Using Learning Theories to Evolve Human-Centered XAI: Future Perspectives and Challenges
Pith reviewed 2026-05-13 23:03 UTC · model grok-4.3
The pith
Learning theories from human education can be integrated into XAI to create learner-centered explanations that increase human agency and simplify risk mitigation.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The central claim is that a learner-centered approach to Explainable AI, built by applying learning theories to the XAI lifecycle, can enhance human agency and ease the mitigation of associated risks, thereby evolving the practice of human-centered XAI.
What carries the argument
The learner-centered approach to XAI, which treats explanations as scaffolds for human learning processes rather than static information transfers.
If this is right
- Explanations would be assessed by how well they promote measurable learning outcomes rather than only immediate comprehension.
- XAI design processes would incorporate educational principles such as scaffolding and active knowledge construction.
- Risk mitigation would become easier because users with deeper understanding could better anticipate and handle AI failures.
- Human agency would rise as people move from passive recipients of AI outputs to active learners who can question and adapt those outputs.
Where Pith is reading between the lines
- XAI evaluation metrics might shift toward long-term retention and transfer of knowledge instead of snapshot accuracy scores.
- Current XAI methods could be shown to treat users as information consumers rather than learners, limiting their effectiveness.
- Adaptive XAI interfaces could emerge that adjust explanation style based on detected user learning progress, similar to tutoring systems.
Load-bearing premise
Learning theories created for human-to-human teaching apply directly and usefully to how people learn from AI explanations without major new adaptation or testing.
What would settle it
An empirical study comparing learner-centered XAI explanations against standard ones and finding no measurable gains in user agency, learning retention, or risk reduction.
read the original abstract
As Artificial Intelligence (AI) systems continue to grow in size and complexity, so does the difficulty of the quest for AI transparency. In a world of large models and complex AI systems, why do we explain AI and what should we explain? While explanations serve multiple functions, in the face of complexity humans have used and continue to use explanations to foster learning. In this position paper, we discuss how learning theories can be infused in the XAI lifecycle, as well as the key opportunities and challenges when adopting a learner-centered approach to assess, design and evaluate AI explanations. Building on past work, we argue that a learner-centered approach to Explainable AI (XAI) can enhance human agency and ease XAI risks mitigation, helping evolve the practice of human-centered XAI.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. This position paper argues for infusing learning theories from education into the XAI lifecycle to support a learner-centered approach for assessing, designing, and evaluating AI explanations. Building on prior work, it claims that this perspective can enhance human agency, mitigate risks in complex AI systems, and evolve human-centered XAI practices, while outlining key opportunities and challenges.
Significance. If the proposed synthesis holds, the paper offers a useful conceptual bridge between educational psychology and XAI, potentially guiding more effective explanation design that prioritizes human learning outcomes over mere transparency. Its value lies in the interdisciplinary framing and explicit discussion of future challenges rather than any empirical demonstration or formal derivation.
major comments (1)
- [Abstract and opportunities discussion] The central claim in the abstract and introduction rests on the transfer of human-to-human learning theories to human-AI explanation settings; while the manuscript frames this as a discussion of opportunities rather than a validated mapping, the absence of any concrete adaptation mechanisms or counter-examples in the opportunities section leaves the load-bearing assumption unexamined and risks overgeneralization.
minor comments (3)
- The manuscript would benefit from at least one worked example mapping a specific learning theory (e.g., constructivism or situated learning) to an existing XAI technique such as counterfactual explanations.
- [Related work] Several citations to foundational XAI surveys appear dated; updating the related-work section with post-2022 human-AI interaction studies would strengthen the positioning.
- [Challenges] The challenges section lists high-level issues but does not discuss measurement approaches for assessing whether explanations actually foster learning; adding a short paragraph on potential evaluation metrics would improve clarity.
Simulated Author's Rebuttal
We thank the referee for their thoughtful review and constructive suggestion. We agree that strengthening the opportunities discussion with more concrete elements will improve the manuscript and will make targeted revisions to address this point.
read point-by-point responses
-
Referee: [Abstract and opportunities discussion] The central claim in the abstract and introduction rests on the transfer of human-to-human learning theories to human-AI explanation settings; while the manuscript frames this as a discussion of opportunities rather than a validated mapping, the absence of any concrete adaptation mechanisms or counter-examples in the opportunities section leaves the load-bearing assumption unexamined and risks overgeneralization.
Authors: We appreciate this observation. As a position paper, the manuscript deliberately frames the discussion as exploratory opportunities rather than a validated transfer. However, we acknowledge that the opportunities section would benefit from greater specificity to reduce the risk of overgeneralization. In the revised version, we will add brief concrete adaptation mechanisms (e.g., how constructivist scaffolding principles could be operationalized in interactive XAI interfaces) and at least one counter-example where direct transfer from human-to-human learning may not apply (e.g., differences in cognitive load when explanations are generated by opaque models). These additions will be kept concise and clearly labeled as illustrative rather than exhaustive. revision: yes
Circularity Check
No significant circularity; position paper relies on conceptual synthesis
full rationale
This is a position paper whose central argument is a proposal to infuse learning theories into the XAI lifecycle to enhance human agency. It contains no equations, fitted parameters, derivations, or technical reductions. All claims are framed as discussion of opportunities and challenges drawn from prior literature rather than any self-referential chain that reduces a result to its own inputs by construction. Self-citations serve only as background and are not load-bearing for any derivation.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption Learning theories can be infused into the XAI lifecycle to foster human learning from AI explanations.
Reference graph
Works this paper leans on
-
[1]
Amina Adadi and Mohammed Berrada. 2018. Peeking inside the black-box: a survey on explainable artificial intelligence (XAI). IEEE access 6 (2018), 52138–52160
work page 2018
-
[2]
Muhammad Afzaal, Jalal Nouri, Aayesha Zia, Panagiotis Papapetrou, Uno Fors, Y ongchao Wu, Xiu Li, and Rebecka Weegar. 2021. Explainable AI for data-driven feedback and intelligent action recommendations to support students self-regulation. Frontiers in Artificial Intelligence 4 (2021), 723447
work page 2021
-
[3]
Alejandro Barredo Arrieta, Natalia Díaz-Rodríguez, Javier Del Ser, Adrien Bennetot, Siham Tabik, Alberto Barbado, Salvador García, Sergio Gil-López, Daniel Molina, Richard Benjamins, and others. 2020. Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI. Information fusion 58 (2020), 82–115
work page 2020
-
[4]
David Paul Ausubel, Joseph Donald Novak, Helen Hanesian, and others. 1978. Educational psychology: A cognitive view. (1978)
work page 1978
-
[5]
Steve Olusegun Bada and Steve Olusegun. 2015. Constructivism learning theory: A paradigm for teaching and learning. Journal of Research & Method in Education 5, 6 (2015), 66–70
work page 2015
-
[6]
Albert Bandura and Richard H Walters. 1977. Social learning theory. Vol. 1. Englewood cliffs Prentice Hall
work page 1977
-
[7]
Astrid Bertrand, Rafik Belloum, James R Eagan, and Winston Maxwell. 2022. How cognitive biases affect XAI-assisted decision-making: A systematic review. In Proceedings of the 2022 AAAI/ACM conference on AI, ethics, and society. 78–91
work page 2022
-
[8]
Benjamin S Bloom and David R Krathwohl. 2020. Taxonomy of educational objectives: The classification of educational goals. Book 1, Cognitive domain. longman
work page 2020
-
[9]
Zana Buçinca, Maja Barbara Malaya, and Krzysztof Z Gajos. 2021. To trust or to think: cognitive forcing functions can reduce overreliance on AI in AI-assisted decision-making. Proceedings of the ACM on Human-Computer Interaction 5, CSCW1 (2021), 1–21
work page 2021
-
[10]
Michelene TH Chi, Miriam Bassok, Matthew W Lewis, Peter Reimann, and Robert Glaser. 1989. Self-explanations: How students study and use examples in learning to solve problems. Cognitive science 13, 2 (1989), 145–182
work page 1989
-
[11]
K Patricia Cross. 1981. Adults as Learners. Increasing Participation and Facilitating Learning. (1981)
work page 1981
- [12]
-
[13]
Upol Ehsan, Q Vera Liao, Michael Muller, Mark O Riedl, and Justin D Weisz. 2021. Expanding explainability: Towards social transparency in ai systems. In Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems. 1–19
work page 2021
- [14]
-
[15]
Tad Hirsch, Kritzia Merced, Shrikanth Narayanan, Zac E Imel, and David C Atkins. 2017. Designing contestability: Interaction design, machine learning, and mental health. In Proceedings of the 2017 Conference on Designing Interactive Systems. 95–99
work page 2017
-
[16]
Kristina Höök, Jussi Karlgren, Annika Wærn, Nils Dahlbäck, Carl Gustaf Jansson, Klas Karlgren, and Benoit Lemaire. 1998. A glass box approach to adaptive hypermedia. Springer
work page 1998
- [17]
-
[18]
Frank C Keil, Robert Andrew Wilson, and Robert Anton Wilson. 2000. Explanation and cognition. MIT press
work page 2000
-
[19]
David A Kolb. 2014. Experiential learning: Experience as the source of learning and development. FT press
work page 2014
-
[20]
Jean Lave and Etienne Wenger. 1991. Situated learning: Legitimate peripheral participation. Cambridge university press
work page 1991
- [21]
-
[22]
H. Lin. 2022. Large-Scale Artificial Intelligence Models. Computer 55, 05 (may 2022), 76–80. DOI: http://dx.doi.org/10.1109/MC.2022.3151419
-
[23]
Emily G Liquin and Tania Lombrozo. 2022. Motivated to learn: An account of explanatory satisfaction. Cognitive Psychology 132 (2022), 101453
work page 2022
-
[24]
Tania Lombrozo. 2006. The structure and function of explanations. Trends in cognitive sciences 10, 10 (2006), 464–470
work page 2006
-
[25]
Tania Lombrozo. 2012. Explanation and abductive inference. (2012)
work page 2012
-
[26]
Tania Lombrozo. 2016. Explanatory preferences shape learning and inference. Trends in cognitive sciences 20, 10 (2016), 748–759
work page 2016
-
[27]
Jack Mezirow. 1997. Transformative learning: Theory to practice. New directions for adult and continuing education 1997, 74 (1997), 5–12
work page 1997
-
[28]
Jack Mezirow and others. 1990. Fostering critical reflection in adulthood. Jossey-Bass Publishers San Francisco
work page 1990
-
[29]
Tim Miller. 2019. Explanation in artificial intelligence: Insights from the social sciences. Artificial intelligence 267 (2019), 1–38
work page 2019
-
[30]
Jennifer A Moon. 2013a. A handbook of reflective and experiential learning: Theory and practice. Routledge
-
[31]
Jennifer A Moon. 2013b. Reflection in learning and professional development: Theory and practice. Routledge
-
[32]
Mohammad Naiseh, Deniz Cemiloglu, Dena Al Thani, Nan Jiang, and Raian Ali. 2021. Explainable recommendations and calibrated trust: two systematic user errors. Computer 54, 10 (2021), 28–37
work page 2021
-
[33]
Timothy J Nokes and Stellan Ohlsson. 2005. Comparing multiple paths to mastery: What is learned? Cognitive Science 29, 5 (2005), 769–796
work page 2005
-
[34]
Mahsan Nourani, Chiradeep Roy, Jeremy E Block, Donald R Honeycutt, Tahrima Rahman, Eric Ragan, and Vibhav Gogate. 2021. Anchoring bias affects mental model formation and user reliance in explainable ai systems. In 26th International Conference on Intelligent User Interfaces. 340–350
work page 2021
-
[35]
Jean Piaget, Margaret Cook, and others. 1952. The origins of intelligence in children. Vol. 8. International Universities Press New Y ork
work page 1952
-
[36]
Pooja Shikaripur Bheemasena Rao, Dinesh Babu Jayagopi, and Mauro Cherubini. 2022. Explainability in Automated Training and Feedback Systems. (2022)
work page 2022
-
[37]
Bethany Rittle-Johnson. 2006. Promoting transfer: Effects of self-explanation and direct instruction. Child development 77, 1 (2006), 1–15
work page 2006
-
[38]
Waddah Saeed and Christian Omlin. 2023. Explainable ai (xai): A systematic meta-survey of current challenges and future opportunities. Knowledge-Based Systems (2023), 110273
work page 2023
- [39]
- [40]
-
[41]
Dale H Schunk. 2012. Learning theories an educational perspective. Pearson Education, Inc
work page 2012
-
[42]
Burrhus Frederic Skinner. 1954. The science of learning and the art of teaching. Cambridge, Mass, USA 99 (1954), 113
work page 1954
-
[43]
Sarath Sreedharan, Tathagata Chakraborti, and Subbarao Kambhampati. 2021. Foundations of explanations as model reconciliation. Artificial Intelligence 301 (2021), 103558
work page 2021
-
[44]
I agree to the terms and conditions
Nili Steinfeld. 2016. “I agree to the terms and conditions”:(How) do users read privacy policies online? An eye-tracking experiment. Computers in human behavior 55 (2016), 992–1000
work page 2016
-
[45]
David CM Taylor and Hossam Hamdy. 2013. Adult learning theories: implications for learning and teaching in medical education: AMEE Guide No. 83. Medical teacher 35, 11 (2013), e1561–e1572
work page 2013
-
[46]
Anja Thieme, Ed Cutrell, Cecily Morrison, Alex Taylor, and Abigail Sellen. 2020. Interpretability as a dynamic of human-AI interaction. Interactions 27, 5 (2020), 40–45
work page 2020
-
[47]
Edward Thorndike. 2017. Animal intelligence: Experimental studies. Routledge
work page 2017
-
[48]
Pablo Villalobos, Jaime Sevilla, Tamay Besiroglu, Lennart Heim, Anson Ho, and Marius Hobbhahn
-
[49]
arXiv preprint arXiv:2207.02852 (2022)
Machine Learning Model Sizes and the Parameter Gap. arXiv preprint arXiv:2207.02852 (2022)
-
[50]
Human Motivation: Metaphors, Theories and Research
B Weiner. Human Motivation: Metaphors, Theories and Research. 1992. (????)
work page 1992
-
[51]
Joseph J Williams and Tania Lombrozo. 2010. The role of explanation in discovery and generalization: Evidence from category learning. Cognitive science 34, 5 (2010), 776–806
work page 2010
-
[52]
Joseph Jay Williams, Tania Lombrozo, and Bob Rehder. 2013. The hazards of explanation: Overgeneralization in the face of exceptions. Journal of Experimental Psychology: General 142, 4 (2013), 1006
work page 2013
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.