaffolding. Thus digital games have a greater opportunity (and, I would say, responsibility) to scaffold the learning experience.
Expertise Reversal Effect
That the player learns and grows in expertise is important to how the game teaches the player. It’s understood that new players need well-crafted guidance. However, what happens when an expert player receives the same guidance? Expert players have to reconcile this guidance with what they already know.
Before Halo largely standardized how first-person shooters work on consoles, there were many competing control schemes. Some games defaulted to a flight-simulator setup in which pressing down on the right control stick caused the camera to pan up. Other games defaulted to a setup in which pressing up on the right control stick looked up. Players who had no past experience with either could adjust easily to both. However, players who expected one action and received the other often struggled with the controls. This is because their working memory was being spent trying to reconcile what they already knew with what was being presented. Novice players could outperform expert players in some tasks. This is known as the expertise reversal effect.10
10 Kalyuga, S., Ayres, P., Chandler, P., & Sweller, J. (2003). “The Expertise Reversal Effect.” Educational Psychologist, 38(1), 23-31.
The implication of the expertise reversal effect in games is that novices and experts need to be treated differently. Novices need guided instruction because they lack the working memory space to absorb large amounts of material at a time. Additionally, novices, if left to their own devices, largely rely on poor problem-solving strategies, such as trial and error.11 They need the help. Experts, however, find receiving redundant help frustrating. When help does not dovetail with their preconceived concepts, that help can become confusing.
11 Kalyuga, S. (2007). “Expertise Reversal Effect and Its Implications for Learner-Tailored Instruction.” Educational Psychology Review, 19(4), 509–539.
In the chess puzzle game named One Pawn Army (Figure 24.2), players start out with one pawn, and when that pawn captures a piece, it becomes that piece. The opponent never moves. The object is to capture the opponent’s king and avoid having your piece threatened by any opposing pieces. The setup leads to some interesting puzzles that are quite different in play from a normal game of chess. However, when the designer had a coworker who is a chess master playtest the game, he found that the player had a lot of trouble solving the puzzles. He could not “unsee” the board as an actual chessboard. His expertise hindered him.
IMAGE USED WITH PERMISSION.
Figure 24.2 One Pawn Army, by Mark Diehr.
Novices find it easy to miss contextual clues.12 If you end up thinking that it is enough to provide players with the material they need to learn without examining what the load will be on them, then you are likely to frustrate those novice players. Tutorials are necessary,13 but only in the context of effective teaching methods.14
12 DiPietro, Meredith. (2009). “Experience, Cognition, and Video Game Play.” Handbook of Research on Effective Electronic Gaming in Education (pp.776–790). Information Science Reference, IGI Global Publishing.
13 Glajch, S., Shea, A., & Tyrrell, J. (2006, February 28). “Game Tutorial Analysis.” Retrieved from www.wpi.edu/Pubs/E-project/Available/E-project-030206-173827/unrestricted/Game_Tutorial_Analysis.pdf.
14 Kalyuga, S. (2008). Managing Cognitive Load in Adaptive Multimedia Learning. Hershey, PA: IGI Global.
Split-Attention Effect
Another useful concept from the study of learning is the split-attention effect. This effect was noted in the instructional design of geometry problems. When the geometry problem showed a diagram and then had an explanation under the diagram, students showed poorer results than if the same geometry problem had the instructions integrated into the diagram itself (Figure 24.3). The theory is that jumping back and forth between the diagram and explanation adds cognitive load that may interfere with solving the problem.
Split Attention Itegrated
Figure 24.3 Split-attention and integrated diagram designs.
In games, avoiding the split-attention effect has been managed with user-interface elements that combine the game and hint effects. In Figure 24.4, the Elder Scrolls V: Skyrim chest has its instructions tightly integrated because the player is already looking at the chest. In Figure 24.5, the instructions for what to do in the Assassin’s Creed games are away from the action, which requires the player to shift attention from the action to the list of button presses.
THE ELDER SCROLLS V:SKYRIM IS © 2011 BETHESDA SOFTWORKS.
Figure 24.4 The context-sensitive directions appear on the item the player is looking at in Skyrim.
ASSASSIN’S CREED IS © 2007–2015 UBISOFT.
Figure 24.5 The context-sensitive directions are in a corner, away from the action in Assassin’s Creed.
Tutorials and Learning Design
Tutorials are the broccoli of the game design world. Designers hide them in their napkins if they can by placing a dedicated tutorial level behind menus or somewhere less accessible. Players also avoid them if possible, demanding to skip them to get to “the game.” This is not because tutorial learning is bad. Players want to learn how to be successful. They just want to do it in a fun way. Many tutorials are poorly designed and thus feel like punishment. The best tutorial of all time, in my opinion, is World 1-1 in Super Mario Bros. Most people had never played a platforming game at the time of its release, so this level served as a perfect teacher of the basic mechanics of what is, honestly, not a very straightforward game. Yet most players consider it just another level of the game and not a tutorial at all.
In the book Learn to Play, researcher and educator Matthew White lists a number of principles for what he calls “learning design” rather than “tutorials” because of the stigma of the term.15 These principles are based on peer-reviewed psychology research and are distilled for the concept of game design:
15 White, M. (2014). Learn to Play: Designing Tutorials for Video Games. Boca Raton, FL: CRC Press.
• Designers should make sure that all players reach a similar skill level quickly. Thus, the game should follow cognitive principles to offer learning support and inclusive features and feedback for low-skilled players without punishing or holding back higher-skilled players.
• Designers should overtly and obviously punish failures and reward success. Reinforcement should be immediate and the punishments and rewards should be large enough to affect behavior. Give big rewards.
• Digital games offer a vast amount of data for the designer to use. Game designers should leverage that data to test the player’s skill and offer help, challenge, reward, and punishment dynamically. Teachers do this in the classroom innately. They can use visual cues to understand when a person’s attention is waning and change gears to re-engage. A game cannot do that unless it’s looking and unless it can adjust itself to address waning interest.
Summary
• Constructivist theories assume that humans learn by building on their experiences and models of their ideas.
• Cognitive load is the measure of how much information a player can manage at once. It is managed by the way that players can effectively chunk information into fewer elements.
• Scaffolding is the process of supporting a player with just the amount of help she needs, removing help as the player no longer needs it.
• The expertise reversal effect suggests that help that is instrumental to novices may actually hinder experts because the experts need to use part of their cognitive load to manage the differences between how they understand the concept and how the help explains the concept.
• The split-attention effect suggests that help should be closely integrated with where the learner interacts with it. Hiding help in a manual is less effective than contextually giving help onscreen.