The Research Arm of Focor
The science behind the attention layer.
Focor Research is the open research program of Focor Inc. We study Focus Debt, Cognitive Collapse, neurodivergent cognition, and the future of work. Researchers and qualifying institutions get free access to Focor, the attention layer for your AI stack, and to anonymized data.
A research initiative by Focor Inc. For human attention.
The problem we study
Knowledge work is broken. Every AI tool companies adopt makes it worse.
We call it Context Collapse: the moment a knowledge worker's mind can no longer integrate the signal around them. The evidence is already loud.
interruptions per day
Microsoft, 2025
say AI increased their workload
Upwork, 2024
of GenAI investments show no ROI
MIT Media Lab, 2025
annual cost of disengagement
Gallup, 2024
of time spent communicating, not thinking
Microsoft, 2025
Why we run this program
Focor builds a product. Focor Research builds the evidence.
Focor is a for-profit company building the attention layer for your AI stack: a sovereign, private model of how you actually work, assembled from the context your tools throw away, that runs across your stack and separates signal from noise. Focor Research is its research arm.
Running the attention layer at scale means we see something no lab alone can: how attention behaves across real work, across real tools, across real weeks. We have a responsibility to study that openly, with the right researchers, and with people who understand neurodivergent cognition first-hand.
Technology should not replace what makes us human. It should protect it: the judgment, the relationships, the hard conversations, the focus.
Methodology · Specification v2.0
The construct framework.
A-D fully specified. E and F partial; full F-track scope reserved for v2.1, pending DPA expansion.
23 ready in v2.0. 6 reserved for v2.1, mostly under Tracks E and F.
Mark, Leroy, Kahneman, Wood & Rünger, Fogg, Maslach, Ericsson, Bandura, Christakis & Fowler, Burt.
Twenty-nine behavioral constructs across six research tracks. Each construct is specified to a single computable formula, anchored to a published baseline from the behavioral and network science literature, and tied to a directional hypothesis about a Focor feature. Studies are conducted across the full participant population, including neurodivergent cohorts. The specification is versioned: construct identifiers are stable across major versions, and any change in operationalization carries a 90-day notice.
Each construct is specified by six attributes
- Definition
- A natural-language statement of what the construct measures, written for a non-specialist reader.
- Operationalization
- The computable formula applied to participant event data, with all input fields named.
- Unit
- The unit of measurement: seconds, ratio, count, bits, dimensionless index.
- Literature anchor
- The published baseline against which results are compared, or a note that the measure is novel.
- Tested feature
- The Focor product feature whose effect the construct is designed to detect.
- Hypothesis
- The directional prediction, stated in falsifiable form, with the expected direction of effect.
Constructs missing any of the six attributes are excluded from the published specification. Composite constructs CSC, CDI, BRDI, and BPF carry an additional attribute: a defined source-construct formula.
Anatomy of a construct
- Definition
- Time from an interruption to the participant's return to sustained activity on the pre-interruption task.
- Operationalization
t_resume_deep_work − t_interrupt, wheret_resume_deep_workis the start of the next continuous ≥ 10-min activity block on the sametask_context_id.- Unit
- seconds
- Literature anchor
- Mark, Gudith & Klocke (2008). 1,395 s, about 23 minutes 15 seconds. The 23-Minute Mark baseline.
- Tests feature
- Drift Detection + Ambient Priming.
- Hypothesis
- Active Focor users below the literature baseline.
Focus & Attention
How much of the day is actually thinking, and how much is the cost of being interrupted from thinking?
Active Focor users show Refocus Time below the 23-minute Mark baseline, and Context-Switch Cost returns measurable minutes to the workday.
Cognitive Debt
What is the running mental balance of work owed, work owed to you, and decisions not yet made?
The Cognitive Debt Index drops over the active phase as Focor surfaces aging items earlier and closes the backlog.
Decision Quality
Are fast decisions made well, and are slow decisions worth waiting for?
Decision-Metabolism Alignment rises as users learn their own optimal timing per category; reversal rates fall on metabolism-aligned timing.
Habit Formation & Behavioral Drift
When does an intervention become a habit, and when does behavioral drift accumulate over weeks?
Active-phase users show higher Habit Stability and a measurable shift in the deep-vs-shallow work ratio. Drift signals carry directional information about subsequent self-report.
Identity & Professional Growth
How does what you spend your hours on slowly become who you are professionally?
Capability domains expand over multi-month windows, and the gap between self-concept and observed behavior narrows over time.
Network & Collective Intelligence
Which relationships actually feed the work, which are quietly going dormant, and what does a high-cohesion team look like in the data?
Strong ties remain stable; weak ties get more time once Focor surfaces them; high-brokerage users do not pay the standard focus penalty under active Focor.
Construct IDs are stable across versions. A construct's operationalization can change with a major-version bump and 90-day notice, but its ID is permanent. Composite IDs (CSC, CDI, BRDI, BPF) are derived from source IDs by a defined formula. The full readiness matrix, feature map, and v2.1 expansion plan are available to approved research partners.
Apply to the program
Two tracks. Both free. Application-based.
If your work fits any of the six tracks above, or you train and support people whose cognition the tracks are built to describe, we want to work with you.
Anonymized data from a live attention layer.
Focor instruments how people move between tasks, contexts, and tools across the working day. With informed consent and strict anonymization, we make this longitudinal dataset available to qualified academic researchers.
Free access to anonymized, longitudinal data, plus the Focor product for your research cohort.
- Longitudinal attention & task-switching data at scale
- Ethical review and informed-consent pipeline in place
- Secure research API with versioned data releases
- Optional co-authorship and open-methods collaboration
Free access to Focor for institutions serving neurodivergent learners.
Universities, research labs, specialized schools, and programs that study or educate neurodivergent learners can apply for free access to the Focor attention layer for their students, staff, and research cohorts.
Free Focor licenses for your cohort, shared research studies, and curriculum collaboration.
- Free seats on Focor for students, staff, and research cohorts
- Joint studies on neurodivergent cognition and tooling
- Curriculum collaborations and classroom-practice pilots
- Funding opportunities for qualifying graduate research
Non-negotiable, and worth naming.
Consent first, always. Participation in research is opt-in and revocable.
Anonymization is not optional. No researcher sees identifiable data.
Research serves people, not marketing. Findings are published openly.
Neurodivergence is not a deficit. We study it as a form of human variation.
Independence within the company. The program operates at arm's length from product decisions.
Methods are open. Reproducibility is a feature, not a marketing claim.
Apply
Tell us what you are working on.
One form for both tracks. Pick the path that fits, share a short abstract or program description, and we will reply within a week.
Prefer email? Write to research@focor.org or institutions@focor.org.