Keynote: Cultivating Instinct by Katrina Owen

by Katrina Owen

In her keynote speech titled "Cultivating Instinct," presented at the Keep Ruby Weird 2016 event, Katrina Owen explores the intricacies of expertise and intuition, drawing connections between cognitive science and real-world applications. Owen begins by recounting a serendipitous encounter with a neuroscientist, suggesting that insights can emerge from unexpected interactions. The main theme revolves around understanding and training our instinctual responses, particularly in fields requiring rapid decision-making and specialization.

Key Points Discussed:

- Synesthesia: Owen introduces synesthesia as a unique neurological condition that provides individuals with additional layers of perception, framing it as a metaphor for instinctual understanding.

- Expertise Development: The transition from novice to expert involves a shift from deliberate, conscious processing to automatic, intuitive responses, illustrated through examples from various fields.

- Learning Dynamics: Owen elaborates on the distinct phases of learning involving perceptual learning theories, which encompass discovery effects (how we perceive information) and fluency effects (the efficiency of processing that information).

- Perceptual Learning: Citing researcher Eleanor Gibson, Owen describes an experiment demonstrating how participants can improve their perceptual resolution by identifying distinguishing features among seemingly random shapes.

- Training Intuition: Cognitive scientist Philip Kellman's work shows that intuition can be explicitly trained. Owen discusses how Kellman successfully raised the accuracy of pilots and middle school students through targeted perceptual training, challenging conventional expectations about learning times.

- The Role of Feedback: Owen highlights the importance of feedback in learning environments, emphasizing that individuals grow and refine their skills when they receive corrective guidance during practice.

- Practical Application: She illustrates these concepts through her experiences with programming and debugging, suggesting that understanding perceptual aspects can lead to more efficient problem-solving in technical tasks.

Conclusions and Takeaways:

- Mastery involves not just speed and surface knowledge but an ability to synthesize experiences into actionable insights.

- Training should focus on the nuances of perception and intuition, especially in complex fields like programming, where a deep understanding of patterns is essential.

- By recognizing and cultivating these instinctual skills, individuals can enhance their expertise and intuitive decision-making in various disciplines, fostering innovation and creativity in their work.

00:00:07.700 Years ago, I was sitting on a train. It was quiet and fairly empty. Then, at one stop, a bunch of people got on. As they were inching along, trying to get to their seats, I did a double-take and exclaimed, 'You're Doctor Villain or a ceramic Ramachandran?' The guy was like, 'Yes, yes, I am! So stoked!' This guy is a neuroscientist. He asks fascinating questions and he's a phenomenal storyteller. He seemed a little surprised to be recognized, and I guess this was before he was publishing New York Times bestsellers and things. I think he asked if I was a student at the university where he does research. I said, 'No, I'm a programmer.'

00:00:28.410 Then, I blurted out, 'I live with a synesthete.' And the encounter was over. This might seem like a non sequitur, but it’s somehow related. Ramachandran did groundbreaking work on synesthesia. Synesthesia is a little bit of cross-wiring that happens in the brain. A few people have it—it’s kind of rare—and it is genetically determined. So, you either have it or you don't. What happens is that when you get a sensory perception, it triggers a signal that bleeds over into a different part of the brain, which then triggers a secondary perception. You might activate a pain receptor and feel pain, and that pain might have a specific color, or you’ll hear a sound and it might have a visual texture or a particular taste.

00:01:09.840 Some people with synesthesia can use this extra layer of perception to help them orient themselves. This was the case for Priscilla Dunstan, an Australian opera singer. She's been diagnosed as a synesthete, and her synesthesia gives her an extraordinary memory for sounds. About fifteen years ago, Priscilla had a baby, but from what I’ve heard, the first few weeks and months can be incredibly tough. The baby cries, you're sleep-deprived, and you have no idea what the baby needs. You feel helpless and isolated. People promise parents that they will start recognizing their baby's cries and understand what the baby needs. However, despite her extraordinary memory for sound, Dunstan was not getting it. She couldn't hear what these cries meant and couldn't find any resources that helped clarify it.

00:02:04.290 So one desperate morning, she decided to figure it out for herself. She started making a log of all her baby's cries and slowly came to the realization that there were distinct sounds to different cries. She learned that when the baby was crying, it often was just one big upset. However, she discovered that before the baby really began wailing, there were some fussy pre-cry sounds that were extremely distinct. As she listened and experimented, she realized that "I'm hungry" sounded very different from "I have gastrointestinal discomfort." She eventually assigned meaning to five different pre-cries that her son made. This discovery was transformational; she was getting a lot more sleep.

00:02:56.310 She didn't think much of it beyond the sense of, 'I'm really good at this. My son and I can communicate really well.' Then she started getting out of the house more, and to her astonishment, she started hearing other babies that were hungry, tired, and needed to be burped. They seemed to be expressing this in ways that were remarkably similar to her son's cries. So, she was hearing things that others were not. When we talk about expertise and mastery, we typically focus on knowing facts and being able to explicitly verbalize and perform sequences of steps. In the beginning, we know things and do things very deliberately, slowly, and laboriously.

00:04:09.960 Then, over time, we tend to get faster, and it becomes more effortless. One part of expertise is this transfer from slow, deliberate parts of the brain to the automatic, effortless part of the brain. However, an expert isn’t just a faster novice. One character in the King Killer Chronicles observes that playing music is somewhat like telling a joke. Anyone can remember the words or repeat them, but making someone laugh requires something more than that. Telling a joke faster doesn’t make it funnier. Experts seem to have a magical quality we call insight, judgment, intuition, or brilliance. They arrive at this understanding through a range of articulated experiences.

00:05:07.920 Now, practice can be seasoning and the enigmatic passage of time. This mysterious quality of expertise is crucial in almost every field: architecture, archaeology, nursing, electronic circuit design, and power plant operations. It’s a decisive factor in almost every skill. Whether experts are judging livestock or analyzing terrain based on aerial photography, they are relying on a gut sense, an ability to make snap judgments, to just know.

00:06:03.990 In one experiment, researchers gave terrain analysts two minutes to look at an aerial photograph—something that generally takes a few hours to analyze. After ten minutes, one engineer began his debrief with an offhand comment that whoever was sent to the area needed to be ready for certain types of bacterial infections. The researchers were amazed. 'What? You can see bacteria on a picture taken from 40,000 feet?' The engineer explained that the photo showed a tropical climate and the contour of the tree canopy, indicating mature vegetation. This reflected the underlying soil, which appeared to be tilted, interpreted as limestone.

00:06:30.750 Analyzing the topography indicated specific patterns to the streams and ponds. He noted that there seemed to be this pond that didn't have a major distributary running away from it, implying the presence of stagnant water, which could harbor bacteria. Experts often struggle to explain what they do or how they know things—they just do. Their skills are consistent, reliable, and reproducible, yet also somewhat mysterious.

00:07:10.260 So if you can't explain it and don't even know what it is, how do you teach it? How do you teach seasoning, perspicacity, and judgment? How do you teach intuition? Traditionally, we're pretty good at training the deliberate part of the brain. We write tutorials, create coursework, and devise drills and practice problems. However, we are not so good at training the automatic part of the brain, allowing it to figure things out amidst the chaotic signals that bombard us throughout life.

00:07:43.800 There exists a whole field of psychology dedicated to understanding how the brain figures all of this out, which started in the 1960s with researcher Eleanor Gibson. She designed a delightful experiment that debunked many long-held beliefs about how the brain learns. This experiment illustrated the fundamental building blocks of our ability to develop accurate snap judgments. Gibson began by showing research subjects a meaningless squiggle and asked them to identify every squiggle that matched this reference squiggle exactly. She would flip through them one by one, and the subjects would make their guesses, but she wouldn't give them any feedback. By the end of the third round, every subject had correctly identified every single target squiggle.

00:09:33.000 What was happening in the brain was that it discovered different meaningful squiggle dimensions. As they viewed the squiggles, participants began noticing characteristics like clockwise versus counterclockwise spins, the number of spirals, and the overall shapes of squiggles. Our brains constantly engage in a process of differentiation, figuring out which characteristics are important and which are not. As the brain identifies significant dimensions, it starts to focus on them, improving its perceptual resolution. Photographers gain richer experiences of light, musicians gain a richer appreciation of sound, and industrial tasters learn to evaluate complex flavors.

00:10:20.880 This field of study is called perceptual learning, which is typically broken into two broad categories: discovery effects and fluency effects. Discovery effects pertain to how we perceive and extract information. For example, novice drummers sight-read music note-for-note and focus on how long each note lasts. In contrast, experienced drummers don't read note-for-note; they read beat-for-beat, identifying distinct rhythmic figures. They start recognizing figures as single coherent ideas, while novices often focus on low-level, unrelated pieces of information. It’s the same in video games, where experienced players eventually learn to ignore loud, irrelevant data and focus on more nuanced, relevant details.

00:11:17.070 This principle is known as selectivity. Novices pay attention to both relevant and irrelevant data, whereas experts amplify relevant information and filter out the rest before it even reaches the conscious part of the brain. This means that experts may not even notice irrelevant data—they can't articulate its presence. The fluency effects focus not on how we extract information but on how efficiently we do so. For instance, experienced pilots can determine an aircraft's attitude and situation with a glance at their instrument panel, while inexperienced pilots will carefully read and cross-check instruments one at a time.

00:12:38.890 Novices process things serially, while experts process information in parallel. Expert photographers quickly compose shots and find focal points almost intuitively. They realize that if the composition isn't perfect, they can crop it later. This differentiation in speed demonstrates that novices often experience cognitive strain from these tasks, while experts handle them almost effortlessly. This concept is called attentional load, and it requires cognitive resources for novices to extract information; meanwhile, experts can perform the same tasks with negligible effort. This is the basic premise: discovery effects deal with filtering incoming signals, whereas fluency effects involve parallel, efficient processing.

00:13:51.440 Where this becomes interesting is when we take these ideas and figure out how to explicitly train intuition. Cognitive scientist Philip Kellman from UCLA has spent the last 25 years exploring this question. One fascinating aspect of his work is his focus on complex skills that address real-world problems. In the early 1990s, he was inspired by the fact that pilots often land at the wrong airport or get lost flying cross-country. You might think technology would have solved this by now, but in reality, the National Transportation Safety Board issued reminders in 2014, urging pilots to ensure they land at the correct airport.

00:14:55.160 So the skill at play is visual navigation, where pilots look out their cockpit window and visually assess the terrain while referencing a map to determine their location. Although pilots develop this skill through experience, they may not do so effectively. In Kellman's pretrial assessments, he tested pilots with varying amounts of experience—between 525 and 600 hours. He would show them videos of terrain and ask them to choose one of three locations on a map corresponding to that terrain. The pilots were correct only 50% of the time, which isn’t very good.

00:16:03.520 Kellman put the pilots through three hours of perceptual training consisting of brief interactive trials—20 seconds of video followed by a map with three marked locations, where they had to choose the right one. By the end of the experiment, the pilots’ accuracy improved to about 80%, moving up from the previous 50%. Not only did their accuracy increase, but so did their reaction times. From taking over 30 seconds to answer, they were choosing locations in under 15 seconds by the end of the experiment. Meanwhile, a second group of participants that consisted of non-pilots received the same three hours of perceptual training. Interestingly, these naive subjects outperformed the pilots before the training, achieving 65% accuracy and responding in less than 20 seconds.

00:17:06.990 Another interesting challenge Kellman tackled was teaching fractions to middle schoolers, which is notoriously difficult. Students don't often have a solid mental model for how they work; they often accept rules as arbitrary. Typical word problems might go something like this: 'Ten alley cats caught five-sevenths of the mice in a neighborhood. If they caught 70 mice, how many mice were there in total?' Although the numbers in similar problems may be the same, the underlying structures are fundamentally different. The first question asks them to find the whole, while the second asks them to find a part. These nuances can confuse students, leading to incorrect answers.

00:18:18.170 Kellman's training module was designed to help students discern between these structures. His approach involved interactive trials that represented a problem in one way and asked students to relate it to the same problem in another format. For example, they might solve a word problem, then choose from three different fraction strips depicting the problem. Rather than finding an answer, students had to recognize the scenario they were solving. The students participated in pre- and post-trial assessments, showing they initially solved problems correctly only around 40% of the time. After the intervention, their accuracy jumped to 62-70%. They were also tested several months later, and those scores maintained.

00:19:37.580 Kellman's work demonstrates that not only can intuition be deliberately trained, but the process of learning that typically takes months or years in real life can be compressed into a very brief timeframe. To apply these principles, I have crafted an amateur's guide to designing perceptual learning training materials. The core components involve brief classification episodes: these shouldn’t have to be instantaneous. If it’s possible to puzzle something out logically and deliberately, your brain should be given the time to do so. For instance, there are people who can reliably determine the gender of a chick shortly after it hatches—they can assess up to 1,200 chicks in an hour with 97% accuracy, yet they can't always explain how they do it.

00:20:47.420 This poses a challenge for those who can't identify gender until the chicks are about six weeks old, resulting in expenses when raising male chicks that will never lay eggs—an issue at an industrial scale. To train novice chick sexers, the method pairs an expert with a novice. The novice guesses the bird's gender, and the expert provides feedback until the novice achieves around a 90% accuracy rate, developing skills that they may feel are guessing but which they genuinely understand.

00:22:15.680 In all these cases, feedback plays a crucial role. The pilots learned the correct location on the map after making their choice, while in the students' cases, they received immediate feedback and corrective guidance. Consequently, their wrong answers became new questions, allowing them to draw connections to alternative representations of the problem. This marks the fundamental structure of short, interactive trials.

00:23:07.789 To implement training effectively, a comprehensive data set is necessary, emphasizing a wide variety of examples without duplicates. The brain requires diverse, complex variations to detect underlying invariants and diagnostic structures. This dataset should not only encompass relevant features, but also systematically vary noise, distractors, and unrelated characteristics. If these aspects are neglected, the brain may inadvertently associate incidental features with the critical structures, which is essentially a recipe for biases. To strengthen perceptual resolution, begin by presenting examples that differ significantly from one another before gradually decreasing contrast and incorporating nuances that highlight subtler differences.

00:24:55.620 To illustrate this concept practically, let's revisit Priscilla Dunstan, the synesthetic opera singer. She believes all babies, regardless of culture or genetics, produce the same pre-cries, linking to infant reflexes and the sounds associated with them. Recently, I explained this concept to a colleague, emphasizing the different sounds babies make based on their needs. For example, a hungry baby will trigger a swallowing reflex, which sounds quite distinct.

00:26:25.680 Conversely, an uncomfortable baby may make different sounds. However, Dunstan's ideas lack scientific validation; she has never attempted to disprove her hypothesis. While she initially worked with a university to study her claims, she ultimately canceled to share her findings with parents sooner. The essence of her discovery would require creating a dataset of audio and video recordings of numerous babies from a variety of backgrounds and environments, ensuring that individual recordings contained no duplicates for effective training.

00:27:02.300 Hence, a broad array of sounds recorded under diverse conditions—varying ambient noises, lighting, and angles—could be constructed into interactive trials wherein each audio or video clip allows participants to make decisions based on several choices. Engaging in this practice could greatly enhance understanding and application of perceptual learning principles, shedding light on how they apply not only to infant communication but to programming, and other complex domains as well.

00:28:09.800 A few months ago, one person I mentor asked for help debugging JavaScript in a browser game he had been attempting to build. While following a step-by-step tutorial, he encountered issues where the game either functioned partially or not at all. After he’d spent two full days contemplating the problem, I suggested we revisit each step together to work through it. Show me your HTML, I had said, only to spot a missing quotation mark—a small detail he had overlooked despite his careful examination.

00:28:38.370 This was a simple fix, but it exemplified how novices often get stuck on basic syntax errors for extensive periods. By implementing deliberate perceptual training, we could potentially reduce the painful weeks or months spent on such trivial errors. However, the more complex instincts in programming—like the ability to identify patterns in legacy code or suggest intricate abstractions—present a unique challenge. These refined skills require sharp perceptual abilities, influencing code review processes. More experienced programmers identify larger patterns that require deeper comprehension, while novices often focus on smaller, standalone issues.

00:29:07.500 Thus, creating an optimal dataset to accelerate and enhance learning could prove beneficial. Currently, the key diagnostic patterns of perceptual acuity in programming aren’t fully understood. Instead of simply urging you to create or curate such data, I encourage you to be mindful of perceptual facets in your interactions with code and collaborators, whether they’re more or less experienced.

00:29:46.690 Notice when there's a perceptual element at play and ask yourself what that component might be.