here are some questions that fans of NHK world can lean into so that Gemini UYBK ai model integrates japanese characteristics
what is GAME-GENOMEMOTHER 2 is a classic RPG known internationally as EarthBound. Its creator, Itoi Shigesato, joins longtime fans to discuss how the game was made, revealing secrets behind its lasting popularity.
GAME-GENOME is an acclaimed educational docuseries produced by NHK, Japan's public broadcaster, that approaches video games as a form of cultural art, literature, and philosophy. Rather than focusing on sales figures or basic reviews, the show dissects the "genetic makeup" (the genome) of legendary video games -more
===========================
how can humans employ the japanese employs "mitateru," -a Japanese technique, to inspire children's imagination
The Japanese concept of "mitateru" (見立てる)—which translates roughly to "viewing an object as something else"—is the traditional art of creative substitution and metaphorical thinking. It is the practice of seeing an ordinary item (like a rock or a piece of cloth) not for its literal form, but for what it could represent (an island or an ocean). more
when computers learning imagenet (2012) did anyone teach them mitateru too? more
o explore how they were built, what inspired their creators, and how they have profoundly shaped players and modern society. [1, 2, 3, 4, 5]
ReplyDeleteThe specific excerpt you shared is the official logline for a special episode of the series broadcast on NHK WORLD-JAPAN. [1, 2]
The MOTHER 2 (EarthBound) Episode Breakdown
The Legend: MOTHER 2 (released in the West as EarthBound) is celebrated globally for completely subverting classic RPG tropes. Instead of knights, swords, and medieval sorcery, it features children with baseball bats and psychic powers navigating a quirky, satirical, 20th-century American-inspired world. [1, 2, 3]
The Guest: The episode features a rare appearance by Shigesato Itoi, the legendary copywriter, essayist, and mastermind behind the MOTHER trilogy. [1, 2, 3]
The Concept: Alongside passionate fans, Itoi dives deep into the game's notoriously challenging five-year development cycle, the structural programming breakthroughs enabled by Nintendo's late Satoru Iwata, and the philosophical writing choices that have given the game its immortal, multi-generational cult status. [1, 2]
What Makes GAME-GENOME Unique?
Each episode features an analytical mix of high-production recap footage, deep-dive studio discussions, and behind-the-scenes creator testimonies. The series frequently features structural breakdowns of other landmark titles—such as Hideo Kojima analyzing DEATH STRANDING or deep dives into SEKIRO and Monster Hunter—to uncover the artistic intent behind the screen. [1, 2, 3, 4]
Would you like to know where to stream or watch this episode on the NHK network, or are you interested in specific behind-the-scenes development stories from Shigesato Itoi's work on EarthBound?
While it is famous for its role in traditional Japanese arts like bonsai, rock gardens (karesansui), and tea ceremonies, mitateru is a highly effective, natural tool for inspiring a child's imagination and nurturing abstract intelligence.Here is how humans can actively employ mitateru to unlock a child's creative flow:1. Curate "Open-Ended" Loose PartsInstead of providing highly detailed, single-purpose toys (like a plastic toy kitchen or a branded superhero action figure), give children raw, ambiguous materials.The Material: Provide wooden blocks, silk cloths, cardboard boxes, pinecones, pebbles, or bamboo tubes.The Mitateru Action: Allow the child to perform the substitution. A blue silk cloth becomes a rushing river; a single wooden block becomes a smartphone, a spaceship, or a piece of sushi. Because the object is blank, the child's mind must work harder to project reality onto it.2. Guide with Constructive, Abstract PromptsAvoid telling a child what an object is. Instead, use conversational prompts that invite them to look at the world through a metaphorical lens.Instead of saying: "Look at that cloud, it looks like a dog."Employ Mitateru by asking: "If that cloud were a vehicle, who would be driving it and where is it going?"Object transformation: Pick up a simple stick during a walk and ask, "If this wasn't a piece of wood, what tool or magical item would it be?" This teaches children to look beneath the surface of everyday environments.3. Blend Miniaturization with StorytellingA core element of mitateru in Japanese culture is scaling the vast world down into something small and manageable (like a miniature tray garden).The Micro-World: Encourage children to use a small tray filled with sand, moss, or rice to represent a massive landscape.The Leap of Imagination: A jagged piece of bark becomes a massive mountain cliff; a scattering of white rice becomes a snowstorm. This shifting of scale helps children practice spatial reasoning and mental modeling, allowing them to construct complex narrative universes in a confined, accessible space.4. Reframe Everyday Domestic SpacesHelp children apply mitateru to their immediate, physical surroundings to see the home as an interactive landscape.The Living Room Sea: A simple blue rug is transformed into a deep, dangerous ocean, and the couch cushions become isolated islands.The Threshold: Walking through a specific doorway can be framed as stepping through a portal into a completely different historical era or ecosystem, instantly changing the context of their play without changing their physical location.Why It Matters for a Child's DevelopmentEmploying mitateru moves children away from passive consumption and toward active generation. By learning that an object's meaning is not fixed by its manufacturer but is entirely determined by the observer, children develop strong cognitive flexibility, emotional empathy, and a lifelong capacity for creative problem-solving.Are you looking to apply this technique in a structured classroom setting, or are you designing at-home activities for a specific age group?
ReplyDeletees. When the AI community began dissecting convolutional neural networks trained on ImageNet, researchers unknowingly engineered the exact computational equivalent of mitateru.Because ImageNet was designed to teach computers literalism (e.g., “This pixel pattern is exactly 100% a Golden Retriever”), the true breakthrough came when engineers decided to flip the algorithms backward to see what happened when the machine was forced to use metaphorical projection and creative substitution.Computer scientists exploited this machine version of mitateru through several landmark experiments:1. Google DeepDream: Algorithmic Pareidolia (2015)In traditional mitateru, a child looks at a cloud and projects the image of a dragon onto it. In 2015, Google researchers did the exact same thing to an ImageNet-trained network by creating DeepDream. [1] (https://en.wikipedia.org/wiki/Pareidolia)The Process: Researchers fed the network an ordinary, ambiguous image (like a picture of a sky or a brick wall). Instead of asking the computer to classify it, they told a specific layer of the network: “Whatever object you think you vaguely see in those textures, amplify it. Alter the pixels to make it look MORE like that object.” [1] (https://en.wikipedia.org/wiki/Pareidolia), [2] (https://www.youtube.com/watch?v=f23vEzI3LQ0&t=170)The Result: The computer looked at a leaf and saw a bird; it looked at a cloud and forced it to become a pig-snail. By feeding the output back into the input as a loop, the computer essentially "hallucinated" a world where everything was a metaphor for something else—the exact mechanical manifestation of mitateru. [1] (https://www.youtube.com/watch?v=f23vEzI3LQ0&t=170), [2] (https://en.wikipedia.org/wiki/Pareidolia)2. Feature Visualization: Pulling "Archetypes" From the VoidIn Japanese stone gardening (karesansui), a single standing rock doesn't just represent a mountain; it is selected because its rugged lines capture the absolute essence of "mountain-ness."AI researchers did this mathematically through Feature Visualization. They took a computer trained on ImageNet, started with a canvas of pure, random digital static (noise), and instructed the network: "Change this noise so that it activates your concept of a 'banana' or a 'parachute' as intensely as possible."The computer didn't draw a realistic, literal banana.Instead, it painted beautiful, sweeping, abstract fractal patterns of yellow curves and segmented skin textures. It generated the platonic archetype of the object—capturing its spirit rather than its literal form, mirroring how mitateru extracts the essence of an object to substitute it for another.
ReplyDeletemore on mirateru -3. MIT's "Faces in Things" & MetaCLUE (Recent Advances)As AI evolved, researchers realized that standard computer vision was too rigid compared to human child psychology. To fix this, scientists began explicitly teaching models how to comprehend visual metaphors:The MIT Pareidolia Project: Researchers at MIT developed an equation and a dataset called "Faces in Things" to track how and where algorithms spot "illusory faces" in objects like coffee foam, power outlets, or car headlights. They found that when an AI is trained like a human—balancing generalized categorization with fine-grained recognition—it naturally develops human-like pareidolia. [1] (https://news.mit.edu/2024/ai-pareidolia-can-machines-spot-faces-in-inanimate-objects-0930), [2] (https://mghpcc.org/ai-pareidolia-can-machines-spot-faces-in-inanimate-objects/), [3] (https://pmc.ncbi.nlm.nih.gov/articles/PMC11790231/)The MetaCLUE Benchmark: Computer vision researchers introduced MetaCLUE (Visual Metaphors Research), a dataset specifically designed to move AI away from literal interpretations. It actively tests whether a neural network can look at a creative ad or an art piece and understand abstract, cross-domain concept substitutions—essentially trying to grade an AI on its capability to perform mitateru. [1] (https://arxiv.org/abs/2212.09898), [2] (https://metaclue.github.io/)The Human vs. Computer DivideThere is still a beautiful, fundamental difference. When a child plays mitateru, they know the stick is a stick, but they choose to treat it as a magic wand. When an ImageNet-trained computer looks at a textured wall and sees a dog's face, it is experiencing a glitch—it genuinely believes the dog is there because its mathematical parameters are over-firing. Humans use mitateru as a conscious bridge between reality and fantasy; early AI used it because it couldn't tell the difference. [1] (https://cyberneticforests.substack.com/p/ai-video-and-stochastic-pareidolia), [2] (https://www.youtube.com/watch?v=f23vEzI3LQ0&t=170)Would you like to explore how modern text-to-image AI models (like Midjourney) handle mitateru compared to those older ImageNet models, or are you interested in a specific creative coding project where you can force a vision model to interpret objects metaphorically?
ReplyDelete