last tended
One of my favorite books of all time is Stanislas Dehaene’s “How We Learn: The New Science of Education and the Brain”. I’ve long had a curiosity for how we learn as early education has long been an interest area of mine. I'll also admit this curiosity deepened when I saw countless posts blaming the friction of web3 on-boarding on the major learning curve required to understand the technology and participate. I wondered why it was so hard to design these experiences in a way that incorporates what we know about our minds to help users more easily learn what was happening and what the technology was doing. "How We Learn: The New Science of Education and the Brain" looked to be the perfect starting point to understand the foundations of how we learn and begin to explore how we might more effectively introduce users to emerging technologies. This summary was primarily to help myself consolidate the learnings, but also to share this great work with others (-:
Dehaene is concise and clear as he reverse engineers recent advances in artificial intelligence to explain how many developments are closely tied to the study of the human mind. He provides a strong foundation of detail to explain what is physically and chemically happening in the brain, and then translates his learnings into practical frameworks that can be applied by the reader. While I highly encourage you to read the entire book, I wanted to do a brief summary on what he has identified as the four pillars of learning. For all intents and purposes, this framework has been my starting point when trying to understand more about how we learn and how I might design for it.
The four pillars of learning are:
We are constantly bombarded with endless information and stimuli. Stepping outside this morning on 2nd Ave in New York City on my way to the office I am crossing the bike lane and a motorized scooter comes within 12 inches of me-- in this moment it is impossible for me to notice the yelp of the woman that drops her coffee half a block up on the other side of the street. Attention is the selection of what information we will consciously attend to, and for me in that moment it is the scooter hurdling towards me. There is a host of information that our sensory circuits are consistently taking in, but the selection of what is consciously acknowledged and processed is the exclusive entry point for what can eventually be consolidated and understood.
Understanding the importance of attention, Dehaene cites Michael Posner and what he has identified as at least 3 of the major attention systems.
Potentially the oldest in evolution, alerting is the first warning or indication that we should be on watch. Often this is initiated by some sort of emotional reaction based on something that stood out to our senses. At a chemical level, neuromodulators are released across the brain (think serotonin and dopamine) and these alerting modulators travel across the entire cortext letting your brain know it should increase wakefulness and vigilance. This activation of the brain has been tied to neuroplasticity and the opening of sensitive periods. This means that during moments of heightened attention our brains are a bit more readily able to reorganize and store, or remember, the associated information. This evidence of neuroplasticity is what ties attention to rapid learning-- during these sensitive periods our minds are able to more quickly remember and learn new information. These rapid learning results have also been observed with video games that mobilize the brains alerting circuits by playing with life and death or activating the rewards system.
Attention is vital to initiate information processing and learning because you cannot learn something if you don’t even know it exists— but with how many stimuli are constantly occurring around us we physically cannot attend to them all. We have to orient and select which information to focus on.
When we select what we focus on two very important things happen— it is a spotlight effect where whatever lies within the spotlight is amplified, and whatever lies outside the spotlight is quieted and minimized. If we take a look at what is going on inside of our brains, the neurons, or parts of the brain, that focus on what’s in the spotlight become even more sensitive and amplify their firing to reach more of the brain. The rest of the neurons, or parts of our brain that are focused on things outside of the spotlight, are quieted so that the ones amplified have more power to influence the rest of the brain. The effect of this is similar to what is happening during alerting, just compounding and adding even more intensity— our brains become even more sensitive and more able to reorganize to encode, or remember, the information associated with what we have selected.
It is important to note what happens to the information outside of the spotlight— our brains are at risk of essentially becoming blind to it as our minds direct efforts and ignore it. In The Principles of Psychology, William James described the function of attention as “Millions of items of the outward order are present to my senses which never properly enter into my experience. Why? Because they have no interest for me. My experience is what I agree to attend to. Only those items which I notice shape my mind.”
But what makes something of interest for me? how do we actually choose what we attend to and know it’s the right thing to attend to? How do we maintain a level of attention to facilitate learning? These were some of the questions I have immediately when reading of the selection of stimuli, and initially I felt Dehaene left a pretty large gap here. I did find the section on active engagement adds a touch more color about what curiosity is, how we pique it, and what sources of motivation might be, but what has less information available is what exactly drives my curiosity for a specific topic. I found this related article, The Psychology and Neuroscience of Curiosity by Celeste Kidd and Benjamin Y Hayden, that summarizes recent research and provides a comprehensive overview of the variety of ways curiosity has been defined.
Something comes to our attention, we’ve selected what parts to focus on, and we’ve got all cylinders firing… now we just need to determine how this information should be processed. This brings our frontal cortex into play, the largest part of our brain that enables our highest level of thinking, our executive control. This is where we finally become conscious of the attended to information and our executive control figures out what mental operator is the right way to process it. Also known as the switchboard of the brain, it orients, directs, and governs our mental processes. The executive control also has strong ties with working memory because to be able to do the operations that enable you to understand the information, you must temporarily store all of the pieces of information necessary to complete the operations. Dehaene identifies this as the global neural workspace, where the operations take place and information is temporarily stored. Our executive control selects what inputs and outputs move in and out, and is able to evaluate the effectiveness of the mental programs happening to correct course if the current process isn’t working. Because so much is happening at this level, we’ve come to find that it’s relatively slow and can only process one piece of information at a time. We think we can multitask but by definition we can only be consciously aware of and processing one thing at a time. Studies have found that there is always a delay in direct proportion to the time it takes to make the first decision. True multi-tasking may only become possible once an activity has reached automation where it is stored in the brain and does not require conscious processing. For some this might be possible for certain activities such as playing an instrument. We will find this can be possible thanks to rigorous practice and the final pillar of learning, consolidation.
In the book Dehaene then goes into the relationship between early education and the development of attention and executive control in children. Many early child development theories can actually be attributed to the structural development of the prefrontal cortex, executive control and learning to attend. While I won’t include that here I greatly recommend reading the book for more. (-:
From the earliest ages infants are acutely aware of faces, and in particular eyes. When someone is speaking to them their first action is to get eye contact with the person speaking, then to follow the individuals gaze to whatever they might be looking at. This is an early display of social attention sharing, and a major foundation for one our greatest advantages as humans, social learning. Social learning is what allows us to learn from each others' experiences and has enabled a significant amount of our societal development for the past several thousand years. Itis also a risk when not balanced with active, independent critical thinking where we test our own hypothesis and don’t just accept all the information we’re given at face value.
Once we have attended to and become consciously aware of information, we are able to actively engage. Active engagement is the previously mentioned process of critical thinking where we engage with and explore our environments. To learn, our brains must form hypothetical mental models of the world around us, come up with predictions on what might happen, project the model onto the environment, and then compare our predictions to the results. The feedback received can then be incorporated into our mental models and tested further, supporting the cycle of learning. What is vital to this being effective is a consistency in interest, motivation, and goals. Learning is only possible if the brain is attentive, focused, and actively testing and incorporating new information into its mental models. This is why actively engaged students often paraphrase concepts in their own words. This relatively simple form of active engagement helps to center the students’ attention, reflect on the information they’re processing, and project their current understanding of the framework to those around them. From there they can receive feedback on how accurate their understanding is and update their mental model accordingly. (Like me at this very moment summarizing Dehaene to form my own mental model of how we learn and then testing how it can expand to inform my design practice (-; )
It’s important here to acknowledge the roles of mental effort and reflection in this process. In contrast to brute memorization by phonics or spelling, research finds that memory performance is higher when in-depth processing is used to learn a list of terms. Dehaene referenced s study that had 3 groups review a list of 60 terms. The first group was asked to determine whether the terms were uppercase, the second whether the terms rhymed with ‘chair’, and the third whether the terms were animals. Memory performance is far greater in the third group thanks to what has been identified as the processing depth effect. When the level of mental effort is greater and active engagement is occurring (by processing the words and recalling information to determine whether they're animals), the processing is more likely to reach and activate deeper areas of the prefrontal cortex and hippocampus, making it more likely for a memory to be formed.
while things like reflection and physical interaction are strategies for promoting depth of thinking, other major factors that help to instigate and sustain active engagement are curiosity and motivation. The rough definition for curiosity is that it is active information-seeking. While a bit more abstract than our need for food, it is known that the seeking out, and acquisition of information activates the dopamine circuit and brings gratification just the same as other rewarded behaviors like eating. Similar to what we discussed previously on neuromodulators in the section on alerting, this hormonal release promotes memory formation further assisting in learning. So what exactly piques curiosity? From a stimulus-based perspective, it is seen as a novel stimuli that focuses our attention and then drives our need for information to help us understand that stimuli. For something like survival this would’ve been the curiosity to explore a territory to understand potential threats. What we know less about is the observed drive for non-stimulus specific information. This is the concept of epistemic curiosity, the pure desire for knowledge in any field or for abstract information.
There’s a fairly established feedback loop that supports why we seek out novel information and how we find the experience to be rewarding, but how can we more clearly define it? “Curiosity is the direct manifestation of children’s motivation to understand the world and build a model of it” It occurs whenever our brain detects a gap between what we already know and what we would like to know. Understanding what we would like to know is key. While we still don’t fully understand the algorithm— it is believed that curiosity guides us to what we think we can learn. This theory insinuates curiosity follows a bell curve. We have no interest in the unsurprising— but we are also deterred from things that are too novel or surprising. Things that may be so confusing that we do not believe we will ever be able to understand it. This range is constantly shifting as we master information and topics become less appealing so we direct our curiosity elsewhere, or we find that the topic is too difficult and our learning is progressing too slowly so we lose motivation to learn it. To sustain motivation and progress there is a delicate balance of setting realistic goals and having a strong feedback loop. If we are unable to achieve our goals or do not have clear feedback, motivation can subside and we redirect our focus away from something that we previously pursued.
Can it really be that simple? Is everything I have no motivation to learn just something I either determine is too simple or will be too difficult to grasp? What about the moments of inspiration when presented with one topic of new information over another? Or the topics that are more challenging yet still we pursue them with a level of motivation we wouldn't apply to something more in range of our capabilities? Is all else the same outside of novelty and level of difficulty? Dehaene references a study done where the gap theory of curiosity is implemented in a robot. When the robot is placed on a baby mat it behaves exactly as a young child would. To learn more about this interesting experiment and see this theory of curiosity in action reference Computational Theories of Curiosity-Driven Learning.
Overall, this definition of curiosity implies the existence of metacognition. To want to learn new information, we must be aware of what we do not already know. “Metacognition” is defined as the existence of higher-order cognition that oversees our mental processes. The system is constantly monitoring our learning, tracking what we know and do not know, evaluating speed, and whether we are wrong or not. Error feedback becomes vital to this system as it is the major input in evaluating whether we are wrong or not and how that is incorporated as the process loops.
“Organisms only learn what events violate their expectations” Rescorla-Wagner learning rule 1972
Our brain can only truly learn if it perceives a gap in what was predicted and what result was received— by being surprised by the outcome and then incorporating those results into their model to explore further. This perceived gap is not reserved to only being wrong. When we make predictions our minds are also associating a level of confidence with the probability of said prediction being correct. Once that gap is perceived, what becomes vital is the quality and accuracy of the feedback we receive to enable us to progress to reduce the gap and uncertainty. Thinking back to what we covered on active engagement, error feedback is vital to continued hypothesis testing as the results of our prediction are evaluated and incorporated in our models to then test again.
Error signals are the result of the minds surprised reaction to a result as it compares to the original prediction. This could be having an entirely incorrect prediction, or having such a low confidence interval that you didn’t expect to be right. The importance of these signals is apparent as their presence has been identified across all regions of the brain. As soon as something happens that we did not predict, the error signal is disseminated to pass the input on to the next level of processing with the hopes of another process and prediction making sense of the outcome. When the opposite happens and the result or sensory input matches expectations, the stimulus is quieted in the mind as to free up space for other information to be actively processed. This predictive mechanism of expected outcomes also applies to rewards-- if there is a signal that allows the mind to predict a reward, the neuronal firing occurs as a result of the signal and not the actual reward. If we expect a stimuli to result in food it is the stimuli that triggers the resulting dopamine circuit whereas the food itself causes no explicit reaction.
Knowing that the quality and accuracy of feedback is a direct determinant in the speed and success of learning— how can we outline what defines the quality and accuracy of feedback? For starters it’s important that clear goals are set and the learner is provided with the necessary information to understand where they went astray and for what reasons. It’s also not just the content that makes the feedback, but the delivery. While the impacts are definitely more prominent in children, delivering feedback as dispassionately as possible by beginning with showing someone what they should have done and where they may have gone astray helps to sustain motivation. When beginning with the message of “you’re wrong” it makes it harder for the child to connect the dots on why or what the correct answer is, leading to reduced motivation as it takes longer for them to figure out why or what to do next.
Dehaene’s main lens when introducing these topics is early education so many of the examples used in the book refer back to children and the education system. While I won’t cover all of the incredible information he shares on schooling specifics, I believe an important and universal example is grading. While defined as a form of feedback, the structure lends itself towards punishment and doesn’t have the flexibility needed for how we all have different learning speeds and needs. Early experiences with structures like grading have been tied to reduced motivation and the development of fixed mindsets. Still, retrieval practice is one of the most effective educational strategies and testing can be one of the most effective methods when employed properly. If we step away from how we think of testing today, what becomes clear is that the value is not having the emphasis placed on the resulting scoring, but the effort made to retrieve information and then the immediate feedback that directs where further development is necessary. This enables self-awareness by reflecting on what you do and do not know, allowing for focus on what can be identified as informational gaps. Taking this one step further is the concept of spacing out learning. This flexes retrieval muscles over time and refreshes novelty promoting deeper memory formation. The more you practice retrieval testing over longer intervals, the longer the information is stored.
After time and repetition of centered attention and active engagement, consolidation allows us to go from slow, high effort processing to unconscious and automatic expertise. When we learn, our prefrontal cortex works to consciously process each piece of new information as we saw in active engagement. Overtime as the skill is mastered our brains continue to relearn it through practice. During this process our minds compile the operations we use and transition them into more efficient routines. The processes are moved to other brain circuits, like the motor circuit or basal ganglia, and shifted away from conscious awareness to free up metal effort space for other learnings. Once we’ve reached this point we can see a shift to unconscious processing and the ability for some experts to complete these tasks automatically, while talking or thinking of other things. Familiar examples might be when musicians play an instrument, or when we drive a car, or other things we see as routine behaviors.
Recall retrieval practice from error feedback— it’s evident that spacing out learnings and testing over time produces better results and longer lasting memories. A significant factor in that construct is the role sleep plays in consolidation. For awhile we’ve known sleep improves learnings by increasing retention and improving performance of learned information. What was found to make these findings possible is that while we sleep our minds replay the activities from the day at a speed accelerated by a factor of 20. This was discovered when Wilson and Mcnaughton found that while rats sleep, without any external stimulation, the place cells in their hippocampus start firing in the same order they moved throughout the day. It was also found that this replay is not localized to the hippocampus, but also occurs in the cortex where much of our learning is occurring. This re-firing spreads the information to other neural networks and enables long-term memory transfer. This process applies to both the strengthening and atomization of existing knowledge, as well as the recording of information from the day in a more abstract and general form. This is what could be leading to eureka moments the following day after being able to “sleep on it”.
——
I love this book because I believe Dehaene provides a wonderful introduction to these major pillars with just enough detail to make me feel confident I understand why they are tied to the associated outcomes and what is making them possible. Acknowledging this value, I also want to highlight the fact that this is a pretty high-level, generalized overview. Each one of these pillars are massive topic areas with impacts across all areas of our lives. While it was extremely helpful to begin with a learning-specific lens and I intend to dive deeper into learning design and edtech, I also look forward to diving into each of the topic areas and looking at them from other angles.