Applying Cognitive Load Theory to English part 6: The Expertise Reversal Effect


efficiency in learning cover

Cognitive Load theory-and specifically the ‘worked example effect’-is dependent upon treating students as being somewhere upon a continuum between ‘novice’ and ‘expert’. The difference between expert and novice is more than a mere labelling exercise and has real implications for instructional and curriculum design.

Novices learn best by studying worked examples, model answers that build up their schemas and background knowledge and effectively act as analogous solutions to subsequent problems. While worked examples are optimum for novices, they can actually hinder the learning of more competent learners. In Efficiency in Learning, the authors point out that ‘once a learner has acquired a basic schema for the skill or concept, he learns best by applying the schema to problems, rather than investing redundant effort in studying more worked examples’. As well as being a potential waste of time for the more competent student, it is also the wrong instructional choice: he or she would be better attempting problems in order to ‘solidify his or her schema by exercising it.’

Expert or Novice?

So how do we decide who is an expert and who is a novice, who is best served by completion problems or worked examples and who will learn better from problem solving?

Although we can debate exactly what makes someone an expert or a novice in a particular domain, we could reliably say that there will be a difference in the amount of knowledge that they have: novices have little prior knowledge; experts have a lot. I would tentatively place most secondary students towards the novice end of this continuum, although as they make progress within a specific area of study, they clearly begin to gain in expertise. According to Clark et al, ‘prior knowledge is the one individual difference that has been consistently shown to interact with different instructional methods’.

If the authors of Efficiency in Learning are correct, then the simplest and most obvious solution is to assess prior knowledge before instruction and group the students accordingly. If students are grouped according to levels of prior knowledge, then we can make informed decisions at a group level about which particular instructional choices should have the greatest efficacy. Although ability setting is a fiercely contested area (Mark McCourt has written an interesting piece on the topic here), Engelmann would also argue that setting students according to ability is of fundamental importance in order to maximize the efficacy of instruction. From my own experience, I have found teaching groups that have been set by ability to be more efficient and effective. Interestingly, the advantage of ability grouping is most apparent when teaching the hierarchical parts of my subject like grammar. If, as Clark et al assert, it is essential to match the instructional approach to the competence of the student, then the expert reversal effect seems to lend further support behind ability setting (as long as ability setting refers to a student’s level of prior knowledge).

How does this impact upon Curriculum design?

Even if you have begun by setting students according to their level of background knowledge, there will still be a wide disparity within a class. When an instructional sequence has begun, there will almost certainly still be a need to choose the optimum method of instruction based on feedback to the teacher as to how students are performing. An important part of not just teaching but also curriculum design is the process of evaluating and refining what is being taught and the sequence that is being used.

Whether a student has been successful or has produced the desired behaviours that you intended is an empirical question that can be answered by looking at student output. If you are teaching a particular sentence style and, after your instructional sequence, it is clear that students have not mastered what you have taught, then it is likely that your instructional sequence is defective and needs adjusting. While student failure could be down to students generating misrules as a result of misinterpreting your communication, it could also be a result of a mismatch between the level of student expertise and the chosen instructional approach. Sometimes this evaluation happens at the end of a unit and can be used to refine the curriculum so that it is more effective when you teach it again. More often, it will require an adaptive and pragmatic approach where the teacher changes their teaching within a lesson or unit in order to ensure that students succeed. Perhaps their low level of prior knowledge means that they need to see further worked examples before attempting guided practice. Perhaps their struggle and lack of comprehension means that the teacher should postpone freer practice and wider application until students have developed and retained the necessary background knowledge needed to complete a task. This is the essence of responsive teaching: adapting instruction based upon evidence. The broad continuum from worked example to completion problem to application (or I-We-You) provides a framework for the teacher, allowing them decide upon the most suitable instructional choice for their students.

3 framework table with expert novice


Following Engelmann’s guidance, the curriculum should be methodically planned out so that each step forward is both manageable and small enough for all students to experience a high success rate. Instruction should broadly follow an I-we-you format, gradually fading out support in order to account for the incremental development of expertise. If learners are deemed to be novices, then at the beginning of the instructional sequence-the ‘I’ stage-worked examples and the alternation strategy should be used in order to provide ‘schema substitutes’, helping them overcome the limitation of their working memory and slowly building their background knowledge. Towards the end of an instructional sequence-the ‘you’ stage-learners should increasingly be asked to apply their knowledge to problems. By retrieving and applying their developing background knowledge, learners will hopefully develop in automaticity as well as increasing the storage and retrieval strength of their schemas.

With our curriculum, all classes receive the same resource booklets and the adaptation and instructional choices are made by the teacher so that they best fit the level of student expertise. As would be expected, Year 7 booklets are simpler than year 11 but we have not gone down the route of creating different iterations of the same booklet in an attempt to match the level of expertise of the class, as this would be an unsustainable approach that would be far too time consuming. Top and bottom sets use the same booklet. Instead, the teacher of a higher level classes may choose to skip a worked example if it is deemed unnecessary. The teacher of a lower level class will be careful to use a worked example, aware that this is the optimum approach for their novice students. At the end of year 11, the systematic usage of worked examples slowly begins to fade away and students are asked to solve problems independently. Some classes may get to the requisite level of expertise needed to complete independent problems before this point: if so, the teacher can choose to ignore the worked examples in the booklet. There is an important balance here: maximizing the efficacy of the instructional materials whilst keeping the workload sustainable.

What else can we do to accommodate different levels of learner expertise?

Differences in learner expertise are not only accommodated by deciding between worked examples and problem solving. Here are some other recommendations from Efficiency in Learning.

  1. Text Coherence

When creating texts for novice learners, we should be investing time in making them as clear as possible. A 1996 study compared the learning of high and low prior knowledge students after reading high and low coherence texts on heart disease.

text coherence

In figure 10.3, you can see the low and high coherence texts. The more coherent text replaces pronouns like ‘it’ with the appropriate reference such as ‘heart’. Extra sentences have also been added in order to explain and clarify. There is also a heading, making the text easier to follow. While you might expect additional clarity and coherence to be of benefit to all types of learners, the results suggested otherwise:

text coherence graph results

High prior knowledge students (experts) who had studied the text with low coherence performed better on a problem-solving test than those who had studied the high coherence text. Low prior knowledge students (novices) performed better when studying a high coherent text. However, despite this finding, the authors assert that ‘we need more research on text coherence before we recommend writing low coherent texts for more experienced learners’.

A further recommendation is that text should contain definitions and examples of unfamiliar terms. In our booklets, we highlight and define high-utility tier 2 words as well as words that are of central importance for understanding the text. Here is an example from An Occurrence at Owl Creek Bridge by Ambrose Bierce, a short story that we teach in year 9:

owl creek2

Although Clark et al say that we should make texts as easy to follow as possible so that novice readers can access them, this poses a problem because it means less challenge. If reading was solely about communicating and acquiring literal information then making texts as clear, coherent and simplistic as possible would be ideal. Reading texts is not like this though: students need to encounter dense and complicated texts where meaning needs to be teased out and several reading are often required before clarity is reached. Students need to get used to following chains of pronouns in a lengthy exposition or narrative. One of the core arguments in Lemov’s Reading Reconsidered is that we should be exposing students to complex and challenging texts if they are to develop and improve as readers. With the help of explicit instruction, whole class reading and teacher led annotation, complex texts can be unlocked for students. A teacher can provide the additional explanation or elaboration that is necessary. A teacher can stop, question and check that students are following if the text has become labyrinthine.

reading recon

2) Interruptions

Clark et al argue that we should not interrupt weaker readers or they will quickly become confused. While this makes a lot of sense as it will hinder their comprehension, it is often necessary, particularly if you are reading a challenging text that requires additional explanation or frequent checks for understanding. In Reading Reconsidered, the authors look at lots of different ways to read a text and make the point that the more challenging a text is, the more likely you will read it using ‘layered readings’, reading it several times with slightly different approaches. Typically, the first read will be a ‘contiguous read’ which is reading with minimal interruptions so that students get the gist or overview of a passage. Subsequent reads will be more methodical and forensic, involving annotations to unlock challenging vocabulary or teacher led questioning to ensure that students understand. Layered readings can help novice learners to access challenging texts.

3) Eliminate redundant content for more experienced learners

When teaching more experienced learners-students who are closer to the label of expert-then we should be thinking about whether information is necessary or redundant. Do students need a diagram as well as the text? Is one or the other sufficient?

diagrm plus text graph results

The table above displays the results from a 1990 study where the researchers wanted to investigate the effectiveness of diagrams when they were added to text on high and low prior knowledge learners. Participants were learning about mechanical processes, including the operations of brakes, pumps and generators. Before the experiment, students completed a survey to ascertain their prior knowledge levels and were then grouped accordingly. As you can see from the results, novice students performed better when presented with text and diagram. According to Clark et al ‘the diagram provided a schema substitute for low prior knowledge learners’. In contrast, the more expert students did not benefit from the addition of the diagram as they ‘had sufficient schema to make sense of the text alone’. This suggests that, for experts, the diagram was redundant information here.

In another study (Kalyuga, Chandler and Sweller 1998), researchers used a staged experiment to evaluate the effectiveness of two different lesson versions designed to teach how to interpret the circuit diagrams of a motor. The researchers were interested in whether there was a different effect for novices and more experienced learners. Comparing text alone with text plus diagram, the researchers tracked the effects across three learning stages as the learners gained in expertise. Here are the results:

diagram results graphs 108

Figure 10.8 demonstrates that at stage 1, the novice students performed better when learning via diagram plus text. Those students who only had a diagram performed worse. At stage 2, learners had learnt more and increased their background knowledge. At this point, both approaches were equally effective. Learners had enough knowledge to benefit from studying using only a diagram but not enough knowledge to be disrupted or hindered by the combination of both diagram and text. As they developed and learnt more, the text became redundant and by stage 3, learners were performing better when studying from only a diagram. This demonstrates the ‘expertise reversal effect’: a strategy that began as optimum for novices (diagram plus text) became redundant as the students developed in expertise.

So how can we apply these findings? Clark et al recommend that ‘when the text alone is self-explanatory, it is more cost effective to eliminate diagrams in lessons designed for more experienced learners. In contrast, if the diagram is essential to the task, drop the text.’ The optimum choice will be based upon the requirements of the content that you are teaching: if you need to display information in a diagram, then do so!

In Summary:

  • Most students are probably more likely to be ‘novices’ and therefore worked examples and completions problems should feature heavily in instructional sequences, particularly in the early stages.
  • Teachers should be aware of the I-We-You continuum and that each stage best suits a broad level of student expertise and background knowledge. Although our assessment of ‘expertise’ is always going to be woolly, being aware that different instructional choices are likely to fit different levels of expertise could help teachers choose more appropriate strategies.
  • Making information simplistic and easy to follow will almost certainly help students to understand. However, the teacher can act as a valuable resource to help students unlock and access more challenging texts.
  • We should think carefully about whether information is necessary or redundant and choose the format that best suits the requirements of the subject

Next Post: A Pedagogical Overview: Recommended further reading

One thought on “Applying Cognitive Load Theory to English part 6: The Expertise Reversal Effect”

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: