Direct instruction blocks pathways to great discoveries

Jump to navigation Jump to search

This text is part of: "I would never send my kids to school" by Piotr Wozniak (2017)

Evidence for the benefits of direct instruction

Education is supposed to produce great people. It is not just to make us pass tests.

I argue here that optimizing for good test performance may actually be harmful. We want to maximize creativity, innovation, and pathways to big discoveries. Optimizing for test scores is largely pushing in the opposite direction. We stand no chance in a competition with robots, but we can still aim higher as humans.

The mythology of schooling and direct instruction is so pervasive that it garners fervent advocates who cannot seem to break the mold of their own cognitive bias.

David Didau is an advocate for schooling, and direct instruction. In his article "Education isn’t natural – that’s why it’s hard", he espouses a myth in which the brain is supposed to be weak at abstract thinking for evolutionary reasons.

The article starts from a worn out critique of free learning:

One of the most troubling conundrums in the field of education is that the common sense observation that children learn so many things simply by virtue of being immersed in an appropriate environment is contradicted by the overwhelming empirical data that explicit instruction outperforms discovery approaches in schools

The allegedly "overwhelming evidence" comes from PISA test correlations and the like. There is no mention of obvious contrary evidence from neuroscience. The "overwhelming evidence" can be summarized as follows: "if you do a lot of heavy cramming, you will do great at standardized tests, and therefore, schools are good" (see: Discovery learning is very effective). The reality is that truly overwhelming evidence shows that schools inflict a great deal of harm in the area of the psychology of learning, a great deal of harm to the learn drive and the creative process, a great deal of harm to self-reliance, and a great deal of harm to the mental health of students. Brain science explains why.

Problem with discovery learning

There is a similar straightforward explanation for the seeming inefficiency of discovery learning. Again, if you wasted a lot of time doing discovery in physics, you would sure get a good backlog of materials in history. Tests do not care, and provide no indication that your efforts in physics might lead to a Nobel level breakthrough. Needless to say, packing kids in a class and instructing them to "please discover!" is as bad as direct instruction. Discovery must be derived from the learn drive, not coercion in conditions of limited freedom. Enslavement is actually more injurious to the creative process than to a simple effort to cram facts.

Classrooms hurt discovery learning more than they hurt the cognitive response to direct instruction. The culprit is the classrooms, not the discovery

For more see: It is hard to research discovery learning, Misleading research in sociology and psychology, and Horrible theory of minimal guidance learning by Kirschner, Clark, and Sweller.

Evolutionary weakness of human cognition

Didau seems to be happy to have discovered a recent book by D.C. Geary and D.B. Berch, in which an artificial distinction is made between "folk knowledge" and "academic knowledge" (aka "biologically secondary knowledge"). Didau says:

Schools exist to teach the hard stuff that children are unlikely to just pick up from their environments. As John Sweller acknowledges, “Since Geary’s formulation, it has become clear that theories like cognitive load theory apply solely to the biologically secondary knowledge for which schools and other educational institutions were invented”

Geary and Berch speak of the hard-to-bridge knowledge gap allegedly caused by the tardiness of evolution. Didau picks it up to justify teaching. In reality, the gap is a sole expression of knowledge complexity, and can easily be reduced by optimizing semantic pathways in modeling knowledge.

The only difference between folk knowledge and academic knowledge is its complexity, applicability, and the semantic proximity to sensory origins

Hierarchy of declarative knowledge in semantic networks

All declarative knowledge stored in the human brain forms a semantic network that can be ordered into a sequence by the timing of coherent acquisition. There are senses and perceptions at the bottom indeed. These are the "biological' or "natural" inputs. All things above in the hierarchy of knowledge simply build on the basics. This is why astrophysics builds on physics which builds on math which builds on finger counting (see: Mountain climb metaphor of schooling).

Whichever metaphor we choose to understand the acquisition of knowledge, we can easily show that the whole evolutionary argument is fake. The same sparse neocortical representation is used for (1) basic knowledge grounded in perceptions, and (2) most advanced abstract knowledge:

If knowledge is like a tree, adding new branches does not make learning harder. Just the opposite. There are branches that boost cognitive power, and meta-learning skills.

The jigsaw puzzle of knowledge does not get harder with the increase in size because unlike the puzzles in the toy box, knowledge puzzles have the constant total number of available pieces (the reality to learn). It is the assembled puzzle part in the brain that keeps growing. With longer edges and with parallel processing, it is easier to make each successive fit. This mechanism powers an exuberant learn drive.

Knowledge crystallization also does not show any discontinuity while depositing additional layers of the crystal

The gap of knowledge in discovery that was hard to bridge for Einstein becomes easier and easier with each generation due to better models, better metaphors, their ubiquity, and the early exposure. Today, a 6-year-old can find catchy explanation of complex models in physics or mathematics on YouTube. She can do it before she can even read. Those transformative changes in knowledge are the prime reason for the Flynn Effect (see the lecture by Jim Flynn himself). The topologically defined semantic distance between two pieces of knowledge is an expression of the structure of the semantic network. Semantic networks are mutable and semantic relatedness is subject to change accordingly. In simple terms, the knowledge gap keeps shrinking, and we need no evolution for that to occur. It is a pure transformation of models, and a cultural shift. The brain is well equipped to crack problems we cannot even conceive today. There is no boundary between simple knowledge and abstract knowledge. All knowledge is stored in the same concept network that keeps self-organizing via conceptual computation. We need no evolution to advance into unimagibable.

Today I struggle to explain the harms of schooling. The kids of tomorrow will see it all as obvious. This will not be a result of the evolution of the brain. This will be the outcome of the evolution of human knowledge

Exploratory learning is driven by necessities

Didau says that "This [knowledge dichotomy] explains why we find it easy to learn to speak, but much more difficult to learn to read and write ".

This ignores the fact that kids need to speak to get food, while reading can be equally natural as long as the environment is soaked with reading messages whereas lack of reading skills implies lack of benefit. We observe it in democratic schools where kids learn reading when they need reading. Not earlier, and not later. Learning to read entails an effort in little correlation with intellectual predispositions, age, or complexity of texts involved.

It is true that we did not see print in the course of evolution, however, the Wernicke’s area can consume input from speech, sign language, or text with comparable ease. In case of reading, the brain only needs to allocate some area of the neocortex to do the necessary preprocessing.

Speech is associated with (1) interactive feedback, (2) direct reward, and (3) dominates auditory input information. Those characteristics are absent when learning to read. This is why natural conceptualization does not occur when a child is handed a book without pictures. However, with the arrival of videogames, child's environment is transformed. Texts become associated with feedback, and rewards of gaming. The participation of natural text input also increases. An increasing proportion of children master or improve their reading skills with the participation of videogames.

Decoding shapes of letters and blobs of words comes naturally to healthy neural networks unless they are interfered with signals such as toxic memories associated with reading. In other words, it is schooling that is a prime culprit in what we see as reading difficulties. Naturally, ignorance of brain science makes people believe that if little schooling leads to bad reading, then more schooling might improve the outcomes. No wonder then that it is often very hard to find underlying neurobiological causes of dyslexia.

Kids learn speech well ahead of reading because of the differences in the interactive availability of comprehensible input in the environment. The structure of the brain or the evolution play a negligible role in that equation. The simplest way to prove that claim would be to engineer text-based virtual reality for children who are deprived of auditory input.

The learning difference between speech and reading is in the environment, not in the brain

Exploratory learning and great discoveries

At the peak of cognition is a new discovery, a breakthrough that moves the civilization forward by a little step. This is exactly when natural thinking and learning processes comes to fore. If schools are necessary to move from the evolutionary bottom of folk knowledge, how do we ever come up with new discoveries. D.C. Geary and D.B. Berch propose a new fallacy that will obscure a clear picture. For the neocortex, classifying foods is based on the same mechanism by which Einstein encoded the model that lead to E=mc2. The same cortex, the same toolset, the same encoding, and only the distance from the first step in childhood to the new location in the concept network was a bit longer for Einstein.

Common sense anthropology by David Lancy

Didau picked a piece from a book he possibly did not even read in its entirety, or which he chose to intentionally ignore in part. Instead of celebrating Geary and Berch, Didau would find some balance by reading passages from Peter Gray or David F. Lancy in the same book. Lancy noticed:

Children who must learn in and from the environment (as opposed to learning from teachers and books) develop characteristically different attention patterns (Gaskins & Paradise, 2010; Rogoff, Correa-Chávez, & Cotuc, 2005). Village children, as well as immigrant children whose mothers have little schooling - invited to learn to make something (e.g. Origami figures) - rely on observing the task as it is carried out by an expert or attempted by other children. A sample of more “schooled” individuals, on the other hand, pay little attention to the demonstration, waiting for (or soliciting) a teacher’s explanation and verbal guidance (Correa-Chavez & Rogoff, 2005)

Lancy also added a good point against the seemingly evolutionary-based proclivity towards teaching and receiving instruction - an inclination that I must have been born without:

If teaching is vital and universal, we should find the majority of adults considered “good” teachers and children “good” pupils. Assuming, for the sake of argument, that everyone is born with a suite of cognitive traits and the explicit motivation and determination “to facilitate learning in others” (Kline, 2015), we might expect to see the majority of the adult population acting eagerly and willingly as teachers. On average, they should be “good at it.” By the same token, children should gravitate readily to the role of pupil and automatically display appropriate behaviors in order to benefit from the lessons. Again, the majority should exhibit considerable native ability to learn from an instructor. On the subject of “natural teachers,” cases that illustrate careful, informed, systematic Vygotskian-style scaffolded instruction are virtually nonexistent before the modern era

Teachers are welcome when invited

Didau makes it short and emphatic: "This is why we need schools". Words like that spoken by a teacher are a heavy offence to human intelligence. Kids do not need teachers to do great and will do better without direct instruction. All "stuff is hard" until it becomes easy. It is just a matter of need and prior knowledge that forms best scaffolding. See: Do we need teachers?.

Children know all too well what I say. They know that I am right. They know that stubborn teachers who insist on teaching against a child's will are wrong. The problem is that kids have no freedom to decide. I have no issue with teaching that was requested voluntarily. I did have a few questions to my teachers too in the two long decades of my education. Kids rarely ask for teachers unless their self-reliance is crippled by schooling. It is always the pressure from busy parents or, worst of all, systemic coercion that makes them victims of coercive teaching (see: Ban on homeschooling).

Building abstract knowledge via modelling will always be a combination of natural abstraction (e.g. via forgetting), directed attentional effort in abstraction, and transmission from human sources of knowledge. A teacher may be at times more efficient than a book, but it is increasing hard for teachers to compete with simple media like YouTube, or more advanced formed on inquiry-based learning like PhET simulations. The key difference between direct instruction and free learning is the optimization of choices that is based on the power of the learn drive.

Harms of direct instructions

Due to arbitrary choices in knowledge acquisition, teaching can lead to:

To understand the value of generalization, compare a child and a well-schooled adult watching the same instructional video. I stop watching when I get overwhelmed with detail. Healthy kids go on, and pick up whatever their brains opt for (unless they get bored). A well-schooled adult will try to cram and capture it all on "brain tape", even if it leads to a war of the networks. This is an inevitable outcome of conditioning by direct instruction: the slow transition from the exploratory learning to the cramming mode.

It does not take a bad teacher to achieve the above injuries. Generalization power withers in absence of generalization. The mere presence of a teacher takes away generalization skills. As for toxic memories, I would want to believe that a good teacher should be able to spot the symptoms and react, however, many teachers do not even realize the phenomenon exists.

Schools diminish the generalization toolset. Instead, kids are often gifted with a rich set of toxic memories

A toddler with a smartphone

As for the cognitive gap caused by tardy evolution, I suggest Didau and other proponents of direct instruction look at toddlers who master smartphones faster and better than their parents. Herein comes the ultimate proof. There is no folk physics involved in decoding the meaning of buttons such as install, loading or play! Here the shortcomings of evolution did not seem to emerge. The key ingredients in child's exploration are passion and the learn drive. When kids are driven to explore, and show the rage to master, there is no limit to what they can achieve cognitively. The motivation aspects are just part of prior knowledge. Not everyone can read astrophysics at four. It comes in incrementally. See: Mountain climb metaphor of schooling