From a learner's perspective, the notion of difficulties is generally not desirable, and from an instructor's standpoint, it may lead to the assumption that simply adding a difficulty or challenge can lead to better learning. "Desirable difficulties, versus the array of undesirable difficulties, are desirable because they trigger encoding and retrieval processes that support learning, comprehension, and transfer," the Bjorks write. "If, however, the learner does not have the background knowledge or skills to respond to them successfully, they become undesirable difficulties. For this reason, it is necessary to consider what level of difficulty is appropriate in order for that level to enhance a given student's learning, and the appropriate level that is optimal may vary considerably based on a student's background and prior level of knowledge."

Therefore, all challenges are not desirable. As such, when applying the concept of desirable difficulties, instructors must be thoughtful and careful in implementing challenges in the training and learning journey.

While learning, humans naturally tend to incline toward tasks that demonstrate their immediate performance during or after training and to avoid tasks that cause discomfort and lead to making more errors. Individuals learn a new skill by keeping the learning conditions constant and predictable, resulting in an immediate demonstration of learned skills (that is to say, performance)—which is a comforting feeling for us all.

For example, someone practicing their golf shots tends to hit the same stroke repeatedly using the same club without much or any delay between strokes. In other words, they keep their conditions of practice constant. That type of massed practice will improve the individual's immediate performance but won't necessarily improve their game in the long run because it lacks the variability required to play golf. They can introduce discomfort in that situation by changing the clubs or adding time between shots, but such changes in conditions of learning and practice lead to more errors due to the added challenge—and humans generally don't enjoy making mistakes. So, they choose tasks that enhance their immediate performance.

Tasks that lead to immediate demonstration of a recently learned skill are easier to complete and demand less effort from learners. As such, they create a perfect mirage of learning. However, learning is not the same thing as performance. Performance is short-lived and hard to re-create with passing time and changing context. The activities that feel harder and demand more effort (think self-testing or adding variability to a task) are the ones that can convert short-term performance to long-term learning. In the presence of such effortful tasks, learning feels sluggish. The essential distinction between performance and learning and a human's natural inclination toward performance-enhancing tasks (which also happen to demand less effort) can help you understand the notion of desirable difficulties.

Peculiarities of the human learning and memory system

When people acquire new knowledge through a training program or by completing an online module, most of them can immediately demonstrate that knowledge in the form of performance, but can they do that as time passes or as the context within which they need to apply the knowledge changes (that is to say, durable and flexible learning)? That is the challenging part. Learners can't achieve those things unless L&D designers change the conditions within which training takes place. Such conditions are what Robert Bjork calls "desirable difficulties" in Metacognition: Knowing About Knowing.

Before exploring desirable difficulties, it helps to first grasp how human memory works as well as how people store information in their long-term memory and then recall it when they need it. In their essay "A New Theory of Disuse and an Old Theory of Stimulus Fluctuation," Robert and Elizabeth Bjork explain that human memory has certain "peculiarities" as to how it stores and retrieves information from long-term memory. Let's look at how the human memory system stores and recalls information.

Step 1: Information enters a person's memory system. Working memory is limited in time and capacity for new knowledge. John Sweller, the creator of cognitive load theory, notes that well-designed instruction, one that aligns with human cognitive architecture, has the potential to enable new knowledge to successfully pass through someone's working memory so it reaches their long-term memory, where the brain stores it.

Step 2: The individual encodes the information. Contrary to popular belief, people do not store new information in their long-term memory as literal recordings that they can play back, but rather by relating that new information to what they already know. In other words, people store new knowledge in semantic (meaningful) ways in terms of what it means to them and how it associates and relates to other information in their memory. Thus, the more often and more ways in which someone stores (encodes), the better.

Step 3: The individual retrieves or recalls the information as a learning event. The act of recalling information from long-term memory changes it, thereby making it more recallable in the future. Robert Bjork calls the act of retrieval (or recall) a memory modifier. In accessing the information, people end up learning it better versus if they didn't try to access it on their own and simply looked at it (think rereading).

The bottom line is that durable and flexible learning is partly determined by how someone encodes and stores the information (how well they understand it) and partly by recalling it (using the power of retrieval as a learning event). An individual can't recall what they don't possess in their long-term memory, but if the information ends up stored in their long-term memory at some point, then an opportunity to retrieve it makes it more recallable.

Conditions that enhance long-term learning

Desirable difficulties are challenges or conditions of learning that an instructor intentionally integrates within the learning and training process that can lead to long-term retention and enhance knowledge transfer. They are desirable because they make learning durable and flexible, and they are difficult because they temporarily slow down the performance by posing challenges to learners. Robert Bjork introduced five manipulations (now known as desirable difficulties) that can foster long-term goals of training.

Distributed practice. Spacing learning over time is a desirable difficulty because the added time between training sessions is a condition of learning that can lead to durable learning. It first slows down learning but produces superior long-term performance.

In contrast, massed practice (that is to say, when a training event happens in one long session) produces better short-term performance but a much inferior long-term recall. Despite the robust evidence in favor of the spacing effect, training design continues in the form of massed practice, as evident from full-day, off-site training sessions or participants blocking off a chunk of their time to complete mandatory online training in one sitting instead of spreading it over various sittings.

Retrieval practice (testing). Retrieval practice is the act of someone recalling what they have learned without having it in front of them. The act of recall modifies the memory and enhances its storage strength. That testing effect demonstrates that someone retrieving what they've learned previously from memory is considerably better for long-term retention than restudying or rereading.

Paul Kirschner and colleagues write in How Teaching Happens, "As is the case with the other desirable difficulties, using tests rather than restudy or increasing the difficulty of such tests, may seem counterproductive during learning as measured by a test at the end of the study session, but leads to better recall after a delay."

Pooja Agarwal, creator of the site RetrievalPractice.org and co-author of Powerful Teaching, shares that retrieval practice is a powerful technique for enhancing long-term and lasting learning. Some of the more common forms of retrieval practice include flash cards, quizzes, self-tests (ideally low stakes), brain dumps (quickly recalling and documenting relevant information covered in a previous training session), and think-pair-share (a collaborative learning strategy in which someone recalls what they know, discusses it with a partner, and the pair shares the collected knowledge with a larger group). Any form of retrieval practice works best when instructors use it as a tool for enhancing learning and not as an assessment strategy. Retrieval practice can serve as an effective learning booster in training sessions.

Varying conditions of practice (interleaving). By introducing variation in a training session, instructors add unpredictability for learners, keeping them on their toes. That involves randomly scheduling practice tasks rather than blocking them by task type. For example, interleaved practice during a customer service training program could include alternating scenarios such as handling complaints, addressing inquiries, and processing orders, much like practicing golf using different clubs and hitting different shots (as someone would while playing the game).

While challenging, interleaved practice produces stronger long-term results versus if an individual chooses to train using blocked practice—for instance, the golfer hitting first a putt stroke 10 times and then a drive shot 10 times and so on. There is an important caveat when integrating interleaving within training programs: Interleaving entails mixing up different yet related concepts, meaning the added variation must not be so different that it creates too much confusion for a learner.

Contextual interference. A form of interleaving, contextual interference is when the task environment—instead of the task itself—becomes more variable or unpredictable. Integrate that condition of learning within training design by changing the context within which a learner needs to apply acquired skills. For example, adding certain interfering conditions, such as varying a task's difficulty level, can slow down the process of gaining mastery of the concept.

So, for the customer service training program, integrate contextual interference via role-playing scenarios that require trainees to adapt their response to different customer personas (some easy-going and others more challenging) and situations. Despite a learner understanding the concept, enhanced difficulty levels can lead to reduced performance at first but has shown to improve retention and transfer performance in the long run. Contextual interference promotes adaptability, deeper understanding, and better retention of skills by exposing learners to diverse contexts or variations during practice.

Reduced feedback. This is yet another unintuitive condition of learning that initially makes life more difficult for a learner during a training event but can enhance their post-training performance. Immediate, frequent, and accurate feedback helps with skills acquisition and supports short-term performance, especially for novice learners. However, Robert Bjork notes that frequent feedback can guide learners toward expected performance and even create a dependency where they rely on the feedback to guide their behavior. Therefore, reducing feedback as a learner develops a certain comfort level with newly learned skills can lead to durable and flexible learning.

Desirable difficulties often face obstacles when it comes to integrating them within a training program. Society is fascinated by performance and efficiency, often achieved through fast and somewhat effortless learning. Desirable difficulties, however, tend to slow down performance speed and can initially introduce errors. But research strongly suggests that once L&D teams progress past that discomfort, they can support learners in achieving long-term learning. As Robert Bjork writes, "For people to be receptive to the types of manipulations of training suggested herein, institutional and individual attitudes toward the meaning of errors and mistakes must change."

Not all difficulties are desirable

Desirable difficulties are only desirable when they trigger encoding and recall that can further support learning and transfer. For example, L&D can only interleave when the learner has first mastered foundational skills, or we can only ask them to retrieve something that they have encoded. If the learner does not possess the necessary background knowledge or skills, it is an undesirable difficulty no matter its level. The challenge should be fair and not outside the training content's scope because an undesirable challenge is overwhelming for learners.

To integrate desirable difficulties in training programs:

• Space it out, make use of micro-learning and remind participants to avoid massed practice.

• Make use of quizzes, polls, and brain dumps to enhance learning.

• Mix up the tasks once your participants have mastered individual skills.

• Pull back on the feedback when the time is right.

The New Theory of Disuse

In 1992, Robert and Elizabeth Bjork developed the New Theory of Disuse framework based on Edward Thorndike's 1914 law of disuse, which states that if someone does not use or access learned habits, they will fade or decay from memory. The framework further assumes that two strengths characterize an item in memory: Storage strength relates to how well someone has encoded and stored the information during the learning process; while retrieval strength refers to the current ease of access to the item—meaning, how easily the person can retrieve information from their memory based on available cues.

Once someone learns an item and encodes it in their memory, they never lose it. Note that retrieval strength is usually high right after an individual encodes new information in their memory, which makes sense. That immediate demonstration of newly learned knowledge is performance, but it is short-lived. As time passes, the retrieval strength decreases and, therefore, performance appears to decline.

That is when the temptation to simply reread the information is high. When someone rereads or re-studies something, they get the illusion of an increased retrieval strength, and they tend to believe that they also increased the storage strength of the learned information. However, that is not the case. The higher the current level of retrieval strength of an item in memory, the smaller the gain in storage strength. The harder it is to retrieve, the bigger the gain in its storage. Put simply, tasks that seem to slow learning down are the ones that will end up making it durable and flexible, and that is the basis of desirable difficulties.

The Undesirable Effects of a Popular Term

Robert Bjork coined the term desirable difficulty in 1994. In a chapter titled "Introducing Desirable Difficulties Into Practice and Instruction: Obstacles and Opportunities" from the 2023 book In Their Own Words, Robert and Elizabeth Bjork report "sometimes" regretting adopting the term to designate those conditions of study and practice. They write that despite the nice alliteration, the phrase can be problematic.