Temporal Coincidence Detector¶
Detector — instantiates Object-Centered Feature Binding
Tests whether feature onsets fall inside the same time window more often than chance would allow, turning simultaneity into a scored — not assumed — binding cue.
A Temporal Coincidence Detector does one narrow thing well: it decides whether two or more features arrived together in time to a degree that would be unlikely by chance. It is not a binder — it produces a single cue, the timing cue, that a fuller binding decision can then weigh alongside space, provenance, and prior. Its defining discipline is the chance baseline: raw simultaneity is nearly worthless, because in any busy stream some features will land near each other for no reason at all, so the detector never reports "these coincided" without also asking "how often would they coincide if unrelated?" and holding the answer to a significance bar. Everything else it does — sliding a window, counting onsets, phase-aligning — is in service of that one comparison against chance.
Example¶
Two robotic sky-survey telescopes on different continents are each firing off transient alerts: brief brightenings that might be a supernova, an asteroid glint, a satellite streak, or detector noise. On a given night each observatory reports hundreds of these. The scientific prize is a source that both saw — a genuine astrophysical transient rather than a local artifact. A Temporal Coincidence Detector takes the two alert streams and, within a coincidence window of a few seconds, looks for pairs whose onset times align.
The naïve version would flag every near-simultaneous pair and drown in false positives, because with hundreds of alerts per night some will always land seconds apart by accident. So the detector first estimates the accidental coincidence rate — how many aligned pairs you would expect if the two streams were completely independent, given their rates and the window width — and only promotes a pair when the observed coincidence sits well above that background. One pair clears the bar by a wide margin; it is passed downstream as a strong timing cue on a candidate transient. A dozen others fall inside the window but are consistent with chance and are dropped. The detector never claims the surviving pair is one object — only that their timing is real evidence that they might be.
How it works¶
The mechanism's substance is entirely in the window and the baseline. It defines a co-occurrence window — a span in time (and, for phase-locked signals, a tolerance in phase) within which two onsets count as coincident. It then slides that window across the incoming streams and counts aligned events. The step that separates it from naïve simultaneity is the accidental-rate model: from each stream's event rate and the window width it computes how many coincidences pure chance would produce, and expresses each observed coincidence as a margin above that expectation. A significance threshold converts that margin into a call — promote the timing cue, or discard it. Widen the window and you catch loosely synchronized events but raise the accidental rate; narrow it and you demand tighter timing at the risk of missing real but jittered pairs.
Tuning parameters¶
- Window width — the span within which onsets count as coincident. Wider windows catch delayed or jittered features but sharply raise the accidental-coincidence rate; narrower windows are stricter and miss loosely aligned pairs.
- Phase tolerance — for periodic signals, how far out of phase two features may be and still count. Tightening it isolates truly locked sources; loosening it tolerates drift.
- Significance threshold — how far above the chance baseline a coincidence must sit to be reported. Raising it suppresses false alarms at the cost of missing weak real coincidences.
- Baseline model — whether the accidental rate is estimated from a fixed prior, a rolling recent window, or a shuffled-stream null. A rolling estimate adapts to bursty periods but lags sudden rate changes.
When it helps, and when it misleads¶
It shines wherever timing is a strong and cheap cue and where the danger is treating "happened together" as "belong together." By forcing every coincidence through a chance baseline, it converts a seductive but unreliable signal into a scored, thresholded one, and it fails loudly rather than silently — a coincidence that does not clear the bar is simply not reported.
Its failure mode is the mirror of its strength: it knows only timing. A detector tuned to a tight window and a strict baseline will still bind two things that coincided for a hidden common cause (a shared clock glitch, a periodic external trigger) because they are genuinely time-correlated, just not co-object. And a burst of activity inflates the accidental rate everywhere at once, so during storms it can suppress real coincidences it would catch on a quiet night — the false-alarm-rate problem that any coincidence search must budget for.[n1] The guarding discipline is to treat the timing cue as one vote, never the verdict: hand it to a multi-cue binder that can check whether space, provenance, and continuity agree before anything is bound.
How it implements the components¶
co_occurrence_context_window— the window (and phase tolerance) is the detector's core object: it defines exactly when and how closely two features must land to count as coincident.binding_confidence_threshold— the significance bar above the accidental-coincidence baseline is the threshold that decides whether a timing coincidence is reported as evidence at all.
It produces only a timing cue; it does not assemble the multi-cue binding_evidence_vector that weighs timing against space and provenance — that is Feature Binding Matrix and Multimodal Fusion Tracker — and it does not intervene to create new evidence. Its nearest confusable, Active Probe Sequence, also gates on a binding confidence threshold, but the probe actively runs an active_disambiguation_probe to generate signal, whereas this detector only passively scores the timing already present.
Related¶
- Instantiates: Object-Centered Feature Binding — supplies the temporal-coincidence cue the archetype's evidence vector draws on.
- Sibling mechanisms: Feature Binding Matrix · Multimodal Fusion Tracker · Active Probe Sequence · Object File Tracker
Editorial Notes¶
Form Classification¶
Form family: Monitoring, Sensing & Alerting
Rationale: Temporal Coincidence Detector operates as ongoing observation, sensing, or alerting that detects and surfaces state without itself executing the response because it tests whether feature onsets fall inside the same time window more often than chance would allow, turning simultaneity into a scored — not assumed — binding cue.
Independent corroboration: The frozen evidence defines Temporal Coincidence Detector as 'Tests whether feature onsets fall inside the same time window more often than chance would allow, turning simultaneity into a scored — not assumed — binding cue', so its operative form is Monitoring, Sensing & Alerting.
Nearest alternative: Analysis, Modeling & Optimization — Temporal Coincidence Detector includes features of an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution, but its defining operation is ongoing observation, sensing, or alerting that detects and surfaces state without itself executing the response.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Neuroscience
Origin pattern: Single lineage
Present-day reach: Specialized
Rationale: The defining operation is: Tests whether feature onsets fall inside the same time window more often than chance would allow, turning simultaneity into a scored — not assumed — binding cue. In the neuroscience lineage, that operation is specifically evidenced by authoritative or primary work that grounds neural coincidence detection and binding in precisely bounded relative onset and spike-timing windows. This makes neuroscience the best historical origin, while the retained alternates document contributing methods and later applications rather than being mistaken for coequal origins.
Related originating lineages:
- Cognitive Science — Cognitive-science research on representation, learning, and recall supplies a parallel or contributing lineage for the mechanism's defining operation: tests whether feature onsets fall inside the same time window more often than chance would allow, turning simultaneity into a scored — not assumed — binding cue.
- Organizational & Management Science — organizational_management supplies a historically relevant parallel or contributing practice for the defining operation—Tests whether feature onsets fall inside the same time window more often than chance would allow, turning simultaneity into a scored — not assumed — binding cue—but the evidence does not make it the best primary lineage.
- Psychology — Experimental, clinical, and behavioral psychology supplies a parallel or contributing lineage for the mechanism's defining operation: tests whether feature onsets fall inside the same time window more often than chance would allow, turning simultaneity into a scored — not assumed — binding cue.
- Statistics & Experimental Design — Statistics, experimental design, and measurement theory supplies a parallel or contributing lineage for the mechanism's defining operation: tests whether feature onsets fall inside the same time window more often than chance would allow, turning simultaneity into a scored — not assumed — binding cue.
- Systems Thinking & Cybernetics — Systems science's feedback, stock-flow, boundary, and regulation tradition provides a formative adjacent lineage for the same temporal coincidence detector operation.
Review resolution: The blind reviewers disagree on primary lineage (organizational_management versus neuroscience), so I adjudicated the mechanism rather than inheriting either label. The defining operation is: Tests whether feature onsets fall inside the same time window more often than chance would allow, turning simultaneity into a scored — not assumed — binding cue. In the neuroscience lineage, that operation is specifically evidenced by authoritative or primary work that grounds neural coincidence detection and binding in precisely bounded relative onset and spike-timing windows. This makes neuroscience the best historical origin, while the retained alternates document contributing methods and later applications rather than being mistaken for coequal origins. The cited Spike-timing-dependent plasticity and temporal patterning directly supports the mechanism-specific operation and its disciplinary lineage. I retain all independently explained historical alternates without a numeric cap. origin_mode=single_lineage records how the mechanism arose; domain_reach=specialized separately records how broadly it can now be applied.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
Notes¶
[n1] False-alarm rate — the frequency at which a detector reports a coincidence that is actually accidental. In coincidence searches it is set by the event rates and the window width, and it is the quantity a significance threshold is chosen to control; ignoring it is the classic way a coincidence search fools itself. ↩