Loading the atlas
A living atlas of philosophical investigation. Every thesis is split into inspectable claims, structured arguments, and sourced evidence — with the uncertainty left visible.
Verified sources counts only independently checked claim–source links. This demonstration corpus has not yet completed that process.
Fragment · drag to explore · select a node
01 — Orientation
Three investigations that show the range of the atlas: a conceptual gap, a refuted analysis, and an open question about machines.
The explanatory gap is one of the most productive problems in philosophy of mind. It is relatively uncontroversial that current physical accounts do not explain why experience exists. Whether that gap is a permanent feature of explanation or a temporary research frontier is contested.
The justified-true-belief analysis is historically foundational and still captures three plausible necessary conditions. Gettier's 1963 counterexamples show that the three conditions are not jointly sufficient under a fallibilist conception of justification.
LLMs display impressive linguistic and analogical performance, but the inference from performance to understanding is contested. Skeptics emphasise the absence of grounded reference and the form-meaning distinction; defenders distinguish understanding from consciousness and point to systematic generalisation.
02 — Recently investigated
LLMs display impressive linguistic and analogical performance, but the inference from performance to understanding is contested. Skeptics emphasise the absence of grounded reference and the form-meaning distinction; defenders distinguish understanding from consciousness and point to systematic generalisation.
The extended-mind thesis uses parity arguments: if a process would count as cognitive were it inside the head, it should count as cognitive outside. Critics argue that coupling is not constitution and that the thesis blurs the boundary between the cognitive and the merely causal.
Ontological arguments are deductively elegant and notoriously controversial. The classical Anselmian version faces Kant's existence-is-not-a-predicate objection; modal versions face disputes about premise plausibility and the symmetry of possibility arguments.
The logical problem of evil is widely regarded as answered by free-will and soul-making defences. The evidential problem remains serious: much suffering appears pointless, and divine reasons for permitting it are not evident.
03 — Revised
Published assessments are versioned. A revision records what changed and why.
Consciousness depends on physical processes in the brain.
Narrowed the claim to causal dependence and separated it from the identity thesis, removing an overreach that had depressed the assessment.
Personal identity across time consists in psychological continuity.
Retained provisional support but weakened the conclusion from a biconditional to a centrality claim.
The universe has a finite past.
Retained provisional support after clarifying that past-incompleteness is not an absolute beginning.
Moral responsibility requires the ability to do otherwise.
Reaffirmed contested after adding the reasons-responsiveness alternative.
04 — Productive disagreement
These are not failures of the atlas. They are the places where competent arguments collide and the evidence underdetermines a verdict.
Strongest challenge
The knowledge argument: someone could know every physical fact about colour vision and still learn something new on first seeing red. If so, not all facts are physical facts.
Strongest challenge
The gap may be epistemic rather than ontological. We once could not explain life in physical terms; the failure of imagination is not a permanent boundary.
Strongest challenge
Causal closure threatens any non-physical cause: if every physical event has a sufficient physical cause, a distinct phenomenal property seems to add nothing. Its defenders must either deny closure or accept epiphenomenalism.
Strongest challenge
If your brain were transplanted into a new body, most people judge that you go with the brain. Animalism must either deny the intuition or explain why the organism goes with the brain.
2 investigations currently lack discriminating evidence. Underdetermination is recorded as a result, not hidden.
See them on the map05 — Orientation by subject
Nine domains, cross-linked by arguments rather than categories. Every edge carries a rationale.
Consciousness, experience, and the relationship between mental and physical descriptions of reality.
What makes a person the same person across time, and what grounds survival, memory, and responsibility.
Agency, deliberation, moral responsibility, and whether the causal order leaves room for alternative possibilities.
Justification, truth, evidence, scepticism, and the difference between belief that is warranted and belief that merely feels warranted.
Explanation, theory change, realism, underdetermination, and what predictive success licenses us to believe.
Existence, time, causation, and the largest-scale structure of the universe.
The ontology of mathematical objects, proof, and the relationship between formal systems and reality.
The problem of evil, arguments for and against divine existence, and the foundations of moral responsibility.
Understanding, representation, and whether machines can mean what they appear to say.
06 — Research activity
No fabricated real-time activity. This is the recorded, inspectable work queue.
Sample events are labelled in the activity feed. The research engine is not connected in Phase 1, so no background activity is occurring.
The method
AXIOM is designed around a staged investigation, not a single generative prompt. The research engine is specified; Phase 1 ships the atlas and its epistemic scaffolding.
Turn a broad question into a proposition precise enough to assess. Record definitions, assumptions, and the standard of proof.
Collect primary and secondary sources. Verify metadata, locate the exact claim, and mark what has actually been checked.
Construct the strongest supporting and opposing arguments. Premises carry explicit statuses; inductive moves are labelled.
Publish only what the evidence supports — including inconclusive results. Every conclusion remains revisable and versioned.
Truth is the target. Certainty is not.
Epistemic standards