University of Illinois Urbana-Champaign

Somagraphic Learning™ Framework: A Human-First, AI-Supported Visual Cognitive Approach

Independent Researcher/ Visual Learning Specialist, Devika Toprani

Permalink

https://hdl.handle.net/2142/133189

Description

Title
Somagraphic Learning™ Framework: A Human-First, AI-Supported Visual Cognitive Approach
Author(s)
Independent Researcher/ Visual Learning Specialist, Devika Toprani
Issue Date
2026-05-08
Keyword(s)
  • Somagraphic Learning
  • embodied cognition
  • human-AI interaction
  • visual orientation
  • cognitive load
  • automation bias
  • desirable difficulty
  • somatic AI literacy
  • metacognition
  • retrieval practice
  • cognitive sovereignty
  • drawing externalization
  • mediational agent
  • neurodiversity
  • multilingual learners
  • relational intelligence
  • AI sustainability
  • UDL
Date of Ingest
2026-05-08T17:39:52-05:00
Abstract
Artificial intelligence systems increasingly generate explanations, summaries, and analytical outputs at speeds that exceed the natural pace of human cognition. While these technologies expand informational access, they may compress the orientation processes through which conceptual understanding normally develops. Experimental research across seven preregistered studies demonstrates that learners who receive LLM-generated summaries develop shallower knowledge compared to those who engage in active construction through web search (Melumad & Yun, 2025). Separate empirical work further suggests that repeated AI writing assistance was associated with significantly reduced neural connectivity in an EEG study, a pattern the authors term cognitive debt (Kosmyna et al., 2025) - though this finding is preliminary and has not yet been peer-reviewed. Somagraphic Learning™ introduces a visual orientation layer that precedes language, explanation, or AI output. In this stage, learners externalize conceptual relationships using simple shapes, spatial arrangements, and motion cues before engaging with symbolic reasoning or AI-generated content. The learning process unfolds through a three-stage cycle: Attempt → Map → Refine. Grounded in embodied cognition (Lakoff & Johnson, 1999; Wilson, 2002), cognitive load theory (Sweller, 1988), human-AI interaction research (Amershi et al., 2019), and desirable difficulty principles (Bjork & Bjork, 2020), the framework positions visual cognition as a structured interface between human reasoning and AI-assisted learning. A central construct is the mitigation of automation bias - the tendency to defer to algorithmic outputs when internal conceptual models are absent (Skitka et al., 1999; Endsley, 2016). This paper presents the Somagraphic Learning™ Framework as a conceptual model and proposes a structured research agenda for empirical testing. It introduces Somatic AI Literacy™ as a proposed competency domain: the capacity to establish embodied conceptual orientation before AI interaction begins. It does not report experimental findings. Version 3 extends the framework's scope in three directions. First, sociocultural theory confirms that generative AI functions as a mediational agent that restructures participation in learning, not merely a tool that delivers information, which grounds the timing argument in a deeper theoretical account of why sequence matters (Tate et al., 2026). Second, converging evidence from workforce research, relational intelligence scholarship, and national education policy signals that the problem Somagraphic Learning™ addresses is not confined to individual classrooms. It operates at the level of professional competency, human flourishing, and governance of AI-integrated learning systems (Gartner, 2025; Hau, 2026; LinkedIn, 2026). Third, the framework's analog-first design carries an accessibility argument for neurodivergent learners, multilingual populations, and low-connectivity contexts that has not been previously articulated in human-AI sequencing frameworks. These extensions do not change the framework's core claim. They establish that the claim matters across a wider set of contexts than originally stated.
Publisher
Open Science Framework
Has Part
  • https://osf.io/preprints/edarxiv/fnk7z_v2
  • https://osf.io/preprints/edarxiv/fnk7z_v1
  • https://www.linkedin.com/in/devika-toprani/
  • https://substack.com/@devikatoprani
  • https://devikatoprani.carrd.co/
  • https://www.ideals.illinois.edu/items/140142
  • https://www.researchgate.net/scientific-contributions/Devika-Toprani-2344939009
Type of Resource
text
Genre of Resource
working paper
Language
eng
DOI
https://osf.io/preprints/edarxiv/fnk7z_v4
Copyright and License Information
CC BY-NC-ND 4.0

Owning Collections

Illinois Research and Scholarship (Open Collection) PRIMARY
This is the default collection for all research and scholarship developed by faculty, staff, or students at the University of Illinois at Urbana-Champaign