What we already know determines what, how, and how well we learn

Paul A. Kirschner & Mirjam Neelen

No two people are the same, not even identical twins. Yes, even two people with exactly the same genes, growing up in exactly the same household, going to the same schools and possibly in the same classrooms with the same teachers, and so forth aren’t the same.. This means two things. First, our genes play a large role in how well we do at school with identical twins scoring very similarly with respect to learning and achievement (Krapohl et al, 2014; Shakeshaft et al, 2013). Second, even identical twins are not the same when it comes to learning and achievement. If this is the case, what about a ‘normal’ class with 20-30 different children? The fact that learners are different and don’t always have the same level of prior knowledge can lead to problems when they need to learn something new. But there’s a solution to this problem, and [spoiler alert!], it’s not personalised instruction!

Now, imagine the following. You’re walking around a school or a company and you see a setup with lots of glassware, tubes, bubbling liquids… If you never took a chemistry course, then you probably have no idea what’s going on except that maybe they’re filming some science fiction film with a mad scientist in the starring role.

If you took chemistry in high school, you might remember that this has something to do with distillation; separating two or more liquids (like gin, whisky, or gasoline) via evaporation and condensation. If, however, you studied chemistry at college, then you might see that the distillation setup is being used for a specific purpose as you see on the water bath instead of a Bunsen burner that the temperature has to be carefully limited to a temperature below 100 degrees Celsius (212 Fahrenheit) and the temperature gauge gives a reading of 78.33 degrees Celsius (173 Fahrenheit) which tells you that ethanol is being distilled. In other words, what you know determines what you see and not the other way around (Kirschner, 2009).

In 1968, the American psychologist David Ausubel wrote an article in which he stated that the most influential factor for learning new things is what the learner already knows. In other words: prior knowledge. According to him, new things that we want or need to learn need to be connected to what we already know. The knowledge we already have is stored as schemata (also called schemas) in our long-term memory. Schemata are basically mental frameworks in which we can integrate new information. They’re like the clothing rack in the hall where there are ‘hooks’ that we can use to hang new information. Without these hooks, we have nothing to which we can connect the new information. Ausubel came up with the concept of using ‘advance organisers’ to make the process of connecting new information to prior knowledge easier. An advance organiser, according to Ausubel, is a text or presentation at a higher level of generality, inclusiveness and abstraction presented before the learning event and which forms a conceptual framework for learning the new information.

Our mental schemata are hierarchically organised, from more generic to more specific concepts. New, to-be-learned knowledge gets integrated in existing structures (assimilation; new knowledge and experiences become integrated in existing structures in our brains) and changes the schemata at the same time (accommodation; new knowledge and experience ensure that the structures in our brains are accommodated). This idea was put forth by Jean Piaget in discussing the cognitive development of children.

For example, if learners have previously learned about how a gasoline-powered car works and now start learning about cars fuelled in other ways, they’ll add those cars and their engines to their knowledge schema with knowledge about how cars function (assimilation). In addition, they’ll adjust those schemas, because now, for example, they see that a diesel engine is very similar to a gas-powered engine except that the latter has sparkplugs and the former has glow plugs which function differently (accommodation). This way, learning is a continuous interaction between what someone already knows and what they learn.

Sometimes, this is called assimilation theory and sometimes it’s called subsumption theory. The following figure (tries) to illustrate Ausubel’s assimilation theory.

With respect to subsumption theory, the following figure shows the four types of subsumption involved in advance organisers: derivative, correlative, superordinate, and combinatorial. Derivative subsumption is when you add new things to existing cognitive structures, linking them to concepts already known. Correlative subsumption is when you add new details to what the you already know, usually a higher-order concept. Superordinate subsumption introduces a new higher-level concept into which already existing categories can be integrated. Finally, combinatorial subsumption is when ideas are linked (combined) between higher-level concepts such as when one knows form physics, for example, that stationary air-spaces insulate helps to better understand the function of hair or feathers in keeping certain animals warm.

There are, basically, four types of organisers that can be used prior to instruction namely expository, narrative, skimming and graphic organisers. Expository organisers provide descriptions of new knowledge that learners will need to understand what follows and is often used when the new learning material is relatively unknown to the learner by relating the new information to what is already known. Narrative organisers present new information in a story format to the learners to activate background knowledge so that learners can make connections to what they know, often creating a personal connection to inspire learning. A skimming organiser gives a helicopter overview of the new learning material, focusing on and noting what stands out in the new material such as headings, subheadings, and highlighted information. Finally, graphic organisers include different types of visuals such as concept maps, pictographs, Venn diagrams, and so forth.

Below are two examples of graphic organisers. One is around key concepts of clouds and how they relate, the other is about rocks.

The importance of prior knowledge can be illustrated through the following meta-analysis of the effects of prior knowledge in a certain domain (domain-specific prior knowledge) that’s available as preprint (Simonsmeier et al, 2018). The article at this moment includes 240 articles reporting 4327 effect sizes obtained with 62,129 participants.

First of all, the researchers have found a strong correlation between learners’ prior knowledge and how much they learned (the more prior knowledge, the more they learned). Specifically, they found a very high correlation between prior knowledge and knowledge at post-test (r+ = .521, 95% CI [.491, .550]). Of course, correlation doesn’t equal causation. For example, there’s a correlation between clouds and rain, but that doesn’t mean that rain causes clouds. However, in this case, the researchers were able to demonstrate a causal relation, thanks to the large number of randomised research with a control group. So, “the association of prior knowledge with post-test knowledge is causal and cannot entirely be attributed to a confounding influence of intelligence” (p 20). They explain that the high correlation (not the causation) indicates a high stability of individual differences in knowledge from before to after learning. This finding supports theories that emphasize that we build knowledge over time in an accumulative manner.

Even more interesting are the findings around the compensatory effect as well as the Matthew-effect of prior knowledge. When learners with little prior knowledge acquire more new knowledge through instruction than their peers with more prior knowledge, the differences between the two groups become smaller. Decreasing the gap between learners’ achievements – in this case thanks to instruction – is called the compensatory effect. However, if at the same time the learners with initially more prior knowledge also continue to acquire more knowledge, then the differences actually increase. This is called the Matthew-effect.

The reading example in the figure below shows that initially there’s a group of learners with more prior knowledge (foundational reading skills) and a group with lower prior knowledge. Because both groups receive continuous instruction, the gap between them widens. The ones who initially had more prior knowledge improve faster than the ones who initially had less prior knowledge.

When comparing the compensatory effect with the Matthew effect, we can see a difference. The researchers show that the compensatory effect can mostly be seen with instruction demanding low cognitive effort from learners (remembering and automatizing facts, following familiar processes, practicing with routine-based solutions, etc). In contrast, the Matthew-effect is mostly seen with instruction that demands high cognitive effort, such as figuring out connections themselves between concepts, analysing, explaining, and drawing conclusions (Stein & Smith, 1998). These types of tasks, the ones that require higher cognitive effort, are easier for learners with more prior knowledge.

Kirschner, Sweller, and Clark (2006) explain why this is. If the learner doesn’t have the basic ‘hooks’ (i.e., relevant prior knowledge) to hang the more complex information on, then they’ll start swimming using trial and error and then drown very quickly as they don’t know how to do it effectively. In other words, the foundational knowledge that the learner needs to do the more complex stuff, first need to be made explicit to the learner and the instruction used needs to accommodate this. Advance organisers are one way of helping the learner create those hooks, as long as the learner takes some time to process and understand the information presented in the organiser itself and of course, the organiser also must indicate the relations among the basic concepts and terms that will be used.

Another way of making new foundational knowledge explicit are worked examples (see our blog here) that then move to fading and partially worked examples (a worked example where you leave small steps that lead to the solution open, which learners then need to complete themselves).

What it all comes down to is that these results are a clear plea for direct instruction. Learners with little prior knowledge benefit from instruction with lower cognitive requirements (more guidance / scaffolding), as that type of instruction requires less from working memory. The available space can therefore be used to acquire, interpret, and process the new knowledge. For learners who do have prior knowledge, higher cognitive demands ensure desirable difficulties (Bjork & Bjork, 2014). These desirable difficulties challenge the learners who already have prior knowledge to deepen and extend their existing knowledge. Learners with little prior knowledge don’t meet the prerequisites to be able to do this.

The conclusion is that, as an instructor or learning designer, you can decrease the knowledge gap between learners through strong direct instruction. This way, you first give them a solid knowledge foundation through which you ensure a compensation effect. When they’re ready, you can then start instruction with higher cognitive demands. First, you explain, then you move on to guided practice, followed by independent practice, and (much) later they can move on to inquiry or discovery learning approaches. In other words, direct instruction a la Barak Rosenshine:

References

Ausubel, D. P. (1968). Educational Psychology: A Cognitive View. New York, NY: Holt, Rinehart and Winston.

Bjork, E. L., & Bjork, R. A. (2011). Making things hard on yourself, but in a good way: Creating desirable difficulties to enhance learning. In M. A. Gernsbacher, R. W. Pew, L. M. Hough, J. R. Pomerantz (Eds.), Psychology and the real world: Essays illustrating fundamental contributions to society, (pp 56-64). New York, NY: Worth Publishers.

Kirschner, P. A., Sweller, J., & Clark, R. E. (2006). Why minimal guidance during instruction does not work: An analysis of the failure of constructivist, discovery, problem-based, experiential, and inquiry-based teaching. Educational Psychologist, 41(2), 75-86.

Krapohl, E., Rimfeld, K., Shakeshaft, N.G., Trzaskowski, M., McMillan, A. & Plomin, R. (2014). The high heritability of educational achievement reflects many genetically influenced traits, not just intelligence. Proceedings of the National Academy of Sciences, 111(42), 15273-15278.

Shakeshaft, N.G., Trzaskowski, M., McMillan, A., Rimfeld, K., Krapohl, E., Haworth, C.M., & Plomin, R. (2013). Strong genetic influence on a UK nationwide test of educational achievement at the end of compulsory education at age 16. PloS one, 8(12), e80341.

Simonsmeier, B. A., Flaig, M., Deiglmayr, A., Schalk, L., & Schneider, M. (2018). Domain-Specific Prior Knowledge and Learning: A Meta-Analysis. Research Synthesis 2018, Trier, Germany. Retrieved from https://www.researchgate.net/publication/323358056_Domain-Specific_Prior_Knowledge_and_Learning_A_Meta-Analysis

Stein, M. K., & Smith, M. S. (1998). Mathematical tasks as a framework for reflection: From research to practice. Mathematics Teaching in the Middle School, 3(4), 268-275.