Introduction to the Catalog¶
A task is an “instructed activity”: a set of goals to be achieved or a set of actions to be performed, or a combination of these, under conditions the experimenter controls, so that what participants do can be attributed to the mental processes the task was designed to engage.
Throughout these pages, the Catalog is the HED Task Catalog, the Atlas is the Cognitive Atlas, and CogPO is the Cognitive Paradigm Ontology.
What the Catalog is for¶
Experiments in cognitive and behavioral neuroscience are built from a fairly small
repertoire of standard tasks, but the datasets they produce rarely say which task
represented the “instructed behavior” of an experiment in a standardized way a
machine can act on. A data repository knows what files a dataset
contains but usually not what instructed behavior the participants did beyond
a free-text label such as task-flanker or task-rest. The same paradigm goes
by different names across laboratories, and related paradigms that engage the
same cognitive processes are not connected at all.
The Catalog is meant to supply the missing layer: a controlled vocabulary made of a list of tasks, a list of the cognitive processes those tasks engage, and the links between them.
Tags for datasets. A dataset tagged with a task identifier from the Catalog, and with the process identifiers that task engages, can be found by a repository search for the task, for any of its aliases, or for a process, whatever the dataset called it locally.
Commonalities across datasets. Two datasets tagged with the same task can be compared directly. Two datasets tagged with different tasks that share a process can be grouped for a question about that cognitive process.
Context for event annotation. A task tag tells a reader of a HED-annotated events file what the trial structure was designed to do, which the event-level annotations alone do not.
The Catalog is being developed as part of HED (Hierarchical Event Descriptors), the standard for annotating what happened during an experiment. HED describes events; the Catalog describes the paradigm the events belong to. How task and process identifiers will be carried in HED annotations and in dataset metadata is not yet decided; the identifiers here are the Catalog’s own and are not HED schema terms.
What the Catalog contains¶
The Catalog consists of two interlinked lists, tasks and cognitive processes, with facets as a proposed third element. Every task states which processes it engages, and every process lists the tasks that engage it.
What |
Count |
Where |
|---|---|---|
Tasks |
103 |
|
Pseudo tasks (rest, fixation, questionnaire and feedback blocks) |
4 |
|
Task families the tasks are filed under |
19 |
|
Named task variations |
776 |
on each task page |
Cognitive processes |
172 |
|
Process categories |
19 |
|
Task-to-process links |
486 |
|
Processes engaged by at least one task |
152 |
Resources at 3 granularities¶
Three resources on this site describe what happens in an experiment, at three levels of detail. The Catalog is the coarsest: a task label names the goal of a block of an experiment, what the participant was asked to do, without saying how any trial unfolded. CogPO is intermediate: it labels the task and the steps within it, each condition with its stimuli, responses and instructions. The Catalog’s counterparts at that level are the proposed facets, single attributes of a task such as stimulus modality or response kind drawn from closed vocabularies that CogPO (or the HED equivalent) supplies, and the pseudo tasks, which name the block-level steps such as rest, fixation and feedback. HED annotates what actually happened, event by event, with a time for each.
Resource |
Labels |
Example, for a Stroop experiment |
|---|---|---|
The Catalog |
a task, the goal in eliciting cognitive processes |
Stroop Color-Word Task |
CogPO |
a condition: stimulus, response, instruction |
colored color words, key press, “name the ink color” |
HED |
an event, with its time |
the onset of the word RED printed in blue, 2.35 s into the run |
HED vs CogPO gives the detail of the two lower levels.
Tasks¶
A task in the Catalog is a structured experimental paradigm that produces a sequence of discrete, time-stamped events, has a specific reproducible procedure, is widely used, and engages identifiable cognitive processes: the Stroop Color-Word Task, the N-Back Task, the Iowa Gambling Task. Particular questionnaires, clinical screening instruments, generic labels such as “memory task”, and one-off designs are excluded. The full rules are in the Task selection criteria.
The Catalog also has 4 pseudo tasks, which are not tasks in the sense above: the Rest Task, the Fixate Task, the Take Questionnaire Task and the Receive Feedback Task. A pseudo task is a block that sets up or holds a brain state, collects a self-report or delivers feedback, rather than eliciting a process through trials. Such blocks appear in most experiments and datasets need to label them. The Catalog includes them for that reason, marks them on their pages, and has not yet linked them to processes. The Pseudo tasks section of the task criteria explains what qualifies as a pseudo task. The pseudo tasks are closely related to CogPO Paradigms and to aspects of tasks that we currently are calling facets. The relationship between Catalog tasks, Catalog task facets, and CogPO Paradigms has not been finalized.
The Catalog currently has 103 tasks. Each task page carries:
A canonical name and its aliases, so that a search for “CWIT” or “Stroop” lands on the Stroop Color-Word Task.
A short definition and a longer description.
An inclusion test in three lists: the steps of the procedure the participant follows, the manipulations the experimenter varies, and the measurements recorded. An experiment is an instance of the task when its procedure matches, it manipulates at least one of the listed variables, and it records at least one of the listed measures. The procedure is decisive: two experiments with the same procedure are the same task even if they target different constructs.
A list of variations, each with a description and a justification. A variation is a named version that changes what the participant experiences or does, so that the event structure of the recorded data differs. Changing the recording equipment, the analysis, the population or the stimulus set does not make a variation; changing the response modality, the stimulus-response mapping or the required actions does. There are 776 variations across the Catalog.
The cognitive processes the task engages, linked to their definitions;
key references that established the paradigm and recent references that give a modern entry point.
Where one exists, the corresponding Cognitive Atlas entry.
Task families¶
Tasks are grouped under 19 task families for ease of search by what the participant does: conflict and interference tasks, response inhibition tasks, recall and recognition tests, economic games, and so on. The Stroop Color-Word Task is in the family Conflict and interference tasks. A family is a browsing aid, not a claim about mechanism; the processes a task engages are the place to look for that. Every task is filed under one primary family and may be cross-listed under others; the assignment is a curation decision that is expected to change, and the task family assignment rules say how it is made. The task index lists the families with their scope.
Cognitive processes¶
A cognitive process is a mental operation hypothesized to occur during a trial, such as response inhibition, working memory updating or reward anticipation: something with an identifiable onset, an eliciting condition and an in-principle measurable signature. A candidate qualifies when it is plausible to say when in a trial it happens, what elicits it, and how it is measured. States and traits, specific emotions, stimulus categories, individual-difference constructs, task parameters, analysis methods and umbrella terms such as “attention” or “memory” are excluded; the umbrella role is carried by the categories instead. The full rules are in the process selection criteria.
The Catalog has 172 processes in 19 categories. A category groups processes by research tradition for browsing and carries a scope statement and a note of what is out of scope. A process is filed under one primary category and may be cross-listed under others. Each process page entry has a definition, aliases where the literature uses more than one name, fundamental and recent references, and the list of tasks that engage it. Some processes are engaged by no task in the current Catalog; they are kept because they are real and the Catalog may grow a task for them.
Task-process links¶
A task is linked to a process when the task’s inclusion test engages that process: the paradigm-defining papers say so, a standard condition contrast isolates it, or a standard measure indexes it. A link is not an ontology assertion and not a claim that the task engages nothing else; it records what the task is designed to probe. The task-process links page lists every link in both directions.
Facets¶
A facet is a single attribute of a task from a closed vocabulary, such as its stimulus modality (visual, auditory) or its response kind (button press, saccade), so that a repository can filter tasks by what the participant is shown, does and is told, not only by name. The idea and the candidate vocabularies come from CogPO, the Cognitive Paradigm Ontology, which describes every paradigm by exactly those three things, and each value has HED tags that express it, so that a facet would sit between a task name and the HED annotation of its events. The dimension mapping page holds the vocabularies; whether and how the Catalog adopts facets is not yet decided, and no task record carries one.
Identifiers¶
Every entity has a typed identifier, so an identifier met anywhere can be classified at sight:
Kind |
Form |
Example |
|---|---|---|
Task |
|
|
Task variation |
|
|
Process |
|
|
Process category |
|
|
Task family |
|
|
The hed_ prefix on process identifiers is the Catalog’s working prefix and is not a
claim of HED schema membership. Identifiers are provisional while the Catalog is being
curated; when they stabilise, renames will become aliases.
Where the Catalog came from¶
The starting corpus was the Cognitive Atlas, a community-built ontology of cognitive concepts and the paradigms that measure them. Its task list was narrowed to paradigms that produce event-structured data, its concepts were reshaped into processes that pass the selection test above, and a gap analysis added paradigms the Atlas never registered. Every task and process here is mapped back to the Atlas entry by entry; the Cognitive Atlas pages describe the Atlas, the relationship page says how the two resources differ, and the Atlas mapping method page says how each correspondence was decided.
A second source for the Catalog is CogPO, the Cognitive Paradigm Ontology, whose stimulus, response and instruction vocabularies are the candidate facets; see The Catalog vs CogPO. The first facets to be filled are the six CogPO-derived ones (stimulus modality, kind and role; response modality and kind; instructions), followed by the Catalog’s own (trial structure, feedback, timing and primary measure). CogPO and HED share a history: early HED terminology drew partly on CogPO, and HED’s plan to adopt CogPO’s paradigm list was dropped in its third generation for want of a standardized task nomenclature, which is the gap the Catalog now fills. See HED vs CogPO.
Where to go next¶
103 tasks in 19 task families, each with its inclusion test, variations, processes and references.
172 processes in 19 categories, each with a definition, references and the tasks that engage it.
The selection criteria, naming rules, inclusion test and the rules that decide what counts as a variation.
What the Cognitive Atlas and CogPO contain, how the Catalog maps onto each entry by entry, and what each adds to the other.
Planning to tag a dataset or to read a task page closely? See How to use the Catalog.