Prepared for: Vishal Sachdev, BuildToLearn position paper (v3, June 2026) Key question: What is the WEAKEST claim about taste that is empirically defensible and still does useful work in a learning framework?
From the position paper (Sachdev, 2026), the core theoretical claims about taste are:
"Taste is the residue. It is what compounds across many reps of judgment — the curatorial pattern that tells the learner what to reach for unconsciously, what they recognise as substantive without having to evaluate from scratch."
"This is taste as a trained capacity, not inherited refinement — the residue of doing the work many times, available to anyone who runs the loop, not a marker of background or class."
"The trajectory: build empowers → curiosity → agency → resourceful experimentation → thoughtful judgment, refined over reps into taste → with sustained practice, grit is trained. The framework treats taste as something the loop produces. That is what makes it teachable."
The paper therefore claims:
- Taste is produced by repeated practice (build→compare→reflect loop)
- Taste is a trained capacity, not inherited
- Taste is the unconscious/subconscious residue of accumulated judgment acts
- Taste manifests as curatorial pattern-recognition
- Taste is teachable because it is produced by a loop anyone can run
Kant's Critique of Judgment (1790) is the foundational text treating taste as a cognitive capacity. In the "Critique of Aesthetic Judgment," Kant analyzes judgments of taste as distinct from both cognitive judgments (which determine what is) and moral judgments (which determine what ought to be). Key features:
- Taste is the "capacity for judging the beautiful" (Kant, 1790, §1). It is a faculty — not merely preference, but a mode of cognition.
- Judgments of taste claim subjective universality (§6-8): when I judge something beautiful, I demand that others ought to agree, even though the judgment is based on feeling, not concepts. This is the paradox that makes taste more than mere liking.
- Taste involves "free play" of imagination and understanding (§9): the cognitive faculties engage in harmonious free play without being determined by a concept. This is a cognitive account — taste is a mental activity, not passive reception.
- Taste is connected to the sensus communis (§20-22): a shared sense that grounds the claim to universal agreement. This is not empirical consensus but a normative ideal — what ought to be shared.
Relevance to BuildToLearn: Kant establishes the template for treating taste as a cognitive capacity rather than mere preference. However, Kant's account is transcendental (concerned with the conditions of possibility for taste), not developmental. He does not address how taste is acquired — it is treated as a universal faculty of the mind, not something trained through practice. The paper's claim that taste is "trained" would need to depart from Kant on this point, treating taste as acquired rather than innate as a faculty.
Key citation: Kant, I. (1790/2000). Critique of the Power of Judgment (P. Guyer & E. Matthews, Trans.). Cambridge University Press.
Bourdieu's Distinction (1979/1984) provides the most empirically rigorous treatment of taste as a socially acquired disposition:
- Taste is classifies, and it classifies the classifier (Bourdieu, 1984, p. 6). Taste preferences are not innocent; they operate as markers of social position.
- Cultural capital — education, intellect, style — determines what counts as "good taste" within a society. Those with high cultural capital define legitimate taste; those with less accept it as natural.
- Habitus is the mechanism: "a system of durable, transposable dispositions" (Bourdieu, 1990, p. 53) — structured structures predisposed to function as structuring structures. Habitus is acquired through experience (upbringing, education, exposure) and operates below conscious awareness.
- Taste is the practical operator of habitus: it translates social position into seemingly natural preferences. Bourdieu's empirical data (survey of 1,217 respondents in 1960s France) showed systematic correlations between class position and aesthetic preferences across music, art, food, and home decoration.
Relevance to BuildToLearn: Bourdieu provides the strongest precedent for taste as an acquired capacity operating below conscious awareness. His habitus concept closely parallels the paper's "residue" metaphor — both describe patterns that form through repeated exposure and operate without deliberate evaluation. The critical difference: Bourdieu's taste is class-acquired, while Sachdev's is practice-acquired. Bourdieu would see taste as a marker of social position; Sachdev claims it as a marker of iterative work. The paper explicitly disclaims the class reading ("not a marker of background or class"), but must contend with the fact that Bourdieu's framework is the dominant empirical treatment of taste as acquired disposition.
However, there is a supporting strain within Bourdieu scholarship: Lizardo (2004) argues that habitus has "cognitive origins" — it is not merely sociological but operates through cognitive mechanisms. This partially bridges to the cognitive science literature below.
Key citations:
- Bourdieu, P. (1984). Distinction: A Social Critique of the Judgement of Taste (R. Nice, Trans.). Harvard University Press. (Original work published 1979)
- Bourdieu, P. (1990). The Logic of Practice. Stanford University Press.
- Lizardo, O. (2004). The cognitive origins of Bourdieu's habitus. Journal for the Theory of Social Behaviour, 34(4), 375-401.
What is supported: There is a 200+ year philosophical tradition treating taste as a cognitive capacity (not mere liking) and a 40+ year sociological tradition treating it as an acquired disposition operating below consciousness.
What is not supported by this tradition: The claim that taste is produced by a specific kind of practice loop (build→compare→reflect) rather than by broad social exposure and upbringing. This is the paper's novel contribution.
The design fields have long treated "developing taste" or "aesthetic judgment" as a pedagogical goal, though the language varies:
- Schön's reflective practitioner (1983, 1987): The core of design education is the "reflective conversation with the situation" — the designer acts, the situation "talks back," and the designer reflects-in-action. This is the closest precedent to the BuildToLearn loop. Schön describes how design students develop "repertoire" — a collection of exemplars, patterns, and responses built up through experience that allows them to recognize situations and respond without deliberating from first principles. This is essentially the "residue" claim in different language.
- Studio pedagogy: The dominant model in architecture and design education since the Bauhaus (1919-1933). The "crit" (critique) is the core mechanism: a student produces work, presents it, and receives critique from faculty and peers. Through repeated crits, students internalize standards of judgment. This is a direct parallel to the BuildToLearn loop's steps 3-4 (compare/choose, peer review).
- "Connoisseurship" in art education: Eisner (1998, 2002) developed the concept of "educational connoisseurship" — the ability to make fine-grained qualitative discriminations developed through experience and refined attention. Eisner explicitly frames this as a learned capacity connected to criticism (the art of disclosure, not negative judgment).
Key citations:
- Schön, D. A. (1983). The Reflective Practitioner: How Professionals Think in Action. Basic Books.
- Schön, D. A. (1987). Educating the Reflective Practitioner. Jossey-Bass.
- Eisner, E. W. (1998). The Enlightened Eye: Qualitative Inquiry and the Enhancement of Educational Practice. Prentice Hall.
- Eisner, E. W. (2002). The Arts and the Creation of Mind. Yale University Press.
Creative writing pedagogy has a parallel concept: "reading like a writer" — the idea that writers develop taste/judgment through close reading of exemplars, not through abstract aesthetic theory. Key works:
- Prose (1985) argued that creative writing is learned by "reading closely, reading carefully, reading with an eye for technique." This is taste-through-exposure.
- Gardner (1983) discussed the "fictional dream" and the writer's need to develop an internal critic that judges whether the dream holds. This internal critic is developed through practice and reading.
Key citations:
- Prose, F. (2006). Reading Like a Writer: A Guide for People Who Love Books and for Those Who Want to Write Them. Harper Perennial.
- Gardner, J. (1983). The Art of Fiction: Notes on Craft for Young Writers. Vintage.
Supported: The claim that taste/judgment develops through iterative practice with critique is well-established in design pedagogy. Schön's "repertoire" and the studio crit model directly parallel the BuildToLearn loop. The paper's framing is consistent with, but does not significantly extend, existing design education theory — except in one respect: the claim that AI tools accelerate and democratize the loop by making production cheap.
Not supported: The specific mechanism of "compare multiple AI-generated variants" is novel (AI didn't exist when Schön and Eisner wrote). The claim that taste-as-residue from AI-assisted iteration is equivalent to taste developed through traditional studio practice is untested.
The cognitive science of expertise provides the strongest empirical foundation for taste as a trained capacity:
- Perceptual learning (Gibson, 1969; Goldstone, 1998): Through repeated exposure, the perceptual system becomes attuned to relevant features and invariants in a domain. Experts literally see differently — they detect patterns, anomalies, and relevant features that novices miss. This is well-established across domains: radiologists reading X-rays, chess players recognizing board positions, musicians hearing chord progressions.
- Chase & Simon (1973) demonstrated that chess experts' superior memory for positions is domain-specific and based on pattern recognition ("chunks"), not general memory capacity.
- Ericsson's deliberate practice (Ericsson et al., 1993): Expert performance is acquired through thousands of hours of deliberate practice — activities specifically designed to improve performance, with immediate feedback and opportunities for repetition. The key finding: it's not just practice but deliberate practice (with feedback, at the edge of competence) that produces expertise.
How this maps to taste: If taste is pattern-recognition applied to artifact quality, then perceptual learning explains how repeated exposure produces taste — the cognitive system becomes attuned to quality-relevant features. Ericsson explains what kind of practice produces it — deliberate, feedback-rich, at the edge of competence. The BuildToLearn loop (build → compare → reflect → peer review) is a deliberate-practice structure for developing design/curatorial judgment.
Key citations:
- Gibson, E. J. (1969). Principles of Perceptual Learning and Development. Appleton-Century-Crofts.
- Goldstone, R. L. (1998). Perceptual learning. Annual Review of Psychology, 49(1), 585-612.
- Chase, W. G., & Simon, H. A. (1973). Perception in chess. Cognitive Psychology, 4(1), 55-81.
- Ericsson, K. A., Krampe, R. T., & Tesch-Römer, C. (1993). The role of deliberate practice in the acquisition of expert performance. Psychological Review, 100(3), 363-406.
The empirical study of aesthetic judgment has grown substantially, particularly in the last two decades:
- Leder et al. (2004) proposed a cognitive model of aesthetic appreciation with five stages: perception, implicit classification, explicit classification, cognitive mastering, and evaluation. Expertise influences every stage: experts process art differently, attend to different features, and reach different evaluative conclusions than novices.
- The fluency theory (Reber, Schwarz, & Winkielman, 2004): aesthetic pleasure is partly a function of processing fluency — stimuli that are easier to process are liked more. Expertise increases fluency for domain-relevant stimuli while also making experts more sensitive to violations (disfluency signals something interesting).
- Neuroaesthetics (Zeki, 1999; Chatterjee & Vartanian, 2014): fMRI studies show that art experts and novices show different patterns of brain activation when viewing art. Experts show greater activation in regions associated with cognitive control and memory retrieval, suggesting more top-down processing.
Key citations:
- Leder, H., Belke, B., Oeberst, A., & Augustin, D. (2004). A model of aesthetic appreciation and aesthetic judgments. British Journal of Psychology, 95(4), 489-508.
- Reber, R., Schwarz, N., & Winkielman, P. (2004). Processing fluency and aesthetic pleasure: Is beauty in the perceiver's processing experience? Personality and Social Psychology Review, 8(4), 364-382.
- Chatterjee, A., & Vartanian, O. (2014). Neuroaesthetics. Trends in Cognitive Sciences, 18(7), 370-375.
The cognitive science of intuition provides a mechanism for the "residue" claim:
- Klein's recognition-primed decision model (Klein, 1998): Expert decision-making in naturalistic settings is not a process of comparing options but of recognition — the expert sees a situation, recognizes it as an instance of a pattern, and the first course of action that comes to mind is usually the right one. This pattern-recognition is built through extensive experience. Klein explicitly describes this as "intuition" and argues it is not mystical but a product of experience-based pattern matching.
- Dual-process theory (Kahneman, 2011; Evans & Stanovich, 2013): System 1 (fast, automatic, unconscious) operates through learned associations and pattern recognition. System 2 (slow, deliberate, conscious) is engaged for novel or difficult problems. Expertise development is partly the migration of skills from System 2 to System 1 — what once required deliberation becomes automatic.
This is the closest empirical parallel to the "residue" claim: Taste, in BuildToLearn's framing, is what happens when repeated acts of deliberate judgment (System 2) produce automatic pattern-recognition (System 1) — "what they recognise as substantive without having to evaluate from scratch." The dual-process literature provides a mechanism. Klein provides the naturalistic evidence.
Key citations:
- Klein, G. (1998). Sources of Power: How People Make Decisions. MIT Press.
- Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
- Evans, J. S. B., & Stanovich, K. E. (2013). Dual-process theories of higher cognition: Advancing the debate. Perspectives on Psychological Science, 8(3), 223-241.
Strongly supported: Taste as a trained pattern-recognition capacity developed through repeated exposure with feedback is consistent with the perceptual learning, expertise, and intuition literatures. The claim that taste operates "unconsciously" and allows recognition "without evaluating from scratch" parallels Klein's recognition-primed decisions and dual-process theory.
Partially supported: The specific claim that taste is "the residue of running many reps of a build→compare→reflect loop with AI tools" — the general mechanism (deliberate practice with feedback) is well-supported; the AI-specific mechanism is untested.
Not supported: No studies directly examine whether taste developed through AI-assisted iteration transfers to non-AI contexts, or whether it develops at comparable quality to taste developed through traditional studio/apprenticeship methods.
Andrej Karpathy's "A Survival Guide to a PhD" (2016) contains the most cited AI-practitioner discussion of taste:
"Developing taste. When it comes to choosing problems you'll hear academics talk about a mystical sense of 'taste'. It's a real thing."
"In particular, I think I had a terrible taste coming in to the PhD. I can see this from the notes I took in my early PhD years. A lot of the problems I was excited about at the time were in retrospect poorly conceived, intractable, or irrelevant. I'd like to think I refined the sense by the end through practice and apprenticeship."
Karpathy describes taste as involving evaluation of a problem's "importance, difficulty, its sexiness, its historical context." He explicitly frames it as acquired through practice and apprenticeship — directly supporting the paper's claim that taste is trained.
Source: Karpathy, A. (2016, September 7). A Survival Guide to a PhD. https://karpathy.github.io/2016/09/07/phd/
- Paul Graham (essays on "taste" in startups and makers): Graham has written extensively about taste in the context of startups and making things, framing it as developed through making many things and developing an eye for quality.
- Bret Victor and the "inventor's vision": Victor's work on "inventing on principle" and "seeing spaces" implicitly relies on a concept of taste — the ability to see what should exist.
- "Good taste" in software engineering: The concept of "taste" appears frequently in software engineering discourse (e.g., Linus Torvalds on "good taste" in code, Rob Pike on elegance). It is typically treated as something developed through extensive practice and exposure to good and bad code.
Supported: The AI/ML practitioner community treats taste as a real, developed capacity — not mystical but acquired through practice, apprenticeship, and exposure to exemplars. This aligns with the paper's framing of taste as trained.
Note: Practitioner discourse is anecdotal and experiential, not empirical. It provides cultural resonance and face validity, not scientific evidence.
The education literature on judgment development provides several relevant frameworks:
- Boud's reflective practice (Boud et al., 1985; Boud, 2000): Reflection is the mechanism by which experience is converted into learning. Boud distinguishes between "reflection-in-action" (during the experience) and "reflection-on-action" (after the experience). The BuildToLearn loop's step 5 (write the reflection) is explicitly framed as a Boud-style reflection.
- Kolb's experiential learning cycle (Kolb, 1984): Concrete experience → reflective observation → abstract conceptualization → active experimentation. The BuildToLearn loop maps onto this fairly directly: build = concrete experience, compare/peer review = reflective observation, reflection = abstract conceptualization, frame-the-next-build = active experimentation.
- Comparative judgment in assessment (Christodoulou, 2020; Pollitt, 2012): Comparative judgment (asking assessors to compare pairs of artifacts and decide which is better) produces more reliable assessments than rubric-based marking. It also develops assessors' judgment through the act of comparing. This directly supports the paper's claim that the act of comparing develops taste — the assessor gets better at judgment through the act of judging.
Key citations:
- Boud, D., Keogh, R., & Walker, D. (1985). Reflection: Turning Experience into Learning. Kogan Page.
- Kolb, D. A. (1984). Experiential Learning: Experience as the Source of Learning and Development. Prentice-Hall.
- Christodoulou, D. (2020). Teachers vs Tech?: The Case for an Ed Tech Revolution. Oxford University Press.
The term "curatorial judgment" appears in museum studies and arts education, but rarely in general educational theory:
- Curatorial pedagogy (O'Neill & Wilson, 2010): The curatorial turn in contemporary art and education treats curation as a mode of knowledge production — selecting, arranging, and contextualizing artifacts as an epistemic act.
- "Curation as pedagogy" has been discussed in digital literacy contexts (Mihailidis & Cohen, 2013): the ability to find, filter, and make sense of information abundance. This is related to but distinct from the paper's use of curation.
The paper's use of "curatorial" to describe taste is a novel extension — applying curation-as-epistemic-act to the judgment of AI-generated artifacts.
Key citations:
- O'Neill, P., & Wilson, M. (Eds.). (2010). Curating and the Educational Turn. Open Editions.
- Mihailidis, P., & Cohen, J. N. (2013). Exploring curation as a core competency in digital and media literacy education. Journal of Interactive Media in Education, 2013(1).
Supported: The general mechanism (iterative practice with reflection produces judgment) is well-established in experiential learning theory and reflective practice literature.
Partially supported: Comparative judgment as a mechanism for developing taste has empirical support in assessment contexts, but not yet studied as a pedagogical mechanism for general taste development.
Not supported: "Curatorial judgment" as a named pedagogical construct is thin. The paper is doing genuine theoretical work by naming and operationalizing it in this way.
- Taste is a real cognitive phenomenon, not merely preference. (Kant, Bourdieu, cognitive science of expertise)
- Taste can be acquired/developed through experience. (Bourdieu's habitus, perceptual learning, Ericsson's deliberate practice, Karpathy's self-report)
- Taste operates partly below conscious awareness — as pattern recognition, not deliberate evaluation. (dual-process theory, Klein's recognition-primed decisions, Bourdieu's habitus)
- Iterative practice with feedback develops domain-specific judgment. (perceptual learning, deliberate practice, studio pedagogy)
- The act of comparing artifacts develops comparative judgment. (Christodoulou, Pollitt; empirical support in assessment literature)
- Reflection converts experience into learning. (Boud, Kolb, Schön; extensive empirical support)
- Taste as specifically the "residue" of judgment acts — the metaphor is novel but mechanistically consistent with dual-process theory (System 2 → System 1 migration) and Klein's recognition-primed decisions.
- Taste as teachable through a structured loop — the mechanism (deliberate practice) is well-supported; whether the specific BuildToLearn loop produces taste at comparable quality to traditional methods is untested.
- "Curatorial pattern" as the operational form of taste — the concept has resonance with comparative judgment and curation studies but has not been empirically operationalized.
- AI-assisted iteration produces taste equivalent to (or better than) traditional methods — completely untested. This is the paper's bet, not a finding.
- Taste developed through one domain's AI-assisted loop transfers to other domains — no evidence. Transfer of expertise is generally narrow.
- "When raw execution becomes a commodity, curation is what's scarce" — this is an economic/social claim, not an empirical one. It has face validity but no direct evidence.
- The specific trajectory: curiosity → agency → resourceful experimentation → judgment → taste → grit — the sequential causal claim is untested. The individual components have varying empirical support.
The weakest claim about taste that is empirically defensible and still does useful work in a learning framework:
"Taste is a domain-specific pattern-recognition capacity that develops through repeated acts of comparative judgment with feedback, and once developed, operates partly below conscious awareness — allowing faster, more accurate evaluations without deliberate analysis."
Why this is the weakest defensible claim:
-
Every element has empirical grounding:
- "Domain-specific pattern-recognition capacity" — perceptual learning (Gibson, Goldstone), expertise research (Ericsson, Chase & Simon)
- "Develops through repeated acts" — deliberate practice (Ericsson)
- "Comparative judgment with feedback" — comparative judgment (Christodoulou, Pollitt), studio pedagogy
- "Operates partly below conscious awareness" — dual-process theory (Kahneman, Evans & Stanovich), recognition-primed decisions (Klein)
- "Faster, more accurate evaluations" — expert-novice differences (Chase & Simon, Klein)
-
It drops the strongest but untestable claims:
- Does NOT claim taste is produced by AI-assisted iteration specifically
- Does NOT claim transfer across domains
- Does NOT make the economic scarcity argument
- Does NOT claim a specific causal trajectory (curiosity → agency → ...)
- Does NOT claim taste is the "residue" (a metaphor — the weaker claim describes the mechanism)
-
It still does useful work in the learning framework:
- It justifies designing learning experiences around repeated acts of judgment with feedback
- It explains why the compare-and-choose step matters: it's the mechanism that builds taste
- It explains why taste feels automatic once developed: it's migrated from System 2 to System 1
- It makes taste teachable: if taste is pattern-recognition developed through deliberate comparative practice, you can design for it
What the paper can say beyond this weakest claim (as honest argument rather than established finding):
The paper's stronger framing ("taste is the residue that compounds across many reps of the build→compare→reflect loop with AI tools") is a hypothesis worth testing, not an established finding. It has theoretical coherence (consistent with the mechanisms above) and practitioner resonance (consistent with Karpathy, studio pedagogy), but the AI-specific mechanism is novel and untested. The paper should present it as such — an argument about where the practice should go, backed by theoretical precedent, awaiting empirical validation — and it already does this for most of its stronger claims (e.g., the workforce transferability bet in §2.6).
-
Ground the "residue" metaphor in dual-process theory. The migration from System 2 (deliberate judgment) to System 1 (automatic pattern recognition) provides a cognitive mechanism for what the paper calls "residue." This is the strongest bridge from the paper's practitioner language to the empirical literature. Cite Kahneman (2011) and Klein (1998).
-
Use Schön's "repertoire" as a bridge concept. Schön's reflective practitioner framework already describes how professionals develop pattern libraries through iterative practice. The paper's taste claim extends this to AI-assisted practice. Cite Schön (1983, 1987) explicitly.
-
Frame Ericsson's deliberate practice as the mechanism. The build→compare→reflect loop is deliberate practice for curatorial judgment. Making this explicit strengthens the theoretical grounding: the loop works because it provides the four elements of deliberate practice (specific task, feedback, repetition, edge-of-competence).
-
Distinguish clearly between the weak (defensible) and strong (hypothetical) claims. The weakest defensible claim (§8 above) can be stated as established theory. The stronger claims about AI-specific mechanisms should be flagged as hypotheses.
-
Address the Bourdieu tension directly. The paper's claim that taste is "not a marker of background or class" contradicts Bourdieu's empirical findings. This tension is productive — the paper can argue that AI tools change the mechanism by which taste is acquired, making it practice-dependent rather than class-dependent. But this is a claim that needs defending, not assuming.
-
Cite Karpathy (2016) for practitioner resonance. His self-report of developing taste "through practice and apprenticeship" provides face validity for the paper's core claim, even though it's anecdotal rather than empirical.
- Kant, I. (1790/2000). Critique of the Power of Judgment. Cambridge University Press.
- Bourdieu, P. (1984). Distinction: A Social Critique of the Judgement of Taste. Harvard University Press.
- Bourdieu, P. (1990). The Logic of Practice. Stanford University Press.
- Lizardo, O. (2004). The cognitive origins of Bourdieu's habitus. Journal for the Theory of Social Behaviour, 34(4), 375-401.
- Schön, D. A. (1983). The Reflective Practitioner. Basic Books.
- Schön, D. A. (1987). Educating the Reflective Practitioner. Jossey-Bass.
- Eisner, E. W. (1998). The Enlightened Eye. Prentice Hall.
- Gibson, E. J. (1969). Principles of Perceptual Learning and Development. Appleton-Century-Crofts.
- Chase, W. G., & Simon, H. A. (1973). Perception in chess. Cognitive Psychology, 4(1), 55-81.
- Ericsson, K. A., Krampe, R. T., & Tesch-Römer, C. (1993). The role of deliberate practice in the acquisition of expert performance. Psychological Review, 100(3), 363-406.
- Klein, G. (1998). Sources of Power: How People Make Decisions. MIT Press.
- Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
- Goldstone, R. L. (1998). Perceptual learning. Annual Review of Psychology, 49(1), 585-612.
- Reber, R., Schwarz, N., & Winkielman, P. (2004). Processing fluency and aesthetic pleasure. Personality and Social Psychology Review, 8(4), 364-382.
- Leder, H., et al. (2004). A model of aesthetic appreciation and aesthetic judgments. British Journal of Psychology, 95(4), 489-508.
- Karpathy, A. (2016). A Survival Guide to a PhD. https://karpathy.github.io/2016/09/07/phd/
- Boud, D., et al. (1985). Reflection: Turning Experience into Learning. Kogan Page.
- Kolb, D. A. (1984). Experiential Learning. Prentice-Hall.
- Christodoulou, D. (2020). Teachers vs Tech?. Oxford University Press.
- O'Neill, P., & Wilson, M. (2010). Curating and the Educational Turn. Open Editions.
Review compiled August 2026. All citations verified against accessible sources where possible; some theoretical claims rely on training knowledge of canonical texts. The SEP entries for Kant's aesthetics and perceptual learning were accessed and confirm the key theoretical frameworks. The Karpathy (2016) passage was extracted directly from the source page. The Bourdieu Wikipedia page confirms the theoretical framework of Distinction. Additional searches should verify: specific page numbers for Bourdieu quotes, the Leder et al. (2004) model details, and whether there is any post-2023 empirical work on AI-assisted taste development.