RINHUMAI logo
Research Methods

How do you study what AI does to people — rigorously?

RINHUMAI combines three methodological approaches: meta-analysis to synthesize what existing research shows about attention, agency, and cognitive change under AI mediation — the Meditation Workstation as an observation platform for studying human attentional states at the baseline — and original empirical experiments designed to test the mechanisms directly. Together they address the question from different angles.

RINHUMAI

Methodological foundations

Studying AI's effects on human cognition requires looking outward at what the accumulated literature shows, looking inward at what human attentional states actually are without external scaffolding, and looking forward through original experiments that test the mechanisms directly. Each method addresses a different part of the same question.

Quantitative synthesis

Meta-Analysis

RINHUMAI applies meta-analysis specifically to the growing body of quantitative research on how AI exposure affects attention, memory, judgment, and agency. Rather than relying on any single study, this approach synthesizes findings across independent datasets — making it possible to detect consistent patterns, resolve contradictions, and identify where evidence is strong and where it is thin.

By pooling findings systematically, meta-analysis can resolve apparent inconsistencies between individual studies, identify moderating variables, and clarify the direction and magnitude of effects that would otherwise remain ambiguous. It is particularly well-suited to a research field where individual studies vary widely in sample size, design, and population — as is currently the case for research on cognitive effects of AI mediation.

Quantitative synthesis Effect size pooling Systematic review
Combines findings from multiple independent studies addressing the same research question
Computes combined effect sizes and variance measures across studies
Improves statistical power beyond what individual studies can achieve
Helps clarify inconsistencies and contradictions between study results
Supports future research directions by identifying gaps and moderating factors
Research instrument

Meditation Workstation — "Event Horizon"

To study what AI mediation changes in human cognition, you need a rigorous way to observe what those cognitive states look like in their most intact, self-directed form — before and outside of AI scaffolding. The Meditation Workstation provides exactly that: a closed-loop research environment for observing and stabilizing deep, unmediated attentional states under controlled conditions.

The platform operates through multimodal sensory stimulation combined with real-time physiological sensing. Machine-learning-driven predictive feedback and biofeedback loops allow the system to adapt continuously to the user's measured state. The design philosophy treats these states as something that can be supported and studied through engineered, observable means — producing replicable data on forms of awareness that have historically been difficult to study systematically.

The system is built around practical engineering and safety constraints, uses available hardware and software components, and is designed to comply with established medical device safety standards. The broader design includes EEG and biofeedback sensing as part of its physiological measurement infrastructure — enabling direct observation of attentional and cognitive states during research sessions. It represents a replicable, technology-based platform amenable to further scientific study and industrial application.

Patent reference: USPTO application no. 19/369,243. Co-inventors: Prof. dr. Dirk K.F. Meijer and Richard Dobson.

Closed-loop system Biofeedback Physiological sensing Multimodal stimulation Machine learning
Multimodal sensory stimulation to guide and stabilize meditative states
Real-time physiological sensing of the user's measured state
Machine-learning-driven predictive feedback adapted to physiological data
Biofeedback loops for continuous closed-loop adaptation
Engineered by observable, replicable means — not asserted as mystical
Built within practical engineering and established safety constraints
Designed as a replicable, technology-based research platform
Empirical method

Experimental Research

Meta-analysis and the Meditation Workstation address what is already known and what can be observed under controlled conditions. A third methodological direction is under development: original empirical experiments designed to study how specific features of AI-mediated environments influence cognitive and deliberative outcomes.

This experimental paradigm allows RINHUMAI to go beyond synthesizing existing research — to actively manipulate structural features of AI-mediated settings and measure their effects on how people reason, deliberate, and decide. It operationalizes the core research questions rather than describing them from a distance.

Controlled experiments Structural manipulation Collective cognition Deliberation outcomes
Systematic manipulation of AI-mediated environment features
Direct measurement of cognitive and deliberative outcomes
Designed to produce replicable, externally reviewable results
Currently in development — first studies forthcoming
Collaborations

Researching cognitive effects of AI, or looking for rigorous ways to study human attentional states?

We welcome contact from researchers and institutions interested in meta-analytic approaches to AI mediation research, or in the Meditation Workstation as a research instrument. Whether your interest is methodological, empirical, or applied, we are open to conversation.

Joint meta-analyses across existing datasets on attention, agency, and AI-mediated behavior.
Research use of the Meditation Workstation platform in clinical or laboratory settings.
Methodological review and alignment with institutional research standards.
Contact us → See our publications →
We typically reply within a few days.