1. GET SMOGON SHIRTS HERE! First time ever. Limited time only & next time the design will be different, so get it while you can!
  2. New to the forums? Check out our Mentorship Program!
    Our mentors will answer your questions and help you become a part of the community!
  3. Welcome to Smogon Forums! Please take a minute to read the rules.

A Pokémon Essay: Cognitive Biases and Heuristics of In-Battle Decision Making

Discussion in 'Trou Du Cul' started by thesilverbandit, Jun 5, 2012.

Thread Status:
Not open for further replies.
  1. thesilverbandit


    Aug 8, 2010

    Pokémon: Biases and Heuristics of In-Battle Decision Making
    Josh Fuller
    Psychology of Decision Making (PSYC 25100)
    June 4, 2012
    Prof. Boaz Keysar
    T.A. Kevin Mulqueeny

    “So it is said that if you know your enemies and know yourself, you can win a hundred battles without a single loss. If you only know yourself, but not your opponent, you may win or may lose. If you know neither yourself nor your enemy, you will always endanger yourself.” – Sun Tzu, The Art of War 
    These days, the premier way to compete in the Pokémon metagame is to use an online simulator. These programs offer the ability to create a six-creature team of any Pokémon you choose with any possible moves and stats. The reason why these programs are so popular is that they offer the ability to do rapid-fire battles; once you finish one, it takes less than ten seconds to start another. I assert that this habit of battling strongly encourages the use of quick-processing heuristics and paves the way for biases to run unchecked and adversely affect decisions. Like in all other scenarios, it is important to identify and understand the ways in which your brain works in a decision-making situation; Pokémon is strictly strategy and decision-making, and this knowledge would benefit any battler. To start, I stress the importance of prediction.
    In a competitive Pokémon battle, prediction is the single most important factor. A good prediction can eliminate a big roadblock on the opponent’s team, while a bad one can sometimes ruin your chance of victory; when making an in-battle decision, it is safe to say that the stakes are high. Prediction, however, isn’t as hard as it sounds because there are some clues and resources to help you. After all, shouldn’t the ability to predict well just rely on one’s ability to put themselves in the other’s shoes? In the Pokémon metagame, this empathy is one of two key elements of prediction. The other is Smogon.
    Smogon touts itself as the “Pokémon University.” In the process of creating a new team, the first logical step is to navigate to this website and look up the Pokémon you’re interested in using. The articles there suggest many things: movesets, held items, teammates, battle strategies, etc. However, Smogon has a monopoly on this market; it isn’t the only website that has a ‘StrategyDex’, but it does have the largest userbase and the most visually appealing presentation. As such, knowledge of what Smogon says about a Pokémon gives you great insight into the metagame as a whole. A majority of other players use the same resource that you did while creating their teams, and often a player (I myself am guilty of this) will copy and paste Smogon’s suggestion directly into their team. Smogon is a jumping-off point for nearly all players, and extensive knowledge about the only accredited Pokémon literature leaves one at a distinct advantage. This exhaustive list of suggestions allows a player to guess the opponent’s moveset and item, and furthermore, to use evidence in-battle and the process of elimination to divine these attributes.
    Here, we come to our first cognitive bias: functional fixedness. Functional fixedness limits a person to using an object in only the way it is traditionally used. This is the cognitive bias of the Smogon copy-and-pasters. Movesets listed on Smogon are effective and have been decided by a group of experts, but the fact that this information is available to everyone limits its usefulness. To be predictable is to fail at the Pokémon metagame, pure and simple; sound predictions can even the score of the most lopsided battles. There are some Pokémon that have an obvious and established niche in the metagame: the best example for functional fixedness is Skarmory. Skarmory’s high Defense and Flying/Steel typing earns it a spot on many teams as a dedicated physical wall. The overwhelming majority of Skarmory have max HP and max Defense or Special Defense, thus allowing them to take hits with maximum effectiveness. When a physical threat sees Skarmory, it’s probable that Skarmory can come out on top in a one-on-one scenario, and the physical threat often switches out to something better suited for the task. However, innovation adds an element of surprise that is very potent in this metagame, and Weak Armor Skarmory is a deadly surprise. Instead of utilizing Skarmory’s ability to wall, the Weak Armor set relies on Skarmory’s niche as a defensive Pokémon to trick the opponent. When a physical threat switches out of Skarmory, Skarmory uses Swords Dance on the switch and doubles its attack power. When it takes damage, Weak Armor causes Skarmory to lose Defense and gain Speed. Its natural bulk allows it to take a few hits even with the Defense drop, and soon it has the Attack and Speed to sweep the opponent’s team. Overcoming the functional fixedness that Smogon brings to the table is key for success.
    Innovation isn’t always the best way to go, however. Pro-innovation bias is a marked preference for innovation that makes a person unable to identify the limitations and weaknesses of the innovation. Pro-innovation bias represents the other extreme to the predictableness problem, and I’ve illustrated this with a Special Attacking Aggron. Aggron has 110 Attack (good) and 60 Sp. Attack (mediocre). This vastly inferior Sp. Attack stat alone is enough reason not to use a specially oriented Aggron, but Aggron has a large Special movepool that would allow it to score super-effective hits on many Pokémon. Expert Belt is an item that boosts a super-effective move’s power by an additional 20%, so good type-coverage and Expert Belt is a potent combination. However, Aggron isn’t the best user of this strategy. In a scenario between an Expert Belt Aggron and a Physically Defensive Scyther, a physical Aggron performs better than the special one, even though Scyther is concentrated heavily on Defense rather than Sp. Defense. At most, max Sp. Attack Aggron can hope to dish out 42% damage to Scyther with a super-effective move boosted by Expert Belt . Alternatively, a physical set (even without Expert Belt) can beat Scyther in one hit with a 4x super-effective Stone Edge 100% of the time, even if Scyther is fully invested in HP and Defense . Playing to a Pokémon’s strengths is just as important as innovation: one must find the balance between the two in order to succeed.
    Pokémon is a guessing game: you must guess what the opponent is going to do next in order to counter it. Bluffing, then, becomes an important strategy. As Sun Tzu famously wrote in The Art of War, “All warfare is based on deception. Hence, when we are able to attack, we must seem unable.” By nature, a bluff is an act intended to use the opponent’s confirmation bias in order to confuse them. Confirmation bias is the tendency to search for information that confirms one’s preconceptions or to interpret information in a way that confirms one’s suspicions. To illustrate this, I will explain Expert Belt Latios. Latios, with 130 Sp. Attack and 110 Speed, is one of the biggest threats in the metagame. Its signature move, Draco Meteor, is an astounding 210 power after STAB, though it cuts the user’s Sp. Attack in half until it switches out. This move is so powerful that often Latios uses only this move because it performs so well. It operates under guerilla warfare: hit-and-run ensures that it remains at full strength and stays safe to wreak havoc again. Because it often only stays in for one turn, a Choice item is commonly used to instantly boost its stats upon entering the battle. However, a Choice item locks you into one move until you switch out, hence the name. This isn’t a problem for Latios, which is commonly used as a one-trick pony. However, to bluff being locked into one move fulfills Sun Tzu’s advice: we have the capability to attack but we seem unable. To deduce that a Latios is Choiced is to use the process of elimination of its commonly run sets. After-turn effects betray the Life Orb and Leftovers sets, leaving only Choice and Expert Belt as common options. If the Latios comes into battle, attacks, and then flees, that signals that the Pokémon may be Choiced. If you bluff the Choice item, however, you can lull the opponent into a false sense of security and strike a deadly blow when they least expect it. Assume that you have bluffed a Choice item for a while into a match and that you just KOed a Hippowdon with a super-effective Surf. Assuming that you’re locked into a Water move, the opponent brings in a Salamence, a Dragon type (which resists Water) that only needs one free turn of set-up to become an enormous threat. The opponent predicts a switch and uses a set-up move, Dragon Dance, in order to prepare to demolish whatever you switch into. However, you surprise him by switching up moves and slam him with a super-effective 420-power Draco Meteor, cutting his sweep short. By using his confirmation bias against him, you lured the opponent into your trap and eliminated a mighty sweeper.
    Due to the rapid-fire nature of today’s metagame, battlers sometimes tend to make hasty decisions. If a battler doesn’t calculate damage accurately, they can risk missing out on a clean KO. Say you have a Choice Band Terrakion and your opponent has switched in a physically defensive Tangrowth to try and stop you. Because Choice Band locks you into one move, you must choose between your most powerful move, Close Combat, or the super-effective move, X-Scissor. At a glance, a distracted or unmindful battler will click on X-Scissor. An 80-power move with the 2x super-effective bonus yields 160, and this will do 44-51% to Tangrowth . However, if I choose Close Combat, a 120-power move with the 1.5x STAB bonus yields 180, which does 49-57% . The chance to KO Tangrowth in two hits with X-Scissor is miniscule, while your chances of 2HKOing Tangrowth with Close Combat are significantly better. This is a result of the focusing effect: when one places too much importance on one aspect of an event that causes an error in predicting the utility of the outcome. A super-effective move isn’t always the best choice; a calculating and attentive mind can counter this cognitive bias.
    One’s perception of and subsequent identification of an opposing Pokémon is key. It answers the question fundamental to prediction: “What is this Pokémon about to do?” Here, an experienced battler has acquired heuristics to help answer this question quickly and accurately. Two of the most important in-battle heuristics are the representativeness and availability heuristics. With these, we take known scenarios and memories and compare them to the scenario in question. According to the representativeness heuristic, similar events can be assumed to have similar probabilities in their outcomes. When judging the representativeness of an event, a person takes into account how similar the event is to a typical event. According to the availability heuristic, something that is memorable or “available” is seen as more likely to happen. Tversky and Kahneman, in their 1974 paper Judgment Under Uncertainty: Heuristics and Biases, give the following example: one is given a description of a company and asked to evaluate its future profit. If given a high rating, one would rate their future profits as high because high profits are representative of a highly rated company; if given a mediocre rating, mediocre profits are predicted. They go on to state, “The degree to which the description is favorable is unaffected by the reliability of that description or by the degree to which it permits accurate prediction.” In other words, if people take only the description into account during prediction, the prediction is “insensitive to the reliability of the evidence and to the expected accuracy of the prediction” (1126).
    I can draw a similar analogy with a switch-or-no-switch scenario. Say you have a bulky Gyarados out on the field against your opponent’s Life Orb Zapdos. The representativeness heuristic gives you the information that Zapdos doubtlessly has an Electric move and will decimate your Gyarados, which is 4x weak to Electric . If you evaluate the utility of using Thunderbolt as very high for your opponent, you would assume that the opponent would go for it. Let’s also assume you have a Ferrothorn and a Heatran on your team (recommended partners on Smogon’s analysis and members of a fire-water-grass core). Ferrothorn has great defenses and resists Thunderbolt, so it’s a good choice to save your Gyarados from certain death by switching to it. You know from Smogon that Zapdos also run Heat Wave, a Fire attack that roasts a Ferrothorn with 4x super-effectiveness . The evaluation of the utility of this Heat Wave is the same as the aforementioned Thunderbolt: it’s too good of an opportunity for your opponent to not go for it. The availability heuristic manifests itself here: you can remember many instances where you’ve destroyed the opponent with a 4x super-effective move and you can also remember some time where you’ve cleverly switched your Pokémon around to avoid a catastrophic blow; these events stick out in your mind because they are notable and because they made you feel a strong affect, thus cementing them in your memory. Your plan is to first switch to Ferrothorn to resist the predicted Thunderbolt and then switch to Heatran to absorb the Heat Wave. From the defensive player’s perspective, this seems like a reasonable series of predictions. However, these predictions are assuming predictability in the opponent; the defender’s predictions are “insensitive to the reliability of the evidence and to the expected accuracy of the prediction.”
    Instead, the defender predicted the initial switch to Ferrothorn and incinerated it with Heat Wave, ruining the synergy of the defensive core and costing the defender a defensive powerhouse. This makes sense if you think about the situation from the opponent’s perspective. No sane battler is going to stay in with a Gyarados against a Zapdos when they have a Ferrothorn, especially if their entire defensive core is still intact; the defender would want to keep them all alive and utilize their type synergy. The switch to Ferrothorn is an obvious choice, even more obvious than the assumed decision to go for Thunderbolt against the Gyarados, and an offensive Pokémon in a switch-forcing situation should always capitalize on the predicted switch. For you, the disadvantaged defender, the loss of your Ferrothorn has triggered significant affect and is now incorporated into your availability heuristic; you will think twice before making another obvious switch.
    Such is the charm of Pokémon battling: the mistakes I’ve made build on each other and, over time, turn into expertise. These are only a few examples of the mistakes one can make, and I’ve made them all. My heuristics have evolved from a blank slate into rough tools and finally into honed weapons. As useful as they are, they can lead me astray. However, recognizing these biases and heuristics for what they are is important to keep them in check and keep them working for me instead of against me.

    Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124-1131.
    Tzu, S. (1963). The art of war. New York: Oxford University Press.
  2. doomvendingmachine

    is a Battle Server Moderator Alumnus

    Nov 8, 2009
  3. Seven Deadly Sins

    Seven Deadly Sins ~hallelujah~
    is a Site Staff Alumnusis a Forum Moderator Alumnusis a Tiering Contributor Alumnusis a Contributor Alumnusis a Smogon Media Contributor Alumnusis a Battle Server Moderator Alumnus

    May 29, 2008
    i, uh...




Thread Status:
Not open for further replies.

Users Viewing Thread (Users: 0, Guests: 0)