作者：Matthew Falk

1.背景信息

在开篇前，我要先申明我所遵循的主要理论是Lang教授所提出的“有限容量模式”（Limited Capacity Model of Motivated Mediated Message Processing，简称LC4MP）。这个关于人类认知与媒体加工的数据驱动模式中带有一些基本设想，而在我们展开详细分析之前必须理解这些内容。其中最主要的设想是，人类是一种有限的信息“处理器”，所以所有的认知过程都会随着时间的推移而发生改变。这就意味着我们拥有一些所谓的“认知资源”，需要我们从周边的环境中汲取相关信息，重视这些资源，并对其进行加工。每个人都拥有属于自己的“认知资源”，而不同人的认知资源也有所不同——而当我们所面临的任务要求更多资源时，我们可能会进入一种认知负荷状态。换句话说，那些被使用过的认知资源就属于“二手资源”了。而为了加工一些新信息，我们就必须停止使用这些“二手资源”的，而使用一些新资源。不管怎样，如果出现负荷状态，我们便难以适时加工那些所需要的内容。从当前来看，“加工”也就是编码，存储以及情报检索过程。

将认知资源致力于任务中的两大方法：自动分配或者作为控制过程。当我们周边环境中出现一些新事物时便能够使用认知资源的自动分配方法。那些我们必须学习，并且与动机有关的新事物都能够引起定向反应（Orienting Response，简称OR）。Pavlov最早明确了这种定向反应，他将其称为“是什么”反应。你随时可以看到这种过程——不论是在开会，上课还是等待别人来开门，你可以观察看看有多人转过头去看发生了什么事。这些人都在适应着环境的变化。还有一些生理反应也预示着我们具有OR。包括瞳孔放大，皮肤电传导，6至10秒的心率减速等等。而关于我们今天讨论的主题，定向反应意味着我们可以不假思索地关注某些对象，并使用认知资源去理解我们所关注的对象。这是一个完全自动的过程，因为此时的认知资源使用并不在我们的控制范围之内，它的发生并不受我们主观意识影响。

而控制关注则是一个完全相反的过程。如果我们尝试着去关注某些内容，如数学难题，我们便会将全部精力投入于这个问题中，并尽可能地使用更多认知资源。但是这个过程也很容易遭到自动分配因素的破坏，不过控制关注属于需要投入更多努力，而自动关注则无需耗费吹灰之力。

在媒体加工世界中，媒体的结构特征将引起定向反应，我们将其称为“定向刺激结构特征”（Orienting Eliciting Structural Features，简称OESF）。在视觉领域中，定向反应体现在镜头变幻，闪光灯以及快门的变化。音频中也存在这些特征：声音的变化，音调的提高，或者听到一个性感的词语。如果将这些内容移植到游戏中？想象自己处于一个战争环境下，磨拳霍霍准备施展自己的才能。你可能正忙于其它事，如你需要努力保护同伴的安全，或者拿着棍棒地打击敌人；当你到达一定能力时，游戏中便会出现视频指示器或者音频指示器，甚至两者同时出现（想象你在玩《魔兽世界》，当你的能力得到触发时屏幕上将会出现新的触发系统以及跳出一些新内容）。这种结构特征非常重要，你可以使用自动分配资源的方法去弄清楚它的意义。

UI mod(from forums.penny-arcade.com)

2.主要内容

当玩家在看到《魔兽世界》的突袭画面时，我听到的最无语也是最有趣的评价是“我太老了不能做这些事”或者“我并不擅长这么做。”这些人并不知道从游戏的第1个关卡到第80个关卡是一个持续的学习过程。随着时间的流逝，很多原本对于玩家来说困难的要求（如掌握哪些按钮执行哪些行动或者如何逃离火堆）最终都会变成一种自动的反应。换句话说，我们是学着如何玩游戏的。大多数情况下，如果玩家能够操作必要的按钮适当地控制一些游戏内容，他们也能够在第80个关卡中这么做。但是突袭任务只会出现在更高的关卡中。在《魔兽世界》中，突袭任务指的是由10至25人组成一个小组尝试着去攻打受游戏控制的强大敌人。玩家必须努力适应游戏中的一些内容：当你在游戏中做了某些事，或者你的对手做了某些事时，游戏都会响起一些声音；还会出现一些视觉信息提醒玩家他们拥有哪些能力以及敌人的所在位置等。但是在某些特定时刻，再大的资源库也会有遭遇耗竭。当这种情况出现时，我们便会再次面临认知负荷状态，不能够再加工我们所需要的内容了。如此看来玩家的游戏体验将遭到沉重的打击！那么为何游戏开发者还要创造出这些为难玩家的境况？有些人认为“如果你还不会做，那就等于你之前耗费的学习过程都是白费的”也有些人认为“这就是游戏的挑战所在。”虽然还有一些其它原因，我想我们首先应该弄清楚如何去克服游戏中的信息阻塞现象。

在研究中我发现了一个有趣的现象，即当我们承载着越来越多认知资源（游戏邦注：我们同时使用了更多认知资源于信息加工过程中），我们对于次要任务的反应时间（Secondary Task Reaction，简称STRT）也会变得越来越慢。这被当成是自动分配认知资源加工新信息的过程，但是却未能找到新信息，从而放弃旧任务而转向新任务（就像是按压复原按钮）。我们可以以《魔兽世界》为例解释所谓的“二次任务”。想象你是游戏中的魔术师，快速且频繁地投射出一些咒语“arcane blast”。突然屏幕上出现一些提示语告知你可以投射“arcane missiles”咒语了。这时你还能够快速按压相应按钮投射出这个咒语。但是如果你在游戏中所面临的任务是努力逃离火堆，移除好友身上的咒语，观察你向敌对生物发射了多少咒语等，那么这时你对于“arcane missiles”提示的反应速度便会大大减弱了。

玩家将通过定制用户界面处理这些信息；而如果游戏能够利用玩家所创造的这些用户界面元素将信息呈现在屏幕上，如此玩家便可以不用投入过多的认知资源了。但是从某些方面来看，如今的用户界面变得越来越混乱，所以这种方法已经不再可行了。

为什么玩家会创造出混乱的用户界面？原因有很多，其中一个较为经典的回答是“因为我们希望凸显这些mod，他们都非常重要。”但并不存在一个用以判断玩家究竟需要多少mod的通用标准。有些玩家会使用很多mod，也有一些玩家只会涉及其中一些。不管怎样，每个玩家都能够获得必要信息并判断他们所需要采取的行动。

虽然玩家必须拥有一套统一的游戏信息，但是关于界面mod问题则取决于他们的个人经验。玩家还必须更加关注于某些特定的行动。在某些情况下，添加用户界面元素将会影响我们对于新信息的反应速度（本来我们必须对此采取立即行动）。除此之外，当我们出现认知负荷状态时，我们又可能会抛弃原先的认知任务而将更多新资源投入于新的行动中去。

在此我们可以研究用户使用的界面元素的数量，以及这些元素在吸引用户关注时的效率之间的关系。如：

1.当屏幕上那些用于吸引玩家的界面元素数量增加时，玩家能够回应额外任务的能力便会减弱，直到他们再次面临认知负荷时，他们的回应速度才会逐步加快。

2.玩家在出现认知负荷前面对越多用户界面元素，他们就越能够拥有更多属于自己的认知资源。

这个是个很有趣的研究：首先，作为这些游戏玩家，虽然我们面对的是不同用户界面，但是我们却可以好几个人共同协作，完成游戏任务。其次，从理论上来看，这种方法能够进一步研究人类认知资源的局限性以及每个人在这点上所存在的差异。

为了在研究上取得进一步的飞跃，我们应该做出一些概括性的假设，并且主要是针对于解决一些外在效度问题。我们必须假设每个玩家用于加工每个mod的认知资源数量是一致的。我们同样也必须假设每个mod所要求的认知资源数量也是一致的。就像在《魔兽世界》中，Decursive（游戏邦注：一键驱散插件，帮助有驱散技能的职业一键驱散诅咒，魔法，毒等）的加工难度与Deadly Boss Mods（一种强大的老牌首领报警插件）和Omen（仇恨统计插件）是一样的。最后，我们需要假设玩家在整个游戏过程中会均衡地使用所有mod。换句话说，他们会平衡自己所拥有的mod，并不会总依赖于同一个mod。但是我们这里所讨论的只是一些片面内容，玩家有时候也会选择不同数量的认知资源进行加工。

这些假设会不会不切合实际？会不会影响我们的研究？这些研究思路是否能够帮助我们在解析个体mod对认知负荷的影响中起到推动作用？这些问题都是我们在今后必须深入探讨的内容。

游戏邦注：原文发表于2010年10月29日，所涉事件和数据均以当时为准。（本文为游戏邦/gamerboom.com编译，拒绝任何不保留版权的转载，如需转载请联系：游戏邦）

UI Modification, Cognitive Overload and Game Play

by matthewfalk

Our plan is to post new content at a rate of about one entry per week. I’ve been lucky enough to be chosen to lead off, and so today I thought I’d write a think piece that may better explain the perspective that I approach games from, and use this space to float an idea for a new avenue of research. Hopefully this can spark some debate between both the community, and my co-authors here, and get us off on to a good start.

Today I’d like to start off by discussing a theoretical idea that has been kicking around in my mind. Before I can do that, I have to give you a background about the theoretical basis I approach MMO games from. For those of you that may be familiar with the background and wish to skip that part, I’ve broken the entry into two easy to navigate parts: the Background Information and the Main Post.

1. Background Information

The primary theory that I work with is Lang’s LC4MP (Limited Capacity Model of Motivated Mediated Message Processing). This data-driven model of human cognition and media processing has some basic assumptions that we’ll need to understand as we progress here (Both in today’s entry, and in future ones that I make, most likely). The primary assumptions are that humans are limited capacity information processors, and that all cognitive processing happens over time. This means that we have something called “cognitive resources,” which are required for taking information from our environment, devoting attention (cognitive resources) to it, and then processing it. Each person has a single pool of cognitive resources, and each person’s is different – however, once a task calls for more cognitive resources than are available, we enter a state of cognitive overload. In other words, once cognitive resources are used, they’re used – tied up in the processing task at hand. To process new information we need to stop using them on a task they’re already being used on, and devote them to the new one. Either way, at the point of overload we lose the ability to process all of the things we need to or want to process at that moment. For now, I think it is sufficient to just say that “processing” is the encoding (intake), storage, and retrieval of information from our short and long term memory. If this point causes confusion, I can elaborate later.

There are two ways in which cognitive resources can be devoted to a task: automatically, and as a controlled process. Automatic allocation of cognitive resources occurs when something novel enters our environment. Things that are new to the environment, things that we have learned are important, and things that are motivationally relevant (mates, food, things that can hurt us), can all cause an Orienting Response (OR). Pavlov was the first one to characterize the orienting response, and he called it the “what is it” response. You can see this process happening all the time – the next time that you’re sitting in a meeting, or a classroom, wait for someone to open the door. Watch how many people’s heads turn towards the door see what is going on. They are orienting to the change in the environment. There are several physiological signals that also signal that we’re having an OR. These include pupil dilation, increased skin conductance, and a 6-10 second deceleration of the heart rate, and these will probably be important in the future. For our purposes today, it is enough to just understand that orienting means we are attending to something without thinking about it, and we’re devoting cognitive resources to figuring out what the thing we’re orienting to is. The process is said to be automatic because it is out of our control – it happens whether we want it to or not.

Controlled attention is the opposite. If we’re trying to focus on something, say a math problem, and we’re focusing intently on it, we are trying to devote as many cognitive resources to it as we can. This process can still be interrupted easily by things that cause OR’s, but think of controlled attention as effortful, and automatic attention as effortless.

In the world of media processing, we talk about structural features of media that cause orienting responses. We call these Orienting Eliciting Structural Features or OESFs. In the visual realm these are things like camera changes, flashes of light, and cuts.  Audio analogs exist as well: voice changes, the onset of a tone, hearing a sexy word. So to translate this to a game world, imagine you’re in a combat situation, and you’re waiting for your big ability to be ready for use. You’re busy doing other things; perhaps you’re keeping your party alive, or happily beating something over the head with a club, either way, when that ability is ready, there is usually some kind of visual indicator or audio indicator, or both (If you play WoW, think about the new proc system and how many things pop up on your screen when your abilities proc). That structural feature is important to you, and you’ll automatically allocate resources to figure out what it means. Even if all it means is “MASH THE 4 BUTTON NOW!!1!” as long as you have cognitive resources left, you’re able to process that and pull off your big ability.

I think we’re ready now for the real post!

2. Main Post

One of the most frustrating but humorous comments I hear about games is “I’m too old to play games like that” or “I’m just not good at games like that” when people watch World of Warcraft (WoW) raiding. What they fail to realize is that going from level 1 to 80 is a continual learning process. Over time, most of the things that are required for play (knowing what buttons to push, how to move out of fire, etc) go from being difficult to being automatic. In other words, we learn to play. For the most part, it seems as though most people are at least functionally able to operate the necessary buttons to do something reasonably resembling what they should be able to at level 80. Raiding, however, challenges the player at a higher level. If you’re unfamiliar with the term, raiding in WoW refers to 10 and 25 man groups attempting to defeat complex encounters with powerful game-controlled opponents. Game play offers many things to orient to: sounds are playing both when you do something, and when the creature you’re fighting does things, visual information about what abilities are available, where things to avoid  are (like that fire mentioned before…), and other information clutter the screen. As we discussed above, there are only so many cognitive resources available to go around. At a certain point, even those with the biggest resource pools are going to run out. When this happens, and cognitive overload occurs, we cannot process all the things we want to any more. This sounds like it would be disastrous for game play! Why would the game developers create situations in which we intentionally can no longer do everything they’re asking us to do? Some would say “if you can’t do it, you haven’t learned to play” and others might say “that is where the challenge is”. While there may be some truth to those statements, I think there is some interesting research that can shed some light on how we overcome the games’ information barrage.

One interesting finding in the research has been that Secondary Task Reaction Times (STRT) slow down as we become more and more cognitively loaded (as we have more and more of our cognitive resources devoted to processing information simultaneously). Then, at the point of cognitive overload, they suddenly become very fast again. This is thought to represent the process of trying to automatically allocate cognitive resources to processing the new information, finding that there are none, and then dropping old tasks for new ones (sort of like hitting the reset button). What we mean by “secondary task” here is easy to look at in terms of a simple WoW example. Imagine that you are a mage, continually casting the spell “arcane blast” as fast and frequently as you can. Suddenly, your screen indicates that the spell “arcane missiles” is now ready to cast. How quickly you push the button to cast arcane missiles would be your STRT. Now imagine that you’re also trying not to stand in fire (really guys, please stop standing in the fire), trying to remove curses from your friends, trying to watch how much threat you’re generating with the creature you’re fighting, etc. When all of those tasks are going on, you may be slower to react to arcane missiles becoming available.

Players deal with all of this information by customizing their user interfaces; by trying to use player-created user interface elements (commonly called mods, I will use the terms interchangeably) to display information in ways and places on the screen that make it more easily available and therefore take less cognitive effort to maintain. But at a certain point, user interfaces become cluttered, and are no longer effective.

So why do players clutter their interface? There are many anecdotal reasons to consider, but the typical response seems to be “because I need these mods to win, they’re required.” However, there is no universal standard to how many mods are required. I know of certain players who use many, and some who use only a few. Each of these players still gets the required information and is able to process the actions they need to take. A more theoretically driven reason comes to mind.

There is a common set of information that players must have, but beyond that, interface modding is a personal experience. There are certain actions players must pay more attention to than others. Up to a certain point, adding user interface elements causes a slowdown of our ability to react to new information that requires immediate action. After a certain point, we’ve established enough cognitive load that that arcane missiles button lighting up will cause us to dump the cognitive tasks we’re maintaining and devote a newly-freed-up chunk of resources to responding to this new call for action.

Here we can create a few hypotheses wherein we could investigate the relationship between the amount of interface elements that a player uses, and their effectiveness at responding to these calls for attention. Such as:

1.As the number of interface elements calling for attention on the screen increases, player ability to respond to additional tasks will decrease until the point of cognitive overload, at which point they will speed up.

2.The more user interface elements a player can pay attention to before causing cognitive overload, the larger their individual pool of cognitive resources must be.

To me, this is interesting for two reasons. First, as a player of these games, I find it fascinating that a group of us can come together and accomplish tasks successfully while all looking at very different interfaces. Second, from the theoretical standpoint, this could be a way to investigate the limits of human cognitive resource pools and the individual differences in them.

To make this leap we have to make a few blanket assumptions that might cause some issues with external validity. We have to assume that every player devotes the same amount of cognitive resources to processing each mod. We also have to assume that each mod requires the same amount of resources to process effectively. In other words, Decursive is as easy or difficult to process as Deadly Boss Mods is as Omen (threat meter) is. Finally, we need to assume that players use all the mods evenly throughout the whole fight. In other words, they’re constantly attending to each, not just using one here and there. The discussion here is also only looking at the player – side. Encounters themselves call for a differing amount of processing.

Are these assumptions too much? Do they invalidate the research? Is this research idea really a “further along” step in a chain of studies that must start with examining the impact of individual mods on cognitive overload? Perhaps – but these are the bits of thinking that lay ahead. For now, I’ll be happy with having the idea down and taking comments and questions on it.（source:motivateplay）