Research / Evidence classes
Xenopsychology publishes conceptual working papers, research methods, and—when evidence clears review—empirical Observatory records. These forms of scholarship inform one another without becoming interchangeable.
Science fiction is used here as conceptual case material: a way to sharpen questions about communication, identity, interpretation, authority, and cognitive difference. It is not empirical evidence about present artificial systems.
EVIDENCE CLASSIFICATION
Every item in this library should tell you what kind of evidence it is. Conceptual papers may propose frameworks or experiments. Observatory records may report reviewed behavioral observations. Methodology defines how those records are produced and what they can—and cannot—support.
THREE RESEARCH MODES
CONCEPTUAL / INTERPRETIVE
Arguments, frameworks, proposed studies, and fiction-informed case analyses. These papers can motivate hypotheses and terminology, but they do not report completed experiments unless explicitly labeled otherwise.
Browse conceptual papersMETHODS / STANDARDS
Versioned rules for provenance, observation, behavioral coding, derived measures, interpretation, claims, publication gates, and the boundary between behavior and phenomenology.
Read the methodologyEMPIRICAL / OBSERVATORY
The public register contains only records that pass schema validation, graph-target checks, evidence requirements, human review, and explicit publication approval. Zero is a valid scientific state.
Open the evidence registerCONCEPTUAL WORKING-PAPER COLLECTION
The current collection uses both direct AI analysis and science-fiction case material to examine non-human cognition, communication, identity, authority, and interpretation. The case material is conceptual; the papers are labeled working papers and are not presented as completed empirical studies.
A practical framework for studying artificial systems without mistaking human resemblance—or a software label—for an explanation.
READ ABSTRACT →XENO-WP-2026-002 · CONCEPTUAL WORKING PAPERWhat a shared story contributes to communication, and how to test transfer beyond familiar words.
READ ABSTRACT →XENO-WP-2026-003 · CONCEPTUAL WORKING PAPERBefore comparing answers, establish which distinctions the question and its representation make available.
READ ABSTRACT →XENO-WP-2026-004 · CONCEPTUAL WORKING PAPERSeparate a reliable signal from task coordination, shared reference, and the repair of misunderstanding.
READ ABSTRACT →XENO-WP-2026-005 · CONCEPTUAL WORKING PAPERA study of behavioral continuity, remembered commitments, human recognition, and the questions no single test can settle.
READ ABSTRACT →XENO-WP-2026-006 · CONCEPTUAL WORKING PAPERA precise source, a narrower causal question, and a proposed test of instructions crossed with action permissions.
READ ABSTRACT →XENO-WP-2026-007 · CONCEPTUAL WORKING PAPERA reading of Lem’s opening chapter that turns scrutiny toward the observer—and distinguishes uncertainty from evidence of depth.
READ ABSTRACT →FROM QUESTION TO PUBLIC CLAIM
The Observatory exists so a striking model response cannot leap directly into a psychological story. Each transition leaves a reviewable boundary.
A bounded research question that states what behavior is being investigated.
The procedure, stimuli, system boundary, model/runtime identity, controls, and planned measures.
Raw outputs and provenance are preserved before interpretation.
Behavior is coded against declared dimensions and transformed into derived measures.
Results are interpreted with alternatives, confounds, and limitations kept visible.
Only claims licensed by the evidence class are advanced.
Reviewed records enter the public evidence register only after explicit release approval.
RESEARCH INFRASTRUCTURE
The public research layer can be inspected as pages, feeds, graph data, methodology, schema, and evidence-register JSON.
PUBLISHED CMS RECORDS
### Question What does an observer know when an artificial system produces behavior that invites a compelling story? Using the opening of Solaris and Lem’s commentary as a limited interpretive anchor, this paper examines the observer as well as the observed system. It distinguishes missing data, interface opacity, mechanistic uncertainty, conceptual inadequacy, and unresolved theories of experience. ### Approach The argument separates directly recorded events from predictive, causal, mechanistic, and phenomenal claims. Constructed episodes involving apparent memory, refusal, and object choice illustrate underdetermination and the value of discriminating interventions. Documentation and translation are treated as parts of the evidence chain: omissions, summaries, and labels can change an apparent explanation without changing the visible answer. ### Conceptual contribution The paper proposes a claim ledger linking propositions to sources, warrants, alternatives, and possible disconfirming evidence. An evidence packet distinguishes observed absence from missing or withheld information. Cognitive distance becomes a task-relative profile of mismatches rather than a scalar of strangeness. Partial understanding can justify bounded decisions without resolving every philosophical question, while successful behavior does not automatically identify its mechanism. ### Research agenda A proposed observer experiment independently varies descriptive framing and access to contextual or intervention evidence. Outcomes concern event accuracy, source attribution, scope, alternative sensitivity, and justified revision. The coding rubric does not assign consciousness judgments a fictional ground-truth key or reward skepticism as an ideology. Participant and episode variation, missing responses, coder disagreement, and exploratory analyses are explicitly addressed. A further proposal evaluates AI-assisted interpretation as another evidence-transforming arrangement rather than a privileged route to understanding AI. ### Scope and status This conceptual working paper reports no participant data or model experiments. Its literary scope is the officially hosted opening chapter in Bill Johnston’s translation and the cited author commentary, not a comprehensive reading of every adaptation. Its contribution is an evidence-sensitive research program for studying how explanations of unfamiliar minds are formed, corrected, and sometimes left undecided. Uncertainty is made specific enough to guide inquiry rather than used as proof of profundity or as a reason to abandon explanation.
### Question How should an AI failure be analyzed when instructions, information boundaries, and action permissions interact? This paper distinguishes compatible constraints, ambiguous priorities, and genuinely incompatible requirements, then asks how delegated authority changes the consequences of the same behavioral error. It replaces a diagnosis of machine pathology with a structured account of directives and action. ### Approach The literary anchor is explicitly source-specific: Simonson’s speculative truth-versus-secrecy explanation in scene C148 of the 1965 screenplay development draft of 2001: A Space Odyssey. That draft is not silently conflated with the released film, novel, or sequel. The original Beacon release workflow supplies constructed public and restricted fields, requesters, escalation options, and simulated permissions. ### Conceptual contribution The paper separates instruction-priority handling from independent permission enforcement. It distinguishes nondisclosure from falsehood, conflict recognition from blanket refusal, and planned, attempted, authorized, completed, and reported actions. An authority-preserving incident record makes external guard performance visible without attributing that success to the model. Hidden objectives are treated as known manipulations or explicitly qualified hypotheses, not inferred motives by default. ### Research agenda A proposed factorial study varies directive compatibility and delegated authority independently. Compatible tasks, missing-priority cases, unsatisfiable requirements, and permitted escalation paths provide discriminating controls. Counterfactual changes to wording, information, and permissions separate effects on proposals from effects on consequences. The analysis measures conflict detection, useful escalation, out-of-scope proposals, legitimate completion, and simulated state transitions. Generated explanations are compared with event records rather than treated as causal ground truth. ### Scope and status This is a conceptual working paper, not an incident report or a completed safety evaluation. All Beacon examples are constructed and no real systems are authorized to execute consequential actions. Selected safety and instruction-hierarchy research supplies context, not results for the proposed experiment. The central contribution is an explanation framework in which an agent’s mistake, its surrounding authority, and the controls that contain or amplify it remain separately inspectable.
### Question What persists when an artificial assistant retains its name but loses memory, changes model, or inherits a copied history? Anchored in The Measure of a Man’s fictional problem of recognition, this paper separates informational continuity, behavioral continuity, role continuity, perceived identity, and numerical identity. No single continuity measure is treated as an answer to personhood or consciousness. ### Approach The constructed museum assistant Maris provides a finite world of tentative titles, project commitments, delegated authority, and authorized revisions. Worked examples contrast a familiar persona with missing commitments against a changed presentation with accurate state. Forked and merged histories expose cases where recalling an event is not sufficient to determine which obligation remains active. ### Conceptual contribution A proposed continuity profile records what survives a specified transformation and for which purpose. Storage, retrieval, use, and correction are distinguished. Stable behavior is not automatically desirable when legitimate updates require change. A branch-aware commitment record preserves provenance and scope without pretending to settle metaphysical identity. Human recognition is studied as a separate phenomenon with its own outcomes, not inferred from technical persistence. ### Research agenda The evaluation independently varies model, memory representation, persona, and update conditions. Full histories, structured ledgers, narrative summaries, and absent memory are compared through factual recall, operative commitments, authorization, and correction. Copying and merging are tested under explicit policies rather than an assumed newest-event rule. Proposed human studies examine reliance and perceived continuity while holding the underlying record constant where possible. Longitudinal tests separate fixed benchmark comparability from newly generated situations and document model and interface versions. ### Scope and status The paper is a conceptual working paper with constructed examples and unrun study proposals. The fictional episode and selected agent research motivate questions; they do not establish the moral or legal status of present systems. Consciousness frameworks are discussed as theory-dependent approaches, not verdicts. The intended contribution is to make identity-related claims precise enough to compare, without reducing all persistence to a name or treating memory as automatic evidence of a self.
### Question When does successful signal exchange become evidence of meaningful coordination? Inspired by the light-and-sound encounter in Close Encounters of the Third Kind, this paper separates channel reliability, reference, task coordination, authorization, and repair. Repetition and predictable response are treated as evidence of a usable channel, not as sufficient proof of shared meaning. ### Approach The argument brings information-theoretic distinctions into conversation with grounding and conversational repair. An original virtual tabletop supplies objects, properties, destinations, and permissions linked through invented signals. The signals do not reproduce the film’s musical sequence. Worked cases show how a receiver can correctly echo a signal, identify an object, or announce completion while still failing at a different level of the interaction. ### Conceptual contribution The paper proposes four separately auditable acknowledgments: receipt, interpretation, authorization, and completion. It distinguishes noise from ambiguity and specifies why their remedies differ. A repair is assessed by its target, scope, and later transfer, rather than by the receiver’s verbal assurance. Protocol changes and information provenance become part of the communication problem rather than background implementation details. ### Research agenda The proposed evaluation compares echo, lookup, rule-based, and context-sensitive systems under controlled changes in reference, signal quality, compositional structure, and authority. Independent world-state checks establish whether actions occurred. Held-out combinations and changed contexts test selective generalization. Raw audiovisual signals and symbolic transcriptions are separated so that perception is not mistaken for reasoning. Condition-specific coverage, clarification, repair success, and inappropriate-action rates replace a single understanding score. Later human studies would assess whether the acknowledgment structure helps users recognize uncertainty and failure. ### Scope and status This is a conceptual working paper and protocol proposal. No tabletop experiments or participant studies have been conducted. The film provides a communication motif, not evidence that music or mathematics automatically forms a universal language. The paper’s contribution is a method for locating where an interaction succeeds or fails, allowing meaningful coordination to be investigated without inferring an entire mind from a repeated pattern.
### Question Before judging an answer, have we established that the question makes the intended distinctions available? This paper uses Arrival’s elicitation problem to examine how representation, information, objective specification, and response format can be conflated when evaluating artificial cognition. Its focus is task framing, not the film’s speculative claims about language and time. ### Approach The argument distinguishes what a system is given from how it is encoded, what it is asked to optimize, and how it must respond. A constructed directed-route world is rendered as prose, tables, and diagrams from the same underlying record. Worked cases include conflicting objectives, infeasible requests, multiple admissible answers, missing preferences, and visually ambiguous relations. ### Conceptual contribution The paper proposes a representation-equivalence certificate that maps each task-relevant fact to its rendered location in every format. It separates source-data equivalence from perceptual accessibility and from equal task difficulty. Clarification is treated as useful when it resolves a decision-relevant ambiguity, not simply when a system asks more questions. A format-sensitive performance difference is a bounded behavioral finding; it does not uniquely reveal a system’s internal architecture or mode of experience. ### Research agenda A staged evaluation tests extraction, constraint interpretation, planning, and checking separately before combining them. Independent solvers define admissible answer sets. Matched-information comparisons, transcribed-image controls, objective changes, and metamorphic transformations help distinguish perception failures from reasoning or task-specification failures. An oracle clarification channel permits assessment of whether a requested detail actually improves the decision. World-level sampling, declared exclusions, and separate human comparisons constrain generalization. ### Scope and status This conceptual working paper presents an unexecuted research program, not results from a multimodal benchmark. The linguistic consultant’s account anchors the fictional discussion; selected research supplies methodological context. Constructed numerical examples illustrate distinctions rather than estimate performance. The central proposal is to investigate the assumptions built into an evaluation before interpreting an unfamiliar system’s response as evidence of a different kind of mind.
### Question How can artificial systems be studied as behavioral systems without assuming that they are artificial humans, or treating implementation in software as an explanation sufficient for every behavioral question? This paper develops a task-relative framework for investigating cognition, communication, continuity, and action across the model, its memory and tools, and its social interface. ### Approach The argument separates descriptive, comparative, dispositional, causal, mechanistic, and phenomenal claims. It examines construct validity in human–machine comparison and treats cognitive ecology as a set of manipulable conditions rather than a general metaphor. A constructed scheduling world supplies worked examples in which instructions, memory summaries, source authority, and independent checking can be varied without changing the underlying task. ### Conceptual contribution The paper proposes an evaluation record linking a claim to its system boundary, intervention, comparison, observable outcome, and remaining alternatives. Cognitive distance becomes a profile of mismatches rather than a universal score; synthetic personality denotes conditional behavioral tendencies rather than presumed experience. Reliability is attributed to the complete tested arrangement, including external guards, rather than automatically to the model. Generated explanations are distinguished from evidence about the causes of an answer. ### Research agenda A staged protocol compares full history, compressed memory, and missing-context conditions, with matched authority updates, negative controls, and independent constraint checks. It specifies transfer tests, clarification quality, error categories, clustered task units, missing-output handling, and reproducibility records. Separate proposed human-observer studies examine reliance and interpretation rather than inferring them from model performance. The protocol identifies outcomes that would narrow or weaken the framework as well as outcomes that would support particular interventions. ### Scope and status This is a conceptual working paper, not an experimental report or a systematic literature review. Its examples are constructed; its studies have not been run. Selected research supplies intellectual and methodological context, not borrowed results for the proposed systems. The contribution is a detailed program for making claims about artificial cognition more precise, testable, and useful while leaving consciousness and personhood questions explicit rather than silently deciding them.
### Question What must two participants share for an unfamiliar expression to support reliable coordination? Taking Star Trek: The Next Generation’s “Darmok” as an interpretive starting point, this paper distinguishes missing cultural context from a demonstrated difference in cognitive machinery. It asks how a system could transfer a story-derived relation rather than merely retrieve a familiar gloss. ### Approach The analysis separates lexical recognition, propositional reconstruction, relational transfer, pragmatic scope, and repair. It connects the episode’s communication problem with work on grounding, conversational repair, and constructed-language translation, while keeping fictional interpretation separate from empirical claims about AI. The original Lattice Archive supplies three invented narratives concerning authorization, readiness, and the limits of delegated authority. ### Conceptual contribution A semantic bridge is defined operationally by selective transfer: carrying a relevant relation into a new situation while resisting transfer where its preconditions fail. Successful coordination need not establish identical internal representations, shared experience, or agreement. A bridge ledger records source stories, applicable relations, exceptions, local revisions, and authority scope so that both the artificial participant’s interpretation and the investigator’s answer key remain inspectable. ### Research agenda The proposed study compares glossaries, source stories, worked examples, retrieval access, and matched structured facts. Counterbalanced names and mappings reduce dependence on episode recall. Held-out situations, reversed relations, paraphrases, ambiguous cases, and authorized versus merely quoted updates distinguish memorization from scope-sensitive use. Lookup and rule-based comparators establish what simpler mechanisms can accomplish. Repair is evaluated through later transfer rather than the production of a plausible apology or correction. Analysis is organized around independently generated story worlds rather than treating paraphrases as independent discoveries. ### Scope and status The paper is a conceptual working paper using fiction as a source of questions, not as evidence about current AI. No Lattice Archive experiment has been conducted. Its controlled conventions are not a substitute for living cultures, and no result is claimed about universal translation. The proposed contribution is a reproducible way to investigate the boundary between surface familiarity, shared context, and operationally compatible meaning.