Perception is a chain of transformations: signal becomes selection, selection becomes experience, experience becomes report, and report becomes inference. OBSERVE reconstructs that chain from the world side. CALIBRATE models the observer from the decision side. SYNTHESIS asks which claims survive both descriptions.
SYNTHESIS is where the two models meet.
Reconstruct an event from heterogeneous evidence. Keep observation, measurement, report, inference, source, time, scope, and unknowns distinct so the model can show exactly where a claim entered the chain.
Separate accuracy, confidence, sensitivity, and criterion. Perturb context and repeat the measurement to discover whether the observer, the decision policy, or the evidence actually moved.
Combine the evidence trace with the observer model. Preserve frames, witness report, matched reliability, and what remained unobserved; then state the strongest inference consistent with all four.
This is a teaching synthesis drawn from Andrews’s public artist statement and recent Brown / MIT work. Her practice uses illusion as a live thought experiment: not to prove that perception is defective, but to make the dynamic process of perceiving and believing visible enough to interrogate.
A scene can contain genuine information that never receives enough selection to become reportable. Missing a detail does not make the detail unreal; noticing it does not explain its cause.
Ask · What was present but not selected?Perceptual organization can complete edges, relations, and objects from partial information. The correct response is not cynicism; it is to distinguish what was specified from what was supplied.
Ask · What did perception add?Task, expectation, timing, and participation can reorganize what becomes salient without changing every physical feature of the stimulus. The audience is not merely a recorder.
Ask · What changed: world, task, or interpretation?Visual and social cues can shape what feels coherent or trustworthy. In a world of synthetic media and persuasive interfaces, authority must be earned by evidence rather than by fluency, familiarity, or framing alone.
Ask · Why did this cue deserve trust?OBSERVE follows what enters awareness and how evidence changes as it becomes report and inference. CALIBRATE models judgment through performance, confidence, sensitivity, criterion, and context. SYNTHESIS asks for a coherent account that both models can support.
The point is not “you cannot trust your senses.” The point is to locate the transformation before explaining it. UNKNOWN is not a failure of perception; it is the correct state when the apparatus or evidence stops short.
Lineage: Jeanette Andrews’s Brown artist statement; Brown Arts Institute’s The Logic of Illogic; MIT CAST residency materials. This is HG Atelier’s teaching synthesis of those published ideas, not a quotation or claim of endorsement.
Reconstruct a defensible state from incomplete evidence while keeping observation, measurement, report, inference and unknown distinct.
Model the observer by separating performance, confidence, sensitivity, criterion, context, and repeatability—and see which quantity moves when conditions change.
OBSERVE supplies an evidence model. CALIBRATE supplies an observer model. SYNTHESIS asks for the strongest inference that remains consistent with both.
The record is local to this browser and bounded to these exercises and conditions. It does not establish a general personal trait.
No integrated OBSERVE progress reported yet.
No integrated CALIBRATE progress reported yet.
This layer reads the existing local record only. It does not alter stimulus timing, trial order, response policy, or scoring logic.
Build the evidence trace, test the observer model in a new context, then reconcile them in SYNTHESIS.