Conditioning in UX Design
How Operant and Classical Conditioning inform learnability and user interactions.
Preventing users from becoming confused about how they got somewhere or what something does while exploring an interface is key to keeping them coming back. By designing interfaces that are highly learnable and easy to use, this understanding is increased. But how do we design for learnability?
One way is with a better understanding of the role conditioning plays in this experience. While in gamified experiences, user behaviors like consistency, hard work, purchases, and social interactions are rewarded with points encouraging users to continue these actions, other experiences rely on cause and effect among other methods to establish a connection between a user action and a ‘reward’ or desired outcome. Users learn to seek these small wins.
What is Conditioning?
Is there a difference between learned behavior and learning how to use an interface? Are the two really so diametric? After all, engaging with an interface and understanding it involves actions and consequences — a click leads to an outcome: a new page, a purchase, an error, a confirmation. Conditioning is this association of actions and consequences.
There are two schools of thought in conditioning theory. Let’s break down each of them to better understand what they have to do with UX.
Part 1: Operant Conditioning
Behaviouralism assumes that all learning occurs through interactions with our environment. As a behaviouralist, twentieth-century influential psychologist B.F Skinner believed that there was no need to look at the internal motivations of a person to explain their behavior, one need only look at the external observable causes of that behavior. Operant conditioning, he theorized, is a way of learning by associating success and failure with rewards and punishment; the idea that behaviors have consequences.
Skinner, posited the theory while studying human behaviouralism at Harvard. His operant conditioning apparatus, which later became known as the “Skinner-box” was an enclosed chamber containing a bar an animal could press to receive food, water, or another form of ‘reinforcement’. This reinforcement was found to increase the likelihood and frequency with which the desired action occurred.
Reinforcement can be anything that strengthens or increases the behavior. For example, when children receive an allowance for completing their chores they learn that this behavior has beneficial results and thus will continue to do so.
Positives and Negatives
Skinner made a distinction between positive and negative reinforcements. With positive reinforcement, there are favorable outcomes like praise or rewards, and something is added to increase the response. With negative reinforcement the unfavorable outcomes are removed which still can act as a reward — like when a teacher cancels a quiz because every student completed their homework. This removal of adverse conditions can encourage the behavior, the same way as the addition of a reward.
These positive and negatives are not meant to equate to good or bad, rather are in the mathematical sense to mean the addition or subtraction of a thing.
Punishment
There are also punishments with the intent to deter actions. When a child touches a hot stove and gets burned, they learn to not touch hot objects and heed warnings. With positive punishment, something is added to the equation — for example, after committing a crime, prison or community service may be given as a punishment. By contrast, a negative punishment involves the removal of a positive reinforcer— for example, paying a fine or taking away a child's toy.
So what does this look like in UX?
Consider the last time you engaged with an interface that was new to you. If it was something you’d never seen before how did you know how to use it? You likely engaged in some form of testing and discovery; you pressed a button, followed a link, and maybe searched for a keyword. If you liked the feedback you received you likely stayed on the site longer than you would have had you received an error.
Positive reinforcement occurs when something favorable is added to the experience. In UX, this can look like gamified rewards but it can also take the simpler less direct form of a Call to Action (CTA) or incentive. From the home page of a website, seeing a CTA in the header can increase conversion rates. If users are given an incentive to create an account, like the promise of 25% off their first purchase, they may be more likely to do so.
Negative reinforcement, the removal of unfavorable conditions, also has a place in UX, though it is not generally seen as a user-friendly method. It is also harder to define a use case for negative reinforcement since it, by nature, means the user had an unfavorable experience with the product to begin with — something we want to avoid.
Perhaps, one such case would be using minimum viability blockers on orders as is common in grocery ordering apps where users must reach the minimum order cost to make the purchase. If a user is looking to order 1 item, unless that item alone is $30, they will have to add more items to their purchase to reach the total before proceeding. When they reach the total by repeating the desired action of adding items to their cart the unfavorable condition of being blocked from proceeding is removed. While this may seem like a favorable outcome has been added — and it has– the user's awareness of the unfavorable outcome (the block) also means that its removal signifies a negative reinforcement.
We typically also don’t see punishment employed outside of gamification since it again is not very user-friendly nor user-beneficial. We help users diagnose and recover from errors rather than punish them for making them since there is no benefit to the business in doing so. While in gamified products users are punished for mistakes by losing lives and may be unable to play for a period of time once they’ve lost all their lives, we don’t do the same when a user fails to enter their password correctly too many times. Or do we? Can it be considered a punishment when a user is locked out of their account? Despite creating solutions to help the user recover from the errors, we don’t let them continue to try indefinitely without consequences.
Response Strength
Back to Operant Conditioning theory. Interestingly, the strength of the behavior or response to the reinforcement is affected by how and when that reinforcement is delivered. There are a few factors that affect this:
- Accuracy: Does the reinforcement deliver the desired response?
- Duration: How long does the response last?
- Frequency: How often does the response occur?
- Persistence: Does the response occur every time?
In these ways, we can measure how effective the conditioning is and make adjustments to encourage more of the desired behaviors. In effect, we reflect these same principles when designing UX interfaces to inspire such behavior. Iterative testing of an interface provides insights into whether these reinforcements are having the desired effect.
- How accurate is the conversion process and where are the issues causing users to drop off or fail?
- How long are users engaging with a flow before they stop?
- Are users coming back and how regularly?
- Are returning users following the same path, doing the same thing?
Classical Conditioning
While Skinner’s Operant Conditioning theory relies on direct correlations of voluntary behavior and response, Classic Conditioning theorizes that associations can be built between a conditioned stimulus and an involuntary response.
Classical conditioning, regularly associated with Ivan Pavlov and his dog, relies on the understanding of an unconscious learning process creating a conditioned response to an unconditioned stimulus and a neutral stimulus. In other words, by reinforcing a natural reflex with a neutral signal (or stimulus), unconsciously we learn to associate the two. In the dog’s case, food that naturally causes salivation was conditioned to be associated with a sound, and eventually, the sound alone could cause salivation.
Even us humans have come to associate food with the sound of a bell as anyone who’s worked a restaurant service pass may know. But do we salivate when we hear it? Maybe not. But it may also not be a voluntary reaction to be drawn to the sound.
What does this look like in UX?
Classical conditioning is why products make sense to us and why we come to associate excitement, happiness, success, or pain, failure, and rejection with these products.
Products rely on interface and UX design to create efficient processes that take the user from logging in to conversion. Through repeated exposure and conditioning, users have come to associate different interface elements with different outcomes.
- Buttons when clicked will have an action > Add to Cart buttons when clicked will add an item to the virtual shopping cart.
- Links when clicked will take the user somewhere else > Main navigation links when clicked will take the user to another page on the site while hyperlinks may take the user off the site.
- Pressing the plus sign will allow a user to add something > Using the plus sign on a social media app will allow users to create a post.
- Notifications signal that something is happening > A message notification signals that the user has received a message.
These outcomes are universally applied to products and recognized by users even when they are using a particular product for the first time. This familiarity is what makes that product easier to learn.
Users can come to form associations between these individual elements and an unconditioned response or feeling, or between a task flow, or even the product as a whole.
Observe yourself the next time you’re making an online purchase. Try to focus on when in the process you feel the rush of excitement or guilt. Is it when you add an item to the cart, when you enter your card details, or when you see the confirmation message? With Classical conditioning, we come to associate the neutral buttons, form fields, and actions with the same feelings we have about making the purchase.
By taking an unconditioned stimulus (excitement) and a neutral stimulus (the Checkout button) and associating them — i.e., making the button click intermediary to making a purchase — the user begins to feel excited when clicking the Checkout button. This is the conditioned response.
We may even come to associate the whole product or site with our feelings towards the outcome. If your bank account is drained, you may start to feel sick when opening your banking app. Looking to move you may feel excited opening the Zillow app. Need a new roof you may feel guilty going on Amazon.
These feelings are not always ones we as designers and product owners have control over especially those that are conditioned by external factors. But we do have control over the conditioning within the product that encourages users to come back and feel positive about their experience. We can lessen the cognitive friction in a flow, reduce pain points, and keep interfaces familiar but also exciting.
The Role of Conditioning in UX
Operant and Classical conditioning work simultaneously in products to reinforce and build associations between actions, elements, behaviors, and products, and their consequences or outcomes. As UX designers, we play a role in what these associations look like and how they keep users happy and engaged.
Learning by Association
When an action is discovered through exploration and results in a favorable outcome, the user will more than likely return to repeat this action the next time knowing that it is there. But if the user is unaware of what they did because it was too difficult to recreate or inaccessible, they will be conditioned to prevent this engagement in the future. The frustration caused by the ordeal will leave a lasting impression that likely means they won’t return.
Operant conditioning is learning by association. Through experience, certain behaviors are reinforced or punished leading to an association with pleasurable or painful outcomes. In UX, this same process affects the way users engage with products or processes. With a positive impression, users will keep coming back, but a negative impression will quickly drive them away.
Painkillers and Vitamins
This reinforcement effect is related to what refers to as painkillers and vitamins in design. Eyal coined the famous analogy that distinguishes products by their user end goal. Painkillers address a need while vitamins are ‘nice to haves’. The conditioning behind these products differs based on their intended use.
Painkillers look to solve unmet customer needs and make users want to use them. These products work to remove adverse conditions (through negative reinforcement conditioning). They become products people need.
- Google Maps removes the pain associated with making a wrong turn, waiting in traffic, or getting the “turn right” command just after you missed the turn.
- Amazon’s Echo becomes the most knowledgable member of the household and removes the need for dozens of other devices — timers, radios, speakers, a computer, etc.
But what behaviors are these products reinforcing? What desired behaviors are they after? Use of course. Like all products, they are designed to be used and the company benefits from their use. To encourage this use they remove or lessen a negative experience for the user.
Vitamins on the other hand help improve a situation by becoming things people should use. These products create more favorable conditions by adding positive reinforcement.
- Duolingo aims to improve the language learning experience by adding rewards like points, badges, and trophies.
- Fitbit reinforces fitness goals by rewarding achievements with points that can be used at retail stores and collecting badges for completing levels.
These products reinforce desired behaviors with direct rewards relying on Operant Conditioning theories that this will increase the frequency of the behaviors and keep users coming back for the same rewards.
Generalization and Similarity
Stimulus generalization is the tendency for similar responses to occur from a conditioned stimulus. For example, if Pavlov’s dog is already conditioned to associate the sound of a bell with food and start salivating, it may start salivating when it hears a sound similar to a bell like a car horn. Discrimination is the ability to determine the difference between these other stimuli but when stimuli are close to too similar to each other, the natural response may supersede the difference.
If one conditioned stimulus is already paired with the unconditioned stimulus then another conditioning will be blocked and therefore ineffective. For example, the dog will not start salivating when it hears the sound of a firetruck siren if it has already been conditioned to associate food with the bell.
In experience design, this means 2 things:
- It is necessary to maintain the right degree of similarity between interface elements;
- But not create elements that look the same (or too similar) and function differently — i.e., a conditioned stimulus with 2 or more outcomes.
Buttons for example should look the same overall but labels should reflect the outcome or action that will result. When a user sees a button, they are conditioned to anticipate that clicking it will cause an outcome. An “Add to Cart” button will add the item to their shopping cart and will not do the same thing as a “Log In” button but both will have a consequence. The text labels indicate the consequence.
This consequence is also conditioned for many labels. The familiarity of these labels and their results mean the user will click them as they have already come to associate the interaction with the desired outcome.
When we deviate from either the conditioned stimulus (the familiar label or button) or from the conditioned outcome (the resulting action) users get lost or frustrated. Imagine if clicking the “Add to Cart” button took you to the FAQ page or if when looking for the FAQ page you instead found a page called “RAQ: Randomly Answered Queries” would you be a bit confused?
This doesn’t just apply to buttons. Nearly all interface elements and features are subject to these conditioning rules. Nothing can be random and reinventing the wheel can easily create confusion so relying on familiar, trusted patterns is key.
Feedback and Extinction
When the frequency of a conditioned response decreases or disappears, extinction occurs. This classical conditioning theory defines what happens when a conditioned stimulus is no longer paired with an unconditioned stimulus. In UX, this effect can happen when users stop associating an outcome with feedback.
When using a product, feedback is the mechanism that informs us we were successful (or in error) and that something has occurred. When used effectively, feedback establishes an association between hearing a ring, feeling a buzz, or seeing a message and our natural reaction to receiving that result.
For example, if you feel stressed by phone calls you will likely come to associate the sound of a phone ringing with that stress. Or, if you are excited about making a purchase, you may come to associate that feeling with the sounds of the card terminal.
But if switched your phone ringer to silent, after a while of not hearing the conditioned stimuli (ringing) when receiving a phone call, you would likely no longer feel the unconditioned stimulus (stress) hearing the same ringing sound. Extinction.
But for other, more pleasurable unconditioned stimuli, is extinction so helpful? Probably not. Rather, products ought to encourage these pleasurable stimuli by maintaining the conditioned responses to keep users engaged and excited.
So what’s the deal with conditioning in UX?
Hint: It probably can be applied to any UX decisions you make in some way.
It’s not simple, and it’s not foolproof, but conditioning plays a major role in the UX design decisions we build into a product, and the way users interact with that product. It’s not always something we can control but those aspects we can control can greatly affect the uptake and usability of a feature. By maintaining similarity and creating differentiation between functions we establish the bounds of learnability so integral to use.
When you start to see and understand how conditioning happens naturally or how to create conditioned experiences, you begin to understand the benefits of monitoring and crafting experiences in a way that makes users want to keep coming back.
