Domain-Specific Knowledge: 1, Domain-Independent Skills: 0

Mirjam Neelen & Paul A. Kirschner

In both the workplace and education, there’s a lot of talk about so-called ‘domain-independent skills’, also called ‘generic’ or ‘transferable’ skills. In general, the perceived need for those skills is based on the premise that we currently live in a knowledge-based new economy, and the associated pressures for lifelong learning as well as the maintenance of employability that come with that require something different than ‘simple knowledge’.  More specifically, in the context of the workplace, the idea is that organisations change so fast and are so complex that it’s no longer feasible to know what kind of domain-specific knowledge and skills people need (they’ll be outdated as soon as you’ve learned them, is the idea) and therefore, it’s better to focus on more generic skills so that people are more flexible and can more easily adapt to change. BruceElkin.com put it forward as follows:

Generic skills are high-order, transferable skills that are common to almost all complex endeavours. They include skills such as communicating, problem-solving, curiosity, patience, flexibility, purpose, persistence, resilience, courage, and creating — that apply across all specific fields. They enable us to organize, adapt, and strategically apply our specific skills in new situations and circumstances.”

We’ll get back to this later.

One example from an educational context, is the Armenian ministry of education claiming that each child should learn chess because it helps them to learn how to think creatively and strategically. Consequently, they’ll be better problem-solvers and become more intelligent. That’s the idea. We’ve discussed this example and explained why that idea is rubbish in a previous blog.

In another blog, we’ve explained why 21^st century skills don’t exist and we now take the opportunity to (strongly!) argue that domain-independent skills don’t exist either (and, in fairness, many of the so-called 21^st century skills overlap with generic skills). We’ll repeat it until we’re heard: Generic, domain-independent skills DON’T exist, period!

They.don’t.exist.

We’ll begin by briefly looking at what domain-independent skills are supposed to be. Then we’ll explore how we build generic knowledge first and domain-specific knowledge next. Finally, we’ll discuss the consequences of domain-independent skills on our problem-solving and other capabilities, which will leave us with only one conclusion: If you want to be ready for the future, forget about domain-independent skills and focus on domain-specific knowledge!

What are domain-independent skills?

A skill (Merriam Webster) is the ability to use one’s knowledge effectively and readily in execution or performance. Interestingly, it’s not so easy to find a strong definition on domain-independent skills. For example, Murdoch-Eaton and Whittle (2001) describe them as a common set of transferable skills, regardless of a domain, which facilitate people’s ability to learn. They also state that terms like attributes and lifelong learning have been used interchangeably with the concept of ‘generic skills’ with implications beyond disciplinary knowledge.

There’s no consensus on which domain-independent skills are most important, but common examples are communication, collaboration, and problem-solving. In this blog, the focus is on the last one but what goes for problem-solving, goes for other generic skills as well (for example, try communicating or collaborating effectively and knowledgeably about a subject you know nothing about). Let’s explore how we acquire and use knowledge and then reconsider the concept of ‘generic skills’ as it will become clear that, without domain-specific knowledge, you’re nothing but a drifter.

No knowledge, no nothing

Before we dive into domain-specific knowledge, let’s define it first, following Tricot and Sweller (2014) who state that it’s “memorised information that can lead to action permitting specified task completion over indefinite periods of time” (p.  3).

Now, let’s start at the beginning, when we start learning, as children. At that point, we first build biologically primary knowledge through our ‘senses’. We watch and listen, we play, we engage in social relations, learn our mother tongue, learn how to recognise faces, learn to communicate using sounds and gestures, and learn how to use a problem-solving strategy such as a means-end analysis. This type of knowledge is acquired easily and unconsciously. It just ‘happens’. As humans, we’ve evolved this type of knowledge over many generations. In contrast, biologically secondary knowledge is knowledge that is heavily culturally dependent. Examples are reading, writing, and all other kinds of topics that are usually taught in schools, such as history and geography. Although the acquisition of biologically secondary knowledge is highly dependent on its primary partner, it doesn’t happen through simply interacting with the environment. This type of knowledge needs to be taught. It’s acquired consciously and it requires mental effort. What both knowledge types have in common, is that they’re stored in our long-term memory (LTM).

Sweller (2004) explains that the one unique aspect of human cognitive architecture is the size of our LTM. It’s only this quantitative aspect of our cognitive architecture that distinguishes us humans from other species when it comes to the extent to what we can learn and how we learn. The centrality of LTM to learning is non-negotiable. After all, if nothing has changed in LTM, nothing has been learned (Kirschner, Sweller, & Clark, 2006). Period.

One thing that’s very characteristic for our LTM, is that the alterations it makes are slow. In other words, accumulating knowledge through learning doesn’t happen overnight, it’s a slow process. One of the reasons why the learning process is so slow, is because of the limitations of our working memory (WM). As the image below shows, the input for WM can come from both sensory memory (SM) and LTM.

What’s critical to understand, is that the capabilities of WM, when dealing with new information coming from SM is fundamentally different from its capability when dealing with knowledge that’s already stored in LTM. This difference makes a lot of sense! When WM is dealing with new (unknown to the individual) information from SM, there are no knowledge structures available to indicate how the new information should be organised. It’s kind of like looking for a needle in a haystack (is there something in the LTM that this new piece of knowledge relates to somehow?) or perhaps you can compare it with a trial and error process. In contrast, when there’s relevant prior knowledge available in LTM, it functions like a central executive (Sweller, 2004). According to Baddeley and Hitch (1974) the central executive acts as a supervisory system, controlling the flow of information from and to its slave systems: the phonological loop and the visuo-spatial sketchpad.

As Sweller states: “The acquired knowledge organises our sensory world and determines our actions. What we think of and even how we think is overwhelmingly determined by our knowledge held in LTM” (p.  17). When you have access to prior knowledge, you can simply ‘skip’ the limitations of WM. You can learn and solve problems more easily and effectively as a result.

Let’s reflect on what this means when we talk about generic skills. Let’s assume you’re dealing with a problem in a domain you know nothing about. In that case, you’d need to use your domain-independent, generic skills, right? If they’re ‘generic’ as claimed, you could use them to solve any problem in any area, for example by thinking of similar problems with known solutions (Tricot & Sweller, 2014). This generic knowledge would be stored in LTM, just like any other knowledge. If this is how it works, in theory people with no knowledge in a specific domain should be able to ‘beat’ the ones who DO have domain-specific knowledge. We’re lucky because there’s research on that exact topic, namely the research on what distinguishes experts (a lot of domain-specific knowledge) from novices (very little domain-specific knowledge).

Experts	Novices
Possess schemas for encoding elements into a single entity	No access to relevant schemas
Skills acquisition without needing to recall the rule	Attempt to remember & process individual elements
Automation important for complex problem-solving transfer	Need to apply cognitive capacity to inefficient problem-solving
Strong problem solving approach[1]	Weak problem solving approach[2]

(Chi, Feltovich, & Glaser, 1979; Chi et al., 1982; De Groot, 1946, 1965; Kalyuga, Chandler & Sweller, 1998; Schneider & Shiffrin, 1997; Wilson & Cole, 1996).

Until A.D. De Groot (1946, 1965) published his work on chess grand masters, it was not so obvious that domain-specific knowledge and LTM’s role were as central as they are when it comes to problem-solving. De Groot’s research clearly shows that the only reason why chess grand masters outperform novice or average chess players is because “they recognise most of the board configurations that they encountered during a game and knew from previous experience which were the best moves for each configuration” (p.  11). In other words, their chess problem-solving skills are derived from a humongous amount of domain-specific (chess!) knowledge that chess masters hold in their LTM. It is that (and ONLY THAT!) knowledge that distinguishes them from newer and lower performing chess players. Many other studies have since confirmed that problem-solving ability is heavily dependent on domain-specific knowledge held in LTM, and not just for chess, also for other topics, such as designing software and symbolic circuit drawings.

In short, you won’t get far solving problems in an area you know nothing or very little about. In other words, no knowledge, no nothing! You’ll get nowhere.

Problem-solving as a generic skill: Imagine that

Some other examples …. Think solving a learning problem in a workplace context when you know nothing about how people learn (oh, wait, this happens every day! SCARY!). Think solving a quantum mechanical or constitutional law problem when you don’t know anything about quantum physics or constitutional law. Think solving a delicate confrontation in a war zone when you know nothing about peace-keeping, the culture, the history, the sensitivities, etc. And so the list goes on and on and… Why can’t you be effective? Because of how our memory works. This is what you’ll most likely do: You’ll try to ‘figure it out’.

For example, you might spend a lot of time looking for information that you can’t judge well because you don’t fully understand what’s relevant and what isn’t. Or perhaps you’ll conduct a means-end analysis (known as a weak problem-solving approach; basically, reverting to something you did when you were still in your ‘biologically primary learning phase’):

When you have a lack of domain-specific knowledge, what it comes down to is that you must combine elements randomly and then test them for effectiveness (Sweller, 2004). This is of course very inefficient and in some instances, like in the peace-keeping example, even terribly dangerous.

To sum it up: It’s impossible to solve problems without domain-specific knowledge, it’s impossible to communicate effectively on a topic if you don’t know what you’re talking about, it’s impossible to collaborate effectively if you have no idea what you’re trying to accomplish with the others and what needs to happen to achieve the goal, and it’s even impossible to manage a project if you have no idea what the project entails (try managing a restaurant without any knowledge of food, hospitality, cooking…).

Tricot and Sweller (2014) write that psychological studies have researched cognitive performance for over 130 years now. They go on to say that “paradoxically, much of that research emphasised generic or domain-generic cognitive skills despite domain-specific knowledge held in long-term memory being arguably the most important factor, and possibly the only factor, determining acquired cognitive performance” (p.  3).

So just remember this: Domain-specific knowledge will bring you places. Domain-independent skills will turn you into a desert-wanderer, at best stumbling upon an oasis and worst getting nothing but lost.

 

References

Baddeley, A. D., & Hitch, G. (1974). Working memory. In  Psychology of learning and motivation  (Vol. 8, pp. 47-89). Academic press. Retrieved from https://app.nova.edu/toolbox/instructionalproducts/edd8124/fall11/1974-Baddeley-and-Hitch.pdf

Baddeley, A. D., & Hitch, G. I. (1986). Working memory. New York:  Oxford University Press.

Groot, A. D. de (1946). Het denken van den schaker. Amsterdam, The Netherlands: Noord-Hollandsche Uitgevers Maatschappij.

Groot, A. D. de (1965). Thought and choice in chess. The Hague, The Netherlands: Mouton

Murdoch‐Eaton, D., & Whittle, S. (2012). Generic skills in medical education: developing the tools for successful lifelong learning.  Medical education, 46(1), 120-128. Retrieved from https://www.researchgate.net/publication/51860566_Generic_skills_in_medical_education_Developing_the_tools_for_successful_lifelong_learning

Sweller, J. (2004). Instructional design consequences of an analogy between evolution by natural selection and human cognitive architecture.  Instructional science,  32(1-2), 9-31. Retrieved from http://www.ucs.mun.ca/~bmann/0_ARTICLES/CogLoad_Sweller04.pdf

Tricot, A., & Sweller, J. (2014). Domain-specific knowledge and why teaching generic skills does not work.  Educational psychology review,  26(2), 265-283. Retrieved from https://www.researchgate.net/publication/258162628_Domain-Specific_Knowledge_and_Why_Teaching_Generic_Skills_Does_Not_Work

[1] A strong problem-solving method as a top-down, domain-specific, approach (expertise) which can be applied directly to the problem

[2] A weak problem-solving method is a general, non-specific approach where there is little domain expertise to apply directly to a problem. Here given a current state and a goal state, the learner searches for an action reducing the difference between the two. The action is performed on the current state to produce a new state, and the process is recursively applied to this new state and the goal state until the goal state has been reached.