What is CogPO?

CogPO, the Cognitive Paradigm Ontology, is a small formal ontology of behavioral experimental paradigms, built around 2010 to give the BrainMap database a controlled vocabulary for the experiments it indexes. Its design is described in Turner JA and Laird AR, “The Cognitive Paradigm Ontology: Design and Application”, Neuroinformatics 10:57-66 (2012), DOI 10.1007/s12021-011-9126-x. The work was funded by NIMH grant R01MH084812.

CogPO’s distinctive idea is that a paradigm is not a name but a combination of observable components. Every experimental condition is decomposed along three dimensions: the stimulus presented (its modality, its kind and its role), the response required (the body part and the kind of overt act) and the instructions given to the participant. A paradigm class such as Stroop Task Paradigm is then a named pattern of such conditions. The paradigm layer came from the BrainMap taxonomy; the dimension vocabularies were new.

This page provides an overview of CogPO and its current state. The page also describes its influence on and relationship to HED (Hierarchical Event Descriptors). The Catalog vs CogPO compares CogPO and the Catalog. HED vs CogPO

Two sources

CogPO exists in two forms, and they do not quite agree.

• The OWL release, version 1.0 of November 2010, at http://www.cogpo.org/ontologies/CogPOver1.owl. This is the ontology proper: 199 classes with definitions, identifiers and provenance annotations, written in RDF/XML on the BFO upper ontology with IAO annotation properties. It declares an import of a companion file, CogPOexternalVer1.owl, but the server returns the main file’s bytes for that URL too, so there is one file. A second import, the Information Artifact Ontology, lives at the OBO Foundry.

• The wiki at www.wiki.cogpo.org, a MediaWiki with one page per term, opened for community comment when the OWL was released. Its curation ran on into 2011 and it holds a few terms the OWL never received. Its TLS certificate is invalid, so it is reachable over plain http only.

The figures below come from an archive of both taken on 2026-09-22. Nothing in either source is dated after 2011, and the “upcoming releases” the 2012 paper speaks of did not appear: CogPO is complete as far as it goes, and unmaintained.

OWL release summary

Layer	Classes
Behavioral Experimental Paradigm and its subclasses	84 (83 paradigm classes)
Stimulus Modality	6 top-level, 10 with the four tactile subtypes
Explicit Stimulus	41
Stimulus Role	0 (the class exists; its four values are on the wiki only)
Response Modality	8 body parts, as unlabeled FMA classes
Overt Response	17
Instructions	17
Condition, Implicit Stimulus, Covert Response, Response Role	classes with no members
Upper-ontology and imported classes	12

Of the 199 classes, 190 carry a label and 181 a definition. The unlabeled nine are the eight body parts, which are referenced by their Foundational Model of Anatomy identifiers alone, and owl:Thing. 149 classes have an identifier of the form COGPO_000nn; 34 CogPO terms have a name for an IRI instead (Button Press, Naming_(Overt)_Paradigm), a sign that the numbering was never finished.

Ten object properties connect the layers: a condition has_stimulus, has_instructions and has_response_type; a stimulus has_stimulus_modality; a response has_response_modality. Every explicit stimulus declares its modality through a restriction, and six of them (words, letters, digits, syllables, pseudowords, Braille dots) declare two. Three named individuals illustrate the design with one FBIRN auditory oddball task and its two conditions.

The paradigm layer

The 83 paradigm classes are direct children of Behavioral Experimental Paradigm, except that Oddball Discrimination has Auditory Oddball and Visual Oddball beneath it. The wiki adds three more (Delayed Match To Sample, Sleep, Tower of London) and lacks the two Oddball subclasses, so it lists 84.

Almost every definition has the same form: “A Behavioral Experimental Paradigm in which, in at least one of the conditions, subjects …”. The “at least one of the conditions” clause is deliberate: CogPO was built to code published experiments, which routinely combine paradigms, and a class is meant to be assigned whenever any condition fits. The definitions also say which classes to co-code (a passive face-viewing study is coded as Face Monitor or Discrimination but not also as Passive Viewing).

The list reflects its origin. Alongside Stroop, Flanker, Go/No-Go, n-back, Sternberg, Simon, Posner, Task Switching, Wisconsin Card Sorting and Mental Rotation stand Acupuncture, Breath-Holding, Chewing or Swallowing, Eating or Drinking, Micturition, Transcranial Magnetic Stimulation, Flashing Checkerboard, Whistling and Video Games: anything BrainMap needed to describe an imaging experiment, whether or not it is a task in any cognitive sense.

The dimensions

The dimension vocabularies are CogPO’s lasting contribution, and they are small enough to list. Stimulus Modality has six values (visual, auditory, tactile, olfactory, gustatory, interoceptive), with electrical, pain, pressure and thermal as subtypes of tactile. Explicit Stimulus has 41 values from words, faces and tones to random dots, fixation point, food, odor, heat, infusion and TMS. Overt Response has 17, from button press, saccade and speech to blink, grasp, swallow and micturate. Instructions has 17 verbs: attend, count, detect, discriminate, encode, fixate, generate, imagine, move, name, read, recall, repeat, rest, sing, smile, track. Response Modality is eight body parts (hand, foot, mouth, face, arm, leg, shoulder, pelvis). Stimulus Role has four values on the wiki (target, non-target, feedback, reward) and none in the OWL.

Every value has a definition, most of them dictionary definitions with the source cited, and 75 classes cite a BrainMap paper as the use that motivated the term. The dimension mapping page lists all of them.

Curation state

The OWL annotations record where the work stopped.

Curation status (OWL)	Classes
pending_final_vetting	142
none recorded	38
definition_incomplete	17
curation_complete	2

Every dated class is dated 2010. The wiki’s later pass marked most terms complete: of 184 term pages, 92 say “curation complete Aug 2011”, 77 “curation complete Dec 2010”, 5 “curation complete Sept 2011” and 9 remain “pending final vetting”. 25 wiki definitions differ from their OWL text, mostly small rewordings; one, the Stroop definition, has acquired a spam link.

Neither source carries a license statement.

What CogPO does not have

CogPO has no layer for cognitive processes: nothing says what a paradigm measures or which mental function it engages. It has no named variations, no inclusion tests and no references per paradigm beyond the occasional BrainMap citation on a definition. And the decomposition into conditions, which is its central design, is populated for exactly one example task. The dimension vocabularies exist; the paradigm-to-dimension assignments that would use them were never published.