Introducing Difficulties and Errors Can Enhance Instruction
Dasha Bunks
Liberty University
PSYC 775 - Teaching of Psychology
Dr. Winn
2022
Introduction
Should errors be avoided during the learning process, or are they an important component of
effective learning? The merits of errorful and errorless learning have been debated intermittently
throughout the history of psychology and education research. Both positions seem plausible: The
argument that making and correcting errors is one key to learning seems compelling, but so does the
argument that it is better to avoid letting students make mistakes lest their errors impede learning of
corrective feedback and/or reemerge at a later time.
Virtues of Errorless Learning
Proponents of errorless learning suggest that learners should be safeguarded against making
mistakes. Skinner (1958), for example, when recommending how technology should be used in the
classroom, argued that learning should be carefully scaffolded to avoid having learners make mistakes.
When learning how to spell the word ‘manufacture,’ for example, a student would first copy the word
and then fill in missing letters in several iterations before writing the entire word from memory. Taking
very small steps ensures that learners do not make errors, as each step is trivially more difficult than the
last. Perhaps the most adamant supporters of an errorless learning perspective are learners themselves.
Part of the appeal of rereading either entire texts or texts that have been previously marked may be that
there is no possibility for learners to make errors, given that they are able to review the source exactly
as presented, often in whole. Various types of text marking, such as highlighting and underlining text
materials, are also common and errorless learning strategies used by students (Kornell & Bjork, 2007;
Hartwig & Dunlosky, 2012). As teachers, we may also be hesitant to create difficulties for our students.
In fact, we may think that it is our responsibility to make learning easy for students and protect them
from making mistakes.
Virtues of Errorful Learning
Research on introducing what Bjork (1994) labeled “desirable difficulties” for learners, questions,
however, the virtues of errorless learning. Desirable difficulties are manipulations of the conditions for
learning or practice that create difficulties and challenge for learners and tend to increase, not decrease,
the frequency of errors during the acquisition process, but then often enhance long-term retention and
transfer. Overcoming such difficulties, which include spacing rather than massing study opportunities,
using tests rather than presentations as study events, and interleaving rather than blocking practice on
separate to-be-learned tasks, appears to engage processes that support long-term retention and
transfer.
Learning versus performance
To establish why eliciting difficulties is effective, it is useful to think of learning along two
dimensions: current performance and “underlying” learning. Imagine a student who is able, on a
Monday, to name all of the countries in Europe on a map. When he comes back to school on Tuesday,
will he be able to fill in a map again, or will Luxembourg and Malta slip his mind? If he fails Tuesday’s
test, you would say that Monday’s success represented current accessibility (perhaps he had just
glanced at the answers seconds before), but not lasting learning. But if he could fill in the map with ease
on Tuesday, then again for the final unit test a week later, we would agree that he had a good grasp on
the material. Monday’s performance, then, could be misleading, because it could reflect either lasting
learning or merely current accessibility, which can be quite fragile.
High current accessibility is not, therefore, a reliable index of learning. It is also the case that low
accessibility does not mean that learning has not occurred. Students can be learning even when a
current test shows little or no progress. One consideration is that memory is cue dependent—that is,
what you can remember depends on how many cues or “hints” you have. For instance, given a blank
sheet of paper, drawing a map of Europe and labeling each country would be a very difficult task. After a
study session on country names and locations in Europe, therefore, a student might do very poorly given
a blank sheet, but do very well if the countries were outlined and the task was to fill in each country’s
name. So despite continued poor performance on a difficult test, improvement on an easier test could
show that a student learned something from a lesson. In sum, when we seek long-term learning, current
accessibility is not a reliable index of how effective a lesson has been.
The relationship between learning and accessibility
The distinction between accessibility and learning, or performance versus learning, is a time
honored one in research on learning (for a review, see Soderstrom & Bjork, 2013). Estes (1955b), for
example, pointed out that all learning theorists of that era had to distinguish between what he labeled
habit strength and what he labeled response strength. In a theoretical framework Bjork and Bjork (1992)
refer to as a “New Theory of Disuse,” which led to the desirable-difficulties idea, they resurrect such a
distinction—using the terms storage strength (learning) versus retrieval strength (accessibility).
Bjork and Bjork argued among other things, that when accessibility is high it is difficult to make
further gains in underlying learning (i.e., storage strength). For example, repeatedly rereading text or
looking at a formula right before implementing that formula may keep accessibility high during the
learning process, but such rereading is often in vain (see, e.g., Callender & McDaniel, 2009). If
accessibility can be reduced, however, learners can make bigger gains in underlying learning.
Introducing difficulties into learning—thereby reducing accessibility—removes the learner from the
experience of having the to-be-learned material at hand. Overcoming the difficulty through subsequent
practice is then more effective and can create a new, higher, level of learning.
Decreasing accessibility need not involve errors per se. Accessibility lies on a continuum. Some
state capitals, for instance, are likely very accessible for a given individual. If you are from California and
someone asks you to name the capital of California, for example, “Sacramento” will pop to mind
immediately. The capitals of other states, however, may be known but less accessible. You may know
what the capital of Michigan is, for instance, even if it does not come to mind immediately. In searching
your memory several cities may come to mind, such as Detroit, Ann Arbor, and Lansing. After pondering
for a moment, you may realize that Lansing is in fact the correct city. Decreasing accessibility, then, only
means making information harder to retrieve. When training conditions lower accessibility below
whatever threshold is needed for a successful retrieval on a particular test, an error will occur, but
learning in the storage-strength sense can still remain—as evidenced, for example, on a later test that
provides a lot of cue support, or, perhaps, by accelerated relearning.
