Teaching the science of learning
Dr Alicia Shaw, Innovation and Evaluation Lead at the IEE
Many teachers are really interested in the field of cognitive science in education at the moment, and with good reason: as the demands placed on memory by the curriculum are increasing, findings from cognitive science offer potential solutions to the difficulties many pupils face remembering all of the information needed in high stakes exams. A number of organisations, including The Learning Scientists, make evidence from cognitive science accessible to educators. But evidence from psychology and cognitive science is often criticised for being carried out with unrepresentative participants learning in artificial situations. Just how applicable is this research to pupils in primary and secondary classrooms in the UK?
To try to answer this question, I looked at Teaching the Science of Learning, Weinstein, Maden and Sumeracki’s 2018 review of the evidence behind the six learning strategies presented by The Learning Scientists – spaced practice, retrieval practice, elaboration, interleaving, concrete examples, and dual coding (these strategies are not described in this blog, but are explained in depth on The Learning Scientists website). I identified 66 papers in their reference list which reported original research that investigated how learning outcomes were influenced by characteristics of the learning materials or the use of strategies based on cognitive science. I should be clear that this was not a systematic review of research into the application of cognitive science in education or even the evidence presented by The Learning Scientists, and that Weinstein et al do not claim that their review is systematic or comprehensive. These 66 papers were merely a convenient (and hopefully fairly representative) sample of some of the evidence which supports the strategies recommended by the Learning Scientists.
I looked at the 66 papers to try to answer these questions:
- Who participated in the research?
- What did participants learn?
- Which strategies have been trialled with school-age participants? Is there any evidence that they were effective?
One of the criticisms leveled at cognitive science is that the research is often carried out with undergraduate students who may not be representative of the general population, so the first thing I did was to look at the ages of the participants involved in each research paper. Of the 66 studies reporting original research, 21 were carried out with pre-school or school-age participants (aged between two and a half and 18 years old).
The fact that 32% of research was carried out with school-age participants was higher than I’d expected, but it still means that over two thirds was carried out with university students and adults. Where universities were named they were generally in the US and many were highly ranked in international university ratings (such as Harvard, Dartmouth and UCLA). It is therefore likely that university samples are biased towards American students who had been successful in learning at school.
Another frequent criticism of cognitive science research is that it uses artificial tasks, like learning lists of unrelated words. I looked at the information participants learnt in the 66 trials and classified them as follows:
- Relevant to current learning: participants were taught something they would be expected to learn as part of their academic studies. The intervention was often (but not always) introduced to timetabled lessons or lectures.
- True but not relevant to current learning: participants were taught information which, while factually correct, wasn’t part of their curriculum. Examples include learning facts about unfamiliar counties and non-chemistry students learning about atomic structure.
- Artificial: participants were asked to learn information which had been created for the experiment such as vignettes and word pairs.
The types of activities being carried out by participants of different ages in the 66 research papers can be seen in this graph:
While the university students and adults did learn a lot of artificial or irrelevant information, the majority of research with school-age participants related to learning information which was relevant to pupils’ current curriculum in fairly realistic learning situations. This is not surprising: cognitive scientists use easily accessible undergraduates to carry out highly controlled, artificial experiments to develop theories, but when they apply these theories in schools they want to know whether the approach can have a practical impact. There is also a pragmatic reason for this difference: it is harder for researchers to get approval from ethics boards, funders and gatekeepers (such as headteachers) to carry out research that asks school-age pupils to learn artificial or irrelevant information during school time than it is to use similar activities with university students or adults.
I was particularly interested in the types of learning strategies tested with school-age participants and the findings of these studies, so I looked in more depth at the 21 papers reporting research with school-age pupils. The following strategies were used in these studies:
Nearly half of the research carried out with school-age participants tested retrieval practice, with elaboration being the second most researched strategy; research into these strategies was fairly well spread across primary and secondary-age pupils. Very little of the research with school-age pupils included in Weinstein et al’s review reported on the impact of using dual coding, spaced practice or concrete examples.
What did these studies in schools actually find?
Broadly speaking findings from research with school-age participants were similar to research with adult populations, with evidence that retrieval practice and elaboration in particular may benefit school-age learners. In line with research carried out with adults and university students, feedback can be beneficial when pupils do not recall information correctly during retrieval practice. A study which used multiple-choice tests found that feedback following retrieval practice helped stop the 6- to 8-year-olds participating in the research from remembering incorrectly recalled information as facts and increased the likelihood that these children would remember correct answers which they hadn’t been very sure about in the initial test.
The research on interleaving was less clear – only three studies were included in the review and one of them had mixed findings, so these papers do not provide a convincing confirmation that findings from university and adult participants generalise to school-age learners. And not enough research into spaced practice, concrete examples and dual coding was included in the review to draw conclusions about the potential effectiveness of these strategies in schools.
Where these approaches were found to be effective some modifications to the way they are applied are likely to be needed for students of different ages. For example, research by Karpicke, Blunt, Smith and Karpicke (2014) into retrieval practice highlights the importance of ensuring that instruction is appropriate to student age, skills and understanding. As a former Year 3 teacher, it did not surprise me that the 9- to 11-year-olds in this study found it very difficult to complete a free recall activity. Free recall (writing everything you remember about a topic) is a strategy which undergraduates can use effectively as a form of retrieval practice, but as these elementary school students could not complete the free recall activity they did not experience retrieval practice and so the activity did not aid later recall. But when these students were given additional structure, either by providing them with partially completed concept maps or through guided retrieval, the retrieval practice effect seen with undergraduates was found in this younger sample. Findings from individual pieces of research into elaboration, dual coding and concrete examples with school aged participants also suggest that taking into account the expertise of the learner and the learning objectives of the task are likely to be important when applying cognitive science in the classroom.
I was pleasantly surprised by the proportion of research included in Teaching the Science of Learning that had been carried out with school-age participants, that most of this was carried out in fairly realistic learning situations and that it broadly supported conclusions from research carried out with university students and adults. The research referenced in this review suggests that at least some learning strategies based on cognitive science may work in a similar way with learners of different ages, but that materials and content may need to be adapted to suit the age, skills and knowledge of the learner. This suggests that it is important for teachers to understand the theory and evidence well so they are able to adapt strategies faithfully to match the current skills of pupils in their class.
However, more research is needed into the use of these approaches with school-age learners, especially interleaving, dual coding, spaced practice and concrete examples, before we can draw any conclusions with confidence. Of course this research may already have been done but not included in the Weinstein et al review; it needs to be remembered that I am not presenting a systematic review and additional research may have been carried out in this area. As this is an area of research many teachers in the UK are using in their classroom, a systematic literature review of the application of strategies from cognitive science with learners below the age of 18 would be useful to ensure teachers can make decisions based on all of the available evidence. This would review all the evidence in the area, whether findings were positive, negative, or neutral, and help add richness to the guidance for teachers when implementing these strategies.
If you would like to know more about systematic reviews, and how to read and interpret research and advice based on research, take a look at our Engaging with Evidence guide.
At the Institute for Effective Education we would always recommend evaluating the impact of a new approach, or an adaptation of an existing approach, before rolling it out across the school, especially where the evidence is limited. This report of an evaluation of the impact of retrieval practice in primary science lessons is an example of how this can be done. Look at our website for information about in-school evaluation and how we can support you with this. Even if you are not able to carry out a formal evaluation of the approach, you should certainly monitor implementation and impact closely to ensure that your students have the best chance of benefiting from the new and exciting research in the field of cognitive science.
A very big thank you to Megan Sumeracki for commenting on a draft version of this blog.