UX Magazine

Defining and Informing the Complex Field of User Experience (UX)
Article No. 799 March 6, 2012

Cognition & The Intrinsic User Experience

Over the past few years there's been a lot of discussion around whether an experience can be designed. But it seems like everyone's just getting hung up on semantics; an experience can be designed, but the user will always have the opportunity to experience it in a unique way. The reason every experience has the potential to be unique to the user is, in part, because cognition is unique to each user.

Cognition is about knowledge and understanding, so there's a ton of psychological principles that fall under the umbrella of cognition. I'll focus on two principles that, once understood, will elevate a UX practitioner’s designs to a whole new level.

Cognitive Barriers vs Cognitive Load

Even when experiencing the same stimulus at the same time, many users will have completely unique experiences. That doesn’t mean an experience can’t be architected that utilizes knowledge about cognition to increase goal conversion. We create experiences to elicit a response from users; those users’ responses are either extrinsic (e.g., subscribing to a newsletter) or intrinsic (e.g., developing brand loyalty).

Some of the most important decisions UX designers make are those they don’t even think about. It’s generally understood that creating an intuitive interface is important, but few people are really good at articulating what makes an interface intuitive. This is where the concepts cognitive load and cognitive barriers play a huge role.

When dealing with web and software development, principles associated with cognition can be distilled into six distinct categories: three related to cognitive barriers, and three related to cognitive load.

Cognitive Barriers

A cognitive barrier is something that prevents a user from performing the action required to complete his goal. Most cognitive barriers are temporary in the sense that they can be overcome just through information processing. For example: John begins to fill out a credit card application online and is met with a series of open form fields asking for his name, address, phone number, etc. He’s able to quickly move from field to field using the Tab key on his keyboard. The last question on the form asks him to select his interests and provides him with an array of checkboxes. The momentary pause required to process that he needs to shift from keyboard input to mouse input is a cognitive barrier, but only requires that he understand what to do in order to resolve the barrier. That said, this still represents a potential abandonment point if John isn’t able to figure out what to do.

Barrier #1: Number of steps

Number of Steps

Everyone has known about this barrier since the beginning of the Internet, and long before then. Why take three clicks when we can get it done in two?

Despite being the most well known barrier, it’s probably also the most misinterpreted because many people don’t understand that all three major cognitive barriers to have to be balanced. User testing and ongoing multivariate testing are two very good options for striking the right balance between number, length, and difficulty of steps in a user journey.

The takeaway: Understand that it’s equally important to know when to add steps as it is when to remove them. Five easy, short steps often impose a lower cognitive barrier than one long, difficult step.

Barrier #2: Length of steps

Perceived length of step

Just like barrier #1, the length of each step needs to be appropriate for a given experience. We can’t adopt a blanket rule that shorter steps make better experiences. In some cases, a longer step upfront could provide a substantially better experience as a whole.

There are two major considerations when examining length-of-step barriers: users expectations, and cognitive load. A user might expect to spend ten minutes applying for a credit card online, but might only expect to spend one minute finding show times for a movie. Additionally, users will only interact with systems they understand. Understanding the principles of decision-making, cognitive recognition, and cognitive recall will ensure users are not overwhelmed, while providing affordances for a complete experience.

The takeaway: Design pragmatic step lengths based on how motivated the user is to achieve his goal. Users will spend longer with sites, tools, apps, and products they enjoy than they will with ones they’re simply required to interact with. Users tend to prefer short steps that only ask them to resolve the immediate issue they’re faced with. For example, when a user lands on the Wikipedia page for the first time, he’s faced with the issue of selecting a language. It’s better to get him to select his language as one step and then get him to enter his search term as a second step rather than requiring him to fill out a series of questions that could be used to personalize his experience.

Barrier #3: Difficulty of steps

Perceived Difficulty of Steps

The difficulty of a given step is subjective, and is a main concern of UX professionals. Generally, it’s better to have easy steps; however, there are a couple of downsides to making things easy. Users tend to develop a greater sense of loyalty toward experiences that they’ve invested time in. Conversely, users tend to be fickle about experiences they’ve not invested much time in.

It’s important to understand that users tend to make quick decisions based on previously experienced conventions. This means that when steps of a process are considered important (e.g., selecting a payee, making a purchase, entering a contest) they need to make use of special design patterns that cause users to slow down. This type of slowdown often involves making steps more difficult to process, but result in less user error.

The takeaway: Don’t create unnecessarily difficult steps, but don’t immediately discount adding difficulty to limit conversion and increase the quality of the converted. Remember, users will be more likely to complete difficult steps if they understand why the step needs to be so difficult.

Cognitive Load

Cognitive load is the amount of working memory required to achieve the user’s goal. This principle forms the basis for Steve Krug’s book Don’t Make Me Think. The less a user has to think about what he needs to do to achieve his goal, the more likely he’ll be to achieve it.

Attribute #1: Number of choices

Number of Choices

Choice/decision architecture is becoming one of the biggest and most important specialties within the UX field. Understanding natural decision pivot points and how to manipulate the saliency of decision-making elements is key to ensuring users are quickly able to make the right choice.

For example, the most effective e-commerce sites focus on getting users to the product they’re looking for as quickly as possible before hitting them with related products/up-sells. These sites make great use of natural decision pivot points. Once a user has found what he’s looking for, there will be a natural point at which he’ll be receptive to additional offers. If there are related products, up-sells, or related promotions, capitalizing on these pivot points is important.

The takeaway: Human working memory is limited. Users are more likely to move around a site with a simple structure than one with a very wide or very deep structure. George A. Miller published a paper in 1956 called The Magical Number Seven, Plus or Minus Two: Some Limits on our Capacity for Processing Information is the quintessential guide to avoiding choice paralysis. It essentially stipulates that the majority of people have the capacity to remember 5–9 things. So if you’re creating a taxonomy, it would be ideal if it were somewhere in that Goldilocks zone. That said, a more recent study suggests that working memory limits are likely lower, possibly as low as four things.

Attribute #2: Amount of thought

Amount of Thought Required

The most important part of understanding cognitive load is understanding how much a person needs to think about a decision prior to making it. Thought processing is somewhat of an abstract concept because is varies substantially from person to person and doesn’t directly relate to real-world time. This means that it’s possible to create a longer experience that has a lower cognitive load, and conversely, to create a shorter experience that has a higher cognitive load.

Each experience has to be evaluated individually to determine whether people would:

  1. understand that they need to take the time to make the desired decision, and
  2. are willing to spend the time required to make the decision.

These are two distinct considerations. Many people are used to making hasty decisions online because they rely on their own experience to interpret design patterns. If they are asked to take the time to make the optimal decision, even if it only requires one second longer than it would to make a satisfactory decision, users will need special design patterns to recognize they’re being asked to do this.

Take the current incarnation of fox.com, for example. The primary navigation has little downward-facing arrows next to each element. Here’s a great example of a design pattern intended to slow users down and make an optimal decision. These arrows indicate that users should not simply select a section, but should expect to see a mega-dropdown with sub-categories.

Fox.com sub-menu

Although this is a valuable design pattern, Fox has failed to use the appropriate interaction design pattern. They’ve decided to expand the mega-nav on click, which is fine, but rather than closing the mega-nav if the user clicks again (i.e., making each navigation element a toggle), Fox takes the user to that category landing page if the user clicks again.

The takeaway: Users rely on their own experience interacting with digital, and non-digital, products. Therefore, users will make decisions they understand first, and will only stop to consider their decision if they don’t understand what to do. If you use standard conventions, you’ll ensure users don’t have to think too hard to use your site, app, or product.

Don’t ask users to select between too many options. Again, the 7±2 rule is a great guideline to adhere to. Don’t have more than 5–9 calls-to-action, categories, or menu items displayed at any given time. This can be achieved by hiding additional options off-screen, or though a well-thought-out taxonomy. Hiding elements should be done using standard conventions, e.g., standard vertical scrolling, “Advanced” buttons, split buttons, collapsible areas, ”Show more” buttons, etc. Avoid hiding list items that need to be evaluated together.

Attribute #3: Confusion and choice

Confusino and Choice Graphic

How would you log into an investment account with your online bank if your bank has two options: “Online Banking” and “Credit Cards”? Most people would use process of elimination to select “Online Banking,” but some users may abandon their goal if the don’t understand the choice. It’s kind of like asking people if they want a fork or a knife to eat their soup.

Many UX professionals get caught in this pitfall by not allowing users to evaluate a complete set of options at a glance. Remember the 7±2 rule? Well, this is where it starts to get slippery. If you’re unable to reduce the breadth of a site to 5–9 top-level categories, it’s better to display all of them than to display a subset of them. For example: John is looking for a set of work gloves and visits the Canadian Tire website. There are eight top-level categories that appear in the primary navigation. John begins to look for which category he thinks might contain work gloves. He doesn’t see a category that makes sense but knows that Canadian Tire sells them. The issue is that the canadiantire.ca only displays a subset of the total number of departments within its primary nav. Along the left rail, there’s local navigation that includes all of the departments, one of which is apparel. Apparel does not appear in the primary nav. It’s okay to show a subset or summary of options upfront if it’s clear that it’s only a subset, and if there’s an option to show all options.

There should never, or rarely ever, be a need to hide a selection of navigational options. It’s fine to hide the navigation as long as there’s a clear way to access it again; but it’s important to show all of the options when the navigation control is displayed.

The takeaway: Users often mistake a selection of options for the complete set of options. It’s easier for a user to understand which option to select when he can see the alternatives. If only five options of a 20-option set are visible at a time, it will be more difficult for the user to decide which option to select.


UX has a lot to do with how users find and consume content. Understanding the cognitive processes and nuances people go through when finding and consuming content is important to architecting an ideal experience or, at least, to architecting a set of conventions that support a user having an ideal experience.


User Profile

Jordan's been a freelance experience strategist for the past 5 years, working with organizations across North America. He's in the process of establishing Hostile Sheep, a boutique experience lab created to help organizations produce better digital products and services. He's worked with brands like Coke, Nike, BMW, Canadian Olympic Committee, HP, GE, Ford, Canadian Tire, Kraft, Telus, P&G and Diageo. He's worked with agencies like Critical Mass, Razorfish, W+K, TAXI, Trapeze, Ogilvy One, Proximity/BBDO, Cheil, JWT, and others. Follow Jordan on Twitter or connect with him on LinkedIn.

Add new comment



not so sure about the 7+/- rule..otherwise a superb article!


Thanks for the article, Jordan. A lot of thought-provoking stuff. I have a few thoughts that I wanted to add to the mix:

In the diagram at the top, I would say that the amount of thought required means the same thing as cognitive load and can be broken down into things like number of choices, amount of previous experience, number of parallels that can be drawn with other related experiences, etc.

Where you talk of cognitive barriers, I'd argue that the act of switching to a mouse requires no reflection or conscious thought. The physical movement is a non-issue as it is such a frequent action in computing and the user knows exactly what happens when checkboxes are clicked. I think the risk of form abandonment using checkboxes is pretty low... Unless I've misunderstood what the article is saying. I'd suggest that there is more of a cognitive barrier around using a keyboard to advance to the next step in a multi-screen form. I can see users asking themselves whether they will be able to tab to the submit button: they may think that a next button with an arrow image isn't linked to the tab index of the rest of the form because it doesn't look like a native form control.

What makes you say that users only interact with systems they understand? I'd suggest that, in many cases, people will just have a go rather than dropping off.

Regarding increased loyalty from longer experiences, I'd say that it really depends on how long they expect the process to take from the outset - more to do with the perceived effort > perceived reward. Even half way through a process, starting the process again over the phone could be quite an attractive option (unfortunately).



I really enjoyed reading your article and share many of your opinions. Nevertheless, I feel that I should also add to the comments on the 7+/-2-issue. First, Miller (1956) did not stipulate any magical number. He has used 7 only as a rhetorical device in the title and introduction of his article (see the last paragraph of the article). His article certainly is one of the most often misinterpreted scientific articles of all times (see Rasmus T’s comment below). Second, Miller reports very strong empirical evidence for cognitive capacity limitations. Third, not just one “more recent study” found capacity limits around four. Four is actually (based on a bunch of recent empirical studies, notably those by Cowan and colleagues and Vogel and colleagues) the currently most widely accepted estimate for human working memory limitations. Of course, also this number should be used with caution (see Damon Dimmick’s comment below). I discuss these issues in more detail here: http://www.centigrade.de/en/blog/article/the-number-seven-is-not-magical-part-1/ and here: http://www.centigrade.de/en/blog/article/the-number-seven-is-not-magical-part-2/


Well done within the constraints imposed on the dilemma by the author. It gives me hope for a more comprehensive, nuanced discussion that goes beyond the simplest cognitive science. That experience is designed and experienced differently is no surprise to anyone. Cognitive theory gives slim hope of getting it right, only how not to get it wrong in one particular way. By the way, are there synonyms for "design" and "experience"? I think not, which is why we have great difficulty talking about them in non-simplistic fashion.


Very nice article, I'll definitely be saving this for reference. For the "Online Banking" or "Credit Cards" choice, I think a more telling analogy is how people react at a nice restaurant when the salad comes out and they realize their place setting has two different forks. They know that one is the "right" fork and the other is the "wrong" fork, but they don't know which is which and don't have a way to find out without feeling stupid.


I think there's a nuance problem in the discussion re: 7+-2. The problem with the 7+2 rule is that it's sometimes applied without thinking about the content. Applications/arguments/thoughts on the 7+-2 rule should always be accompanied with some sort of semantic structure discussion, such as chunking. The original "working memory" aspect of it simply states that it's hard to keep more than 5-9 different items (e.g. numbers, letters, concepts) in working memory simulatenously. (And that I think is true.) By chunking a phone number for example, we make it easier for ourselves and can handle larger amounts of data (three "chunks of two" numbers instead of six individual numbers). An information architecture where the nav items are hard for a user to understand – that's when I think the rule/guideline can come into play (because that requires "knowledge in the head"). But obviously, it's better to design an IA where the items are clearly mapped to end-user taxonomy and vocabulary. Then users will carry out the chunking (and other strategies) effortlessly and continuously; and have no problem with more than 9 categories/concepts.


Gabriela Trindade Perry and Suely Fragoso discussed the issues with Miller's 7 ± 2 paper in an article on UX Magazine: Using Scientific Knowledge to Bring Structure to Design Problems. The conclusions from the paper aren't invalid, they're just limited in scope and applicability.


PS: I should note that the "paradox of choice" idea is still in flux. Lots of great research done, but we're still dealing mostly with theory.


Hi Jordan,

I think you'll find that many of the studies suggesting increased dropoff rates for more choices relate to a completely different set of rules, actually closer to to pure cognitive load theory. Also (just having personally seen some of those studies) these studies tend to ignore the actual choices themselves, which at least in my anecdotal experience tend to be confusing. That is to say, often times it is the info architecture / taxonomy that is broken, not the amount of choices. Also important to remember that dropoff rates can also be a result of introducing choice at the wrong time, rather than the number of choices. 7/+2 is really just about retentive memory although it, on its surface, appears to relate to choice overload. There is absolutely some truth in suggesting that increased number of choices relates to decreases satisfaction, but it has very little to do with active memory. I think that's what most of the feedback here is getting at: the 7/+-2 rule is not linked directly to the situation you are discussing, but is rather a parallel phenomenon which occasionally coincides but is not causal to the much discussed "paradox of choice" in cases where users do not actually have to remember anything. At least that's what I'm interpreting based on my reading.

Rest of the article though is immensely valuable and I have shared it a few times in the last day.


Nice article and I agree with a lot of your points.


That didn't take long... I figured someone would say something about the 7+/-2 rule. Just want to say for the record, I was just suggesting that working memory is limited and should be a consideration when creating hierarchical taxonomies.

The interesting thing is that I'm not convinced that this rule (in the way I've interpreted it) is a complete myth. I don't think it's the only consideration, but I've seen evidence from several different clients site analytics that suggest people make more errors, drop-off more often, and take longer to achieve their goals when asked to decide between 9 or more choices.

And remember, I'm only suggesting that people take longer to process more choices. I think the Miller study caused a lot of people to look at human behaviour and motivations closely, so although I don't think he made exactly the right claims or provided any real evidence, I think there's still merit in how his study influenced UX.

And I'm not necessarily


I generally use max 4 as a rule of thumb to strive for in menu choices. Can't produce a better link than http://www.ncbi.nlm.nih.gov/pubmed/11515286 right now. But it seems to be the max amount of items you can instantly identify and say how many they are without counting.


@John U, can you elaborate a bit? I have seen the 7+/- rule referred to in many books, articles, and so on from G. Millers theory even thought it is from 1956. Do you have some other references that states otherwise?

Thanks for a good article!


I think the information on cognitive load is very valuable, but I agree with John U. The information on the 7+-2 rule is completely wrong under the noted circumstances. That rule is specifically intended to be a guideline to working memory and is not really a useful consideration for persistently displayed information. The rule itself is generally regarded as a myth in academic UX circles (in regards to persistent displays):



This is a good, practical post. The only negative is the use of the 7+/- rule, as this is a myth that doesn't have psychological support. It is one of those pop design rules that don't hold water.

But that is only a minor point, otherwise I enjoyed your post. Always useful to have matrixs to evaluate a design by.