Research at Cognispace

Structured inquiry into
what gets expressed and released

Cognispace research supports the systems we ship: how speech is measured, how AI context is released, how conditions travel through derived software, and where each of those systems has to stop.

What we study

Our research program studies the path from input to observable output, which parts of that path can be measured directly, and how systems can govern the rest responsibly.

01

Provenance & Context Governance

How conditions on datasets, models, and code carry through every derivation, and how AI systems can receive only the context a task needs. The research behind Material Governance Condition and Elevated Context.

02

Speech & Signal Analysis

Research into acoustic, prosodic, linguistic, and timing-derived measurements of recorded speech, and into how those measurements can be reported without claiming to know what lies behind them.

03

Interpretive System Design

Building AI-assisted workflows for classifying and explaining expression signals against a human-centered ontology — not purely technical taxonomy.

04

Ethics & Interpretive Boundaries

Defining the limits and responsibilities of interpretive systems — uncertainty modeling, human oversight requirements, and safeguards against authority overreach in high-stakes contexts.

05

Enterprise Intelligence Systems

Turning measured findings into decision-support workflows for coaching, communication, and institutional settings, with clearly bounded interpretive scope.

Current research tracks

Governance Systems

Provenance & AI Context Release

How a condition placed on a dataset, model, or source artifact carries through forks, builds, and fine-tunes, and how an AI request can be narrowed to the least context a task needs.

Derivation Lineage Typed Authority Least-Context Release Fail-Closed Design
Signal Systems

Speech & Expression Analysis

Research into how observable speech patterns, acoustic, prosodic, linguistic, and temporal, can be measured consistently and traced back to the recording.

Pause Structure Prosodic Patterns Turn-Taking Dynamics Temporal Indicators
Responsible Research

Ethics & Interpretive Limits

Defining what interpretive systems can and cannot responsibly claim — human oversight requirements, uncertainty boundaries, and non-clinical positioning across deployment contexts.

Interpretive Humility Uncertainty Modeling Autonomy Preservation Ethical Use Boundaries
Applied Systems

Enterprise Intelligence Workflows

Turning measured, traceable findings into organizational decision-support tools for coaching, communication analysis, and institutional settings.

Communication Analysis Coaching Applications Interpretive Workflows Responsible Deployment

Papers & preprints

Foundational papers and pre-publication manuscripts are currently in active development. When ready, they will be distributed through recognized preprint and scholarly channels.

Coming soon

Papers currently in preparation will appear here when available. This page will be updated as work is ready for distribution.

Taking part in a Cognispace research survey

This section applies to Cognispace's professional-practice research surveys (workplace investigation, communication coaching, qualitative research, and call review and coaching). It adds survey-specific details to the Cognispace Privacy Policy, which continues to apply.

Taking part is voluntary. After the consent question, you can skip any question, and you can stop at any time. The surveys ask about professional practice. They are not a legal, employment or regulatory process, and your answers are not used to evaluate you or anyone you work with.

What we collect.

  • Your answers to the survey questions.
  • Optional contact details (name, email, and how you prefer to be contacted) if you choose to give them at the end.
  • A label in your survey link that records which outreach channel it came from and whether the response is a pilot or a live response.
  • Technical information that Tally records when you open or submit a form, such as the time of submission.

We do not ask for files or recordings. Please do not include confidential client, employee, case or customer information in any answer, including free-text boxes.

Tally. We collect responses with Tally (Tally BV, Belgium), which stores and processes them for us. Tally states on its GDPR page that form data is encrypted in transit and at rest and is stored in Europe. Tally's own terms and privacy notice govern what Tally does with the data it holds.

Who can see your answers. The people at Cognispace working on this research, and Tally as our survey provider. We copy responses from Tally into a private research store on a Cognispace computer. Contact details are kept separately from your answers and are not included in the files we analyse. We do not sell your information, and we do not publish individual answers together with your name or contact details.

AI tools. The software we use to collect, store and analyse survey responses does not send individual responses or contact details to external AI services. This describes how that software behaves. It does not describe any other tool that people working on the research may use. If we decide to use an external AI service on individual responses or contact details, we will first update this notice and, where required, ask for your consent.

How long we plan to keep it. These are planned practices that we carry out by hand. They are not automatic and not a guarantee of a deletion date.

  • Survey answers: we plan to delete them, or reduce them to a summary that cannot identify you, within 12 months after the survey closes.
  • Optional contact details: we plan to delete them when we no longer need them to arrange follow-up research, and in any case within 6 months after the survey closes.
  • Pilot and test submissions: we plan to delete them within 30 days after the pilot review.
  • Summaries and findings that contain no individual answers may be kept longer.

Limits on deletion. We delete responses in Tally and in our private research store ourselves, so timing can slip. Tally keeps deleted submissions in its trash for up to 90 days before they are permanently removed, so a deleted response can remain with Tally for that time. These plans cover the copies we manage directly (in Tally and in our private research store). We do not promise that every other copy, such as a backup made by a device or service we use, is removed on the same schedule.

Privacy questions and removal requests. Email privacy@cognispacehq.com and include the date you took part, and the contact details you entered in the survey if you gave any. If you gave no contact details, we may still be able to find your response from the date and time you took part, but we cannot promise to, and we will tell you if we cannot. We aim to reply within 30 days. Removal from Tally follows the trash period above.

Reporting. We report results in aggregate or de-identified form. We do not quote free-text answers unless they cannot identify you or anyone else.

Limits. We do not promise anonymity: your answers are linked to a respondent code in our files, and your contact details, if you gave them, link back to you. Email is not a secure channel, so please do not send confidential information to us by email.

Research questions. Email research@cognispacehq.com.

Last updated: 2026-11-17.

How we approach the work

Cognispace research is grounded in interpretive humility. We operate as an interpretive bridge, not a clinical authority. Every output defines its uncertainty boundaries, scope limitations, and the human oversight it requires.

Interpretive Humility

Our analytical models describe patterns and configurations, not ground truth about internal states. We never claim predictive or diagnostic authority beyond the interpretive layer.

Clearly Defined Scope

Every research output specifies what it is and is not. Our methodology is not a clinical model, not a personality typing system, and not a neurological theory. Scope boundaries are non-negotiable.

Human-Centered Review

Interpretive outputs require human review before informing decisions. No system we build may present its outputs as deterministic conclusions.

Respect for Autonomy

Research applications must preserve individual agency. Interpretive analysis is offered as a structured lens — never used to constrain, rank, or judge individuals.