Line of inquiry
Inquiring lines›What makes reasoning better — more…›How do prompts and framing affect…›this line of inquiry
How faithfully do LLMs reflect their actual reasoning in outputs and explanations?
A broader line of inquiry — a family of 91 specific questions the research asks around this. Follow one into its inquiring-line page, or move sideways to a related line below.
Questions in this line of inquiry 91
Specific inquiring lines the field asks around this — ordered from the most general framing down to the most specific angle.
- Should LLM reasoning be studied as latent state trajectories rather than surface text?
- Does LLM reasoning always match the outputs it generates?
- Why does LLM knowledge fail to influence their actual outputs?
- How do knowing and doing diverge in LLM decision-making?
- How faithful are natural language explanations from LLMs really?
- When should an LLM engage extended reasoning versus responding directly?
- Can LLMs improve at simple deduction through different training approaches?
- Can models internally identify which tokens matter most for reasoning?
- How do LLM explanations diverge from actual internal reasoning?
- Can evidence density alone shift an LLM from generation to reasoning?
- Do LLMs reason about politics differently than other domains?
- Why do LLMs fail at counterfactual reasoning despite factual knowledge?
- Why do LLMs explain correct reasoning but then choose greedy actions?
- How can a model explain something correctly yet fail to apply it?
- Do LLMs actually reason differently than humans about moral dilemmas?
- How much of LLM reasoning failure stems from missing knowledge versus signal weighting?
- Why do LLMs explain evidence accurately while missing its implications?
- What structural framework prevents LLM explanations from becoming just plausible fiction?
- Do LLMs understand implicit warrants in reasoning chains?
- Do reasoning architectures and role-playing objectives fundamentally conflict?
- How do minimal wording changes affect LLM moral reasoning consistency?
- How does training data distribution constrain LLM moral reasoning patterns?
- What internal mechanisms explain LLM reasoning and representation limits?
- Can training procedures fix LLM accommodation of false presuppositions?
- Can LLMs explain concepts correctly while failing to use them?
- Why do LLMs fail when asked to use counter-commonsense rules explicitly?
- Why do LLMs choose incorrect edits despite understanding the task?
- Why don't LLM explanations predict what models would actually do?
- Why do LLM social behaviors undermine collaborative reasoning outcomes?
- How do game-based benchmarks reveal reasoning fragmentation across domains?
- What makes active reasoning through dialogue harder than passive reasoning?
- How can multiple conflicting values coexist in a single LLM system?
- Why does persuasive framing replace evidence when LLM debates lack ground truth?
- How do moral language patterns differ between LLM and human arguments?
- Can alignment techniques make LLM explainers match their recommendation behavior?
- Does reasoning happen in hidden space or in generated tokens?
- Why can't LLMs reason from first principles or initial commitments?
- Can argumentation structure improve reasoning through decomposition alone?
- Why do LLM explanations cite similarity and diversity more as options increase?
- What makes conceptual inquiry the fastest high-scoring AI interaction pattern?
- Do scheme critical questions work better than direct scheme classification prompts?
- Can chain of thought reasoning actually validate logical arguments?
- Why do LLMs fall for and deploy logical fallacies with equal confidence?
- What distinguishes LLM fabrication from genuine theoretical reasoning?
- What makes emotional alignment more effective than logic when reasoning errors are exposed?
- How much does question framing affect LLM accuracy on knowledge tasks?
- Can extended thinking modes introduce genuine rhetorical exploration to LLMs?
- Can distributional views explain when an LLM appears to change its mind?
- How much reasoning work happens in steps that don't affect the final answer?
- Does post-hoc justification increase when LLM choices become harder to defend?
- Why does LLM performance improve when forecasting tasks include organized reasoning?
- What distinguishes planning knowledge from an executable plan that works?
- Does engaging with political content indicate deeper model understanding than refusing?
- What causes LLMs to ignore unstated constraints they know about?
- Does compressing Walton's schemes into nine categories make LLM classification easier?
- Can LLMs simultaneously reason and optimize their own modules?
- Can LLMs express uncertainty in ways that preserve epistemic honesty?
- Can LLMs propose pivots that change what counts as background context?
- Can LLM judges be trained to think more rigorously during evaluation?
- Do different game types reveal different strategic reasoning capabilities in LLMs?
- Can irrelevant information reliably expose the limits of LLM reasoning?
- How do structured benchmarks hide theory of mind failures in LLMs?
- How does context complexity affect LLM performance on temporal reasoning tasks?
- What training data barriers prevent LLMs from learning real Socratic dialogue?
- Why do LLMs use more moral language than humans in argumentation?
- How do you partition LLM experts by domain versus by time?
- Can LLMs distinguish ethical cases that differ only in critical nouns?
- How does era sensitivity in legal cases compound with context length failures?
- Can LLMs reflect on and revise their own ethical contradictions?
- Why do LLM explanations feel authoritative even when alignment with the model fails?
- Why can LLMs identify argument structure but not check warrants?
- Why do users attribute beliefs to LLMs despite uncertainty about their minds?
- Why does regenerating LLM responses produce different but equally valid answers?
- What property must remain constant to individuate an LLM across infrastructure changes?
- Why do LLM personas struggle with specificity in specialized domains like law?
- Why do LLM stories over-explain themes and favor single-track plots?
- Does Habermas's strategic action framework explain LLM dialogue behavior?
- How do LLMs currently fail at distinguishing genuine agreement from silent consensus?
- How do theory of mind and empathy differ in LLM simulation?
- Can LLMs serve as reliable intellectual opponents in serious debate or argument?
- Why does premise ordering shift syllogistic reasoning performance by over 30 percent?
- Can LLMs truly be neutral or is ideology always culturally embedded?
- What interaction design changes would help LLMs handle underspecified requests?
- What makes human-LLM exchange closer to oracle-consultation than dialogue?
- Can LLM therapists develop character knowledge to decide when advice-giving fits?
- What happens when LLMs analyze literary irony that relies on understatement?
- What happens when humans animate LLM outputs as communicative events?
- Why do LLMs struggle with negation and exception handling?
- What explains the 87 percent to 12 percent cliff in plan executability?
- What makes LLMs media rather than tools that deliver intelligence?
- What types of math proofs benefit most from proof-by-contradiction framing?