00:00:07.359
I remember a time when I was on a bus, and a bunch of people got on. As they were inching toward their seats, I looked up and had a double take. I recognized Dr. Ramachandran, a renowned neuroscientist. He seemed a little confused to be recognized. He asked if I was a student at the university where he conducts research, and I replied, 'No, I'm a programmer.' I blurted out that I live with a synesthete. That encounter was brief, but it had a lasting impact on me. Ramachandran is known for his groundbreaking work on synesthesia, which involves neural cross-wiring.
00:00:40.200
For someone with synesthesia, the activation of one sensory receptor can trigger a secondary perception in another. For example, activating a pain receptor could also result in hearing a tone or experiencing a particular taste. One of my classmates at university organized her desk items in pairs because she found two to be a cheerful number, while she thought three was dreary and avoided it entirely. This secondary perception is remarkably stable over time. Interestingly, some individuals with synesthesia can leverage this unique perception for orientation; for instance, one woman recognized when it was time to visit the dentist when her toothache turned orange.
00:01:17.400
There's an Australian opera singer named Priscilla Dunston whose synesthesia provides her with an extraordinary memory for sounds. Around 25 years ago, she became a new mother, which is an incredibly challenging time. Sleep-deprivation combined with a crying baby can leave new parents feeling hopeless and isolated. While others advise that parents can eventually learn to differentiate between cries, Dunston struggled with this. In desperation, she began keeping a log of her baby's cries and sounds, leading her to realize that there were specific sounds at the beginning of a cry that indicated different needs.
00:02:06.280
Through careful listening, she distinguished between five distinct pre-cry sounds, which reflected different needs like hunger or gastrointestinal discomfort. This insight enabled her to venture out more, where she observed other babies expressing similar needs in remarkably similar ways. Thus, she was able to perceive what others could not. This brings us to the broader discussion of expertise and mastery. We often consider expertise in terms of knowing facts or being able to verbalize and execute concepts explicitly.
00:02:59.120
One character in 'The Kingkiller Chronicle' likens playing music to telling a joke; anyone can remember and repeat the words, but making someone laugh requires a deeper understanding. A faster delivery doesn’t create a better joke. Experts possess a kind of insight, judgment, intuition, or brilliance derived from a myriad of unarticulated experiences or subtle influences from time. In an experiment, researchers tasked terrain analysts with analyzing aerial photographs in just two minutes—a task that typically takes hours. After this limited time, one analyst casually mentioned that whoever went to the area needed to be prepared for certain bacterial infections. The researcher was baffled, questioning how he could discern such a detail from a photo taken from 40,000 feet.
00:04:30.160
The analyst explained his reasoning—based on signs of tropical climate and uniformly mature vegetation, he could deduce the underlying soil and its relationship to the geography of the area. Experts often struggle to articulate how they know what they know; their expertise operates in a consistent and mysterious manner. So, if it’s challenging to articulate and understand expertise, how do we teach it? How do we impart seasoning, judgment, and intuition? An example from Japan illustrates this; at the turn of the previous century, some individuals could accurately determine the sex of day-old chicks. This skill was crucial because without designated chick sexers, the poultry industry faced heavy losses.
00:05:23.720
Employing chick sexers eliminated weeks of unnecessary feeding, and Japan established a school in the 1920s dedicated to training new chick sexers. Interestingly, the trainees often couldn’t explain how they identified the chicks' sex. The training involved pairing novices with experts to guess the sex, receiving feedback based on their guesses. After two years, these newly graduated sexers could accurately identify up to 1,200 chicks per hour with a 97% accuracy rate, all without being able to articulate how they did it. Decades later, someone simplified the process to an 80% accuracy rule, but the original method worked phenomenally well.
00:06:12.840
Traditionally, education has emphasized instructing the deliberate brain. We create tutorials, design drills, and provide homework. However, we struggle with training the automatic brain, which processes chaotic signals and finds implicit solutions. The study of psychology has examined how the brain sorts through this noise, beginning in the 1960s with Eleanor Gibson's experiment. She demonstrated the vital building block of snap judgments through simple squiggles, asking subjects to identify matching figures. Without feedback, over time, subjects could recognize and differentiate various squiggle dimensions, enabling their brains to improve perceptual resolution.
00:07:01.560
In this field, known as perceptual learning, studies investigate the differences between how experts and novices perceive. For instance, novice drummers will read rhythms note by note, while experienced drummers read beat by beat, recognizing rhythmic figures as coherent ideas. This distinction leads experts to make faster and more insightful judgments concerning rhythmic patterns. Navigating unfamiliar public transit systems, we often analyze every sign and signal, hoping to make the right choice. As we become more familiar with the city, our brain starts filtering out irrelevant details, honing in on the cues that matter.
00:08:07.360
This selective attention enables experts to quickly process information. Experienced pilots can determine aircraft status by glancing at their instruments, while inexperienced pilots have to cross-check each one meticulously. This difference in processing efficiency between experts and novices illustrates how expertise evolves within our cognitive systems. The cognitive resources of experts remain unaffected, while novices often feel drained after performing similar tasks. The study of these differences highlights how efficiently information is extracted and the implications for training intuition.
00:08:59.800
Philip Kellman, a cognitive scientist, has explored ways to train intuition intentionally. His research focuses not only on simple skills but on real-world complex problems. One such project involved helping pilots enhance their visual navigation skills to avoid wrong landings and getting lost in cross-country flights.
00:09:04.480
Kellman's team showed pilots 20 seconds of cockpit video followed by a map with three locations. They took around 30 seconds to identify the correct spot, succeeding only 50% of the time. While they developed this skill through experience, it wasn’t always well done. His experiment with perceptual training improved their accuracy to about 80% in just three hours, while non-pilots also managed to achieve a 60% accuracy rate, showcasing how perception can be improved through training.
00:09:54.880
Kellman also tackled another challenge: teaching fractions to middle school students. Many students struggle with the conceptual understanding of fractions and often approach problems arbitrarily. Rather than training them to solve problem types directly, he designed interactive trials to help them recognize different representations of fraction problems. Surprisingly, despite no direct teaching of fraction solutions, students' scores improved remarkably from 40% to 70% after training. This improvement persisted over time, indicating the effectiveness of this perceptual training method.
00:10:40.560
As our brains sift through noise, we begin to recognize valuable signals, and Dr. Kellman's work shows that we can deliberately refine this learning process. The question I ponder most often is how these principles apply to programming. What do skilled programmers perceive that others do not? A few years back, I wrote a blog post inspired by a friend’s tweet, showcasing code and receiving feedback. My friend pointed out a race condition that I had overlooked, illustrating how some individuals can identify potential issues instantly, but it goes deeper than that.
00:11:13.120
When reviewing code, the types of problems noticed vary significantly with expertise. Less experienced individuals often focus on low-level, standalone issues, while more seasoned developers tend to identify systemic problems. During my time at GitHub, I worked on GitHub's REST API, which had complex authorization logic. I noticed that our integration tests were bogged down by excessive database calls, leading to failures in continuous integration due to time constraints.
00:12:02.640
After submitting a patch to reduce database calls by 40%, I realized that uncovering performance issues often relies on keen observational skills. Remarkably, while I thought of myself as an average developer, those who excel in troubleshooting possess an innate ability to identify issues instinctively without needing to articulate their reasoning. The important thing is that this intuitive ability is often developed over time as one engages with complexities.
00:12:45.480
During debugging sessions, it’s not merely seeing the problem that leads to resolution, but rather connecting the dots based on prior knowledge. For instance, when grappling with a stubborn bug, identifying the underlying issue may depend on recalling similar past experiences—when informed, one colleague recognized that our error stemmed from defining a foreign key incorrectly, leading to a moment of clarity and resolution.
00:13:34.000
Conversations with seasoned experts can trigger insights. Ten years ago, I discussed a challenging exercise with Sandy Metz, who highlighted how she identified elegant solutions to a complex problem that had caused me frustration. Her ability to see through the noise and propose viable solutions reinforced my understanding of the complexity involved in programming and the insights gained through experience.
00:14:04.800
Despite efforts to articulate knowledge in programming, a significant part of expertise stems from perception—the gut feeling that allows for rapid judgments. In pursuit of improving this perceptual learning, I have compiled a guide outlining strategies for effective training, based on Kellman’s research. This guide emphasizes the significance of engaging the learner with brief classification exercises that integrate complex data sets, enabling the brain to differentiate meaningful patterns amidst noise.
00:14:56.320
For training to be impactful, data must cover a broad spectrum of variations, ensuring learners can discern relevant features while disregarding irrelevant characteristics. Teaching perceptual skills involves exposing novices to a wealth of examples, allowing them to recognize patterns and signal differences unambiguously. This approach promises to develop their cognitive skills significantly over time.
00:15:39.000
Applying these strategies to programming requires identifying the skills we want to foster. This task becomes challenging when facing complex, chaotic environments, especially during production incidents. Research in naturalistic decision-making aims to extract tacit knowledge from experts engaged in real tasks, revealing how they navigate complicated situations. While I won't delve into the specifics, I encourage exploring related literature, such as Gary Klein's 'Sources of Power: How People Make Decisions,' which provides a detailed understanding of decision-making processes.
00:16:20.000
Determining effective targets for perceptual training can take two forms. The first entails defining clear, unambiguous tasks, such as identifying whether a chick is male or female, or determining if a fraction problem is asking for a whole or a part. This clarity allows the brain to explore the relationship between relevant diagnostic features naturally. The second approach targets well-defined activities, as seen in the visual navigation tasks of pilots who rely on intricate data but must categorize it without explicit guidelines.
00:17:03.760
Once we delineate these objectives, our focus shifts to acquiring or curating a suitable dataset. Recent estimates suggest the number of programmers doubles approximately every five years, meaning a significant portion may lack exposure to effective practices. A considerable percentage of developers are likely to contend with an overload of irrelevant data, impeding their capability to spot patterns. However, intentional learning experiences can mitigate these challenges, allowing programmers to refine their skills more effectively.
00:18:12.680
Consequently, we can strive to create comprehensive training methods that compress the learning curve, facilitating new developers in spending less time wrestling with technical details and more time addressing meaningful challenges in programming. Thank you.