Research  ›  Case library  ›  Case 11: Mary’s Room / Knowledge Argument
Case 11 / 31  •  Theme: Doing

Mary’s Room / Knowledge Argument

Complete information and practical competence are different claims — without this resolving Mary’s philosophical argument.

Frank Jackson’s Mary knows all the physical facts about colour but has never experienced colour. The philosophical conclusion is contested, and Jackson later rejected his own anti-physicalist interpretation. For organisations, a narrower question remains useful: what can information alone teach, and what requires practice, feedback and transfer?

✓
Research statusOpen research question
☆
AI relevanceAI may amplify an existing effect
⚙
Core ideaDeclarative knowledge ≠ capability
⌾
Evidence baseSources and limitations summarised; independent review pending.
◇
Case typeThought experiment / knowledge argument
“
In one sentence

Complete information and practical competence are different claims — without this resolving Mary’s philosophical argument.

The storyHow the case arises and how it can mislead our thinking.

Mary is a brilliant scientist who learns everything about colour vision in a black-and-white environment. When she first sees red, the question is whether she learns something new. Jackson used this as the Knowledge Argument against physicalism, but later distanced himself from that conclusion.

What do we know?Evidence, research, critical insights and what remains unknown.

Jackson’s thought experiment does not settle the learning question. In prose-learning experiments, Butler (2010) found better transfer after one week from repeated testing with feedback than from restudying. That study does not establish our AI hypothesis or separate the effects of testing and feedback.

What is the mechanism?Which explanation best fits what we observe.

Equal access to declarative information does not guarantee equal discrimination, transfer or adaptive capability when learning experiences differ.

What does AI change?How AI may amplify the mechanism or make it visible.

AI transfer to investigate: Generative AI can provide explanations and examples that appear complete; whether they support transfer must be tested separately.

What do we want to test?The specific question we want to investigate further.

Proposed test, not yet conducted: Compare AI explanations alone with the same explanations plus active discrimination and feedback. Keep core information and time spent learning as comparable as possible. Measure immediate factual knowledge, delayed transfer and performance on novel cases.

What do we take away?What this case provisionally means — and where its limits lie.

Mary is primarily a narrative warning against equating information with capability. The empirical claim must be tested independently.

An explanation can appear complete without demonstrating transfer. A practical test can investigate learning transfer but does not settle the argument about subjective experience.

SourcesPrimary and key scholarly sources.

Jackson (1982), Epiphenomenal Qualia — p. 130

Jackson (2003), Mind and Illusion — opening passage on changed position

Butler (2010), Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying — Experiments 1b–3; pp. 1128–1131

Jackson (1982; 2003) discusses the philosophical argument and his changed position. Butler (2010) examines learning transfer. The organisational application involving AI is a separate hypothesis.

Give feedback