Deep versus Surface Learning

Right as I am working my way through various readings and teaching preparation for my new course I am occasionally creating artefacts that might be useful.

Unfortunately WP does not let me embed. So follow the link for the full resource: https://drntasler.github.io/Deep_versus_surface_Learning_2026/

Note:
These categories are not an equivalent of the actual learning that happens, they focus on how the learners think about or approach their engagement with learning material. They are also cultural and context dependent and often studies rely on self-reporting.

Critiques of the model

Collated by perplexity research assistant

Three recent critiques

1. Kovač, Nome, Jensen & Skreland (2023/2025), “The why, what and how of deep learning: critical analysis and additional concerns”, Education Inquiry 16(2), 237–253. Published online March 2023. Education Inquirytandfonline

• They call “deep learning” a “goulash”: a mixed concept whose ingredients change with context and preference.
• Policy enthusiasm for it far outweighs the evidence behind it.
• The deep/surface split is too categorical, and surface learning is unfairly cast as always negative.
• They suggest flexible combinations of approaches, including memorisation and testing.
• They also warn that a one-sided push for “deep” learning risks quietly excluding some learners.
• This is the best fit for your “overrated” argument.

2. Thompson (2024), “A comparison of two learning approach inventories and their utility in predicting examination performance and study habits”, Advances in Physiology Education. Advances in Physiology Educationjournals.physiology

• He compared two common questionnaires (the R-SPQ-2F and the short ASSIST) in an anatomy course.
• Once GPA was controlled for, neither predicted exam performance, including on higher-order questions.
• Deep scores were in fact negatively correlated with higher-order performance, the opposite of what the model predicts.
• Neither questionnaire consistently predicted study habits either.
• He notes that approaches may shift with subject, time and assessment, and that students may report what they think they should do.
• This directly supports your point that the link to actual learning is tentative.

3. McDaniel, Wally, Frey & Bates (2025), “Individual Differences in Student Learning: A Comparison Between the Student Approaches to Learning and Concept-Building Frameworks”, Behavioral Sciences 15(8), 1055. Behavioral Sciencesmdpi

• This is a cognitive-science study. Self-reported deep and surface scores on two questionnaires showed no link to how students actually learned on an objective concept-learning task.
• Bayesian analyses supported the “no link” result.
• The authors’ explanations include inaccurate self-knowledge about one’s own learning, misread items, and the gap between intending to understand and achieving it.
• It fits your note that these categories describe how learners think about their engagement, not the learning itself.

General Critique

• Howie & Bagnall (2013), “A critique of the deep and surface approaches to learning model”, Teaching in Higher Education 18(4), 389–400. They point to weak supporting evidence, imprecise concepts, ambiguous language and circular reasoning. This one is closest to your argument.[tandfonline]
• Haggis (2003), “Constructing images of ourselves?”, British Educational Research Journal 29(1), 89–104. Author questions the assumed links between conceptions of learning, approaches and outcomes.[eric.ed]
• Dinsmore & Alexander (2012), Educational Psychology Review 24(4), 499–567. A review of 221 studies finding inconsistent and ambiguous links between processing level and performance. Definitions and measures depended on each study’s theoretical frame, and the relationship was moderated by context.[digitalcommons.unf]
• Lindblom-Ylänne, Parpala & Postareff (2019), “What constitutes the surface approach to learning…?”, Studies in Higher Education. Students with near-identical high surface scores used surface processes quite differently, including a “deep approach with memorisation” profile. In other words, the questionnaires miss real variation, which supports your self-report point.[eric.ed]
• Asikainen & Gijbels (2017), Educational Psychology Review 29, 205–234. Across 43 longitudinal studies there was no clear evidence that students systematically shift towards deep approaches (commentary summary).[link.springer]

Culture and the strategic learner

• Kember (1996), “The intention to both memorise and understand”, Higher Education 31, 341–354. Evidence of a combined approach that helps explain the “paradox” of high-achieving Asian students who are labelled rote learners.[link.springer]
• Richardson, Abraham & Bond (2012), Psychological Bulletin 138(2), 353–387. A meta-analysis of grade predictors in which approaches to learning were one of five domains, while self-efficacy, grade goal and effort regulation stood out. I still need to check the exact deep, surface and strategic correlations in the full text before you quote them.[discovery.ucl.ac]

Cognitive science

• Morris, Bransford & Franks (1977), Journal of Verbal Learning and Verbal Behavior 16, 519–533. “Shallow” rhyme-based encoding beat semantic encoding when the test matched it. Depth is relative to the task, not good in itself.
• Karpicke & Blunt (2011), Science 331, 772–775. Retrieval practice beat elaborative concept mapping, even on inference questions. So a “surface-looking” strategy can produce meaningful learning. A counterweight: a recent meta-analysis found only a small retrieval advantage (g = 0.14), depending on feedback.[pubmed.ncbi.nlm.nih][link.springer]
• Hattie & Donoghue (2016), npj Science of Learning 1, 16013. They treat “surface” as a necessary phase of learning, not a deficit, with strategies working best at particular phases.[nature]
• Deslauriers et al. (2019), PNAS 116(39), 19251–19257. In active classes students learned more but felt they learned less. This is direct evidence against relying on self-report.[pubmed.ncbi.nlm.nih]

Neuroscience

• van Kesteren, Ruiter, Fernández & Henson (2012), “How schema and novelty augment memory formation”, Trends in Neurosciences 35(4), 211–219. Information that fits existing knowledge, and sometimes novel information, is better remembered, through interaction between memory and prefrontal brain regions. This suits your “drivers” framing: prior knowledge shapes how well a strategy works, whatever the “approach” label.[pubmed.ncbi.nlm.nih]