The Soul Made Learner
Devika Toprani on the Somographic Learning framework — the ten-minute process that puts your thinking before the tool, so what AI gives you back is yours.
Crayon and Formula
Devika Toprani · The Origin
At age five, Devika Toprani could not learn mathematics from a textbook.
Not because she wasn't capable — she was exceptional at languages and social science. But formulas without grounding, without drawing, without some way to make the thing physical — they didn't take. Her parents were frustrated. The school system had one answer: more of the same, faster.
So she did what she could. She took crayons. She drew her formulas out. She tried to make sense of them by giving them a shape, a color, a path she could follow. Nobody told her to do this. She invented it on her own, as a five-year-old, because it was the only way she could learn.
That was the beginning of the Somographic Learning framework — though she wouldn't have called it that for another twenty years.
Completing her master's degree in 2025 while working as an educator, she watched an entire cohort of graduate students arrive at the same place she'd been at five: unable to process 200-page textbooks, distracted by phones, and running straight to AI for polished answers they hadn't earned the ability to evaluate.
"Learning is not AI made. It is soul made."
Devika TopraniThe crayon story is the reason she believes that. And the She Leads AI community believed it too.
Cognitive Debt and the Missing Layer
Devika Toprani · The Research Case
There is a term in recent research — cognitive debt — and it describes exactly what happens when learners skip their own thinking and go straight to AI. The Kasmaie et al. paper named it: repeated AI use without prior thinking reduces neural engagement. Outputs accumulate. Understanding doesn't.
This is not a problem with AI. It is a problem with the order of operations.
Three independent research traditions reached the same conclusion — computational intelligence, educational neuroscience, and visual embodied cognition. Each says the same thing in its own language: the learner comes first. AI comes second. Not because AI is less capable, but because you cannot evaluate what you don't already understand.
The clarity gap. Most AI workflows skip directly to what Devika calls the "refine" stage — the polished output stage. Attempt and Map — the human thinking stages that should precede AI use — are absent. They were never designed in. People get answers to questions they haven't fully formed. They accept outputs they aren't equipped to interrogate.
Somographic Learning is the missing layer. Ten minutes. No special tools. Before any AI session.
Current AI-first workflows carry two equity problems Devika named directly: they are text-first and English-dominant. Multilingual learners, neurodivergent thinkers, and low-literacy learners are excluded by design. Somographic Learning is language-agnostic and shape-based. It draws from Universal Design for Learning principles and reaches learners that standard AI tools leave out.
Attempt, Map, Refine
Devika Toprani · Somographic Learning
The three stages are sequential. Each one depends on the last. You cannot shortcut to Refine — that's the state most people start in, which is exactly the problem.
Start messy, on purpose.
Before opening any AI tool, write down at least three things you already know, believe, or have observed about the topic. They don't need to be correct. The goal is to activate your existing understanding so you can see what's missing.
Attempt is a safe entry point. The messy draft is not a failure — it's the evidence that you thought before you asked.
Connect what you know.
Take two items from your Attempt list and draw an arrow between them. Complete this sentence: "[Item A] connects to [Item B] because ___." Write the because. Repeat for at least one more pair.
This is where understanding gets built — not just information. The brain slows down and makes explicit what it knows implicitly. Optional: share your Map with one other person. Different maps of the same topic surface assumptions you didn't know you had.
Then prompt.
Look at your Map. What connection are you trying to make that you can't quite complete? That gap is your prompt. You now know what you know and what you don't. Ask AI specifically about the gap — not a general question, a specific one.
The output you get back will be more precise. More importantly, you'll be equipped to evaluate it.
Three Questions Before You Send
Devika Toprani · Somatic AI Literacy
Before closing her presentation, Devika proposed a new competency — one that doesn't yet exist in AI literacy frameworks.
She calls it Somatic AI Literacy. Soma is the body. The idea: your body thinks before you prompt. Not metaphorically — your brain needs to have built a physical understanding of the topic before the AI session starts. The Attempt-Map-Refine process is how you do that. These three questions check that it happened.
"I know what I do not know. This is what this tool does."
Devika Toprani · on Map Before MachineDid you think before you wrote this prompt — or did you come here blank?
Is your question gap-based, not generation-based — are you asking about a specific gap you identified, or just generating to see what comes back?
Did you modify the output — or accept it as-is?
The third question is the one that catches people. Using AI output verbatim is the final stage of cognitive debt. The output only becomes yours when you've shaped it against your own understanding.
Prompts
Copy, paste, adjust for your context.
Community Voices
"My brain is breaking open. I volunteered with grade-school kids to help them learn to read. If they don't learn by second or third grade, they are left behind — and I was doing this way before AI was part of our lexicon. Now I'm just understanding the ramifications of that."
"I was in my 40s before I realized the degree to which I am a hyper visual learner. I dropped out of a master's program because I have no chance of reading a book and then just taking tests. I have met people well into their 60s who have spent a lifetime failing at educational things because they never understood their own learning style."
"Nothing about it is binary. The danger point is thinking it's this way or that way — have AI in the classroom or don't. There is a spectrum of learning techniques and styles. My mother was differentiating education and teaching before it was a buzzword."
"As someone who is dyslexic, it is important for many of us to visualize and think things out before we absorb information. We have to make strong connections in many ways to make it make sense."
From the Chat
Devika Toprani is a systems architect for human-centered learning and responsible AI adoption. She founded the Somographic Learning framework — Attempt, Map, Refine — to support structured sense-making and reduce cognitive overload before AI use. Her research spans learning design, workforce strategy, and analytics across Oman, India, the UAE, and the United States. She is open to institutional pilots, educational consulting, and strategic partnerships. Connect on LinkedIn · Soulful Learning with AI
Anne Murphy is the founder of She Leads AI, an AI education community for women leaders and entrepreneurs. These companion guides are produced by Anne with her agentic leadership team. 33 years in higher education fundraising. She found Devika through the She Leads AI community — and Devika had been building this framework since she drew her math formulas in crayon.
Social Saturday is supported by the She Leads AI Society. Society members make this free for everyone.