nexus

nexus

Personality Test with EEG Data

Personality Test with EEG Data

project type

Interactive Installation with EEG Data

Year

2025

Role

Technical direction

technology

BCI (EEGNet), Unity, TensorFlow

nexus is an interactive installation that simulates a job interview of the future by measuring brain waves for an implicit personality test.

PROJECT DESCRIPTION

nexus

nexus is an interactive art installation and research experiment. Within a controlled simulated job‑interview, an EEG headset records participants’ brain activity while curated video segments are presented. A deep‑learning model (EEGNet CNN) infers the individual Big‑Five personality profile in real time, compares it with predefined team archetypes and returns a concise “Fit / No Fit” assessment.

CONCEPT

What if we know you?

Conventional job interviews depend on self‑assessment, a process research shows is skewed: women tend to understate and men often overstate their abilities. nexus replaces verbal answers with a biometric read‑out, asking whether unbiased neural data can surface genuine potential more equitably - and, conversely, make exaggeration visible. At the same time it confronts visitors with the unsettling possibility that employers might predict personality traits before any conscious self‑disclosure. The work therefore opens a debate on brain‑data ownership, security architectures, and the regulatory guard‑rails society will need as BCIs move from labs into workplaces.

Conventional job interviews depend on self‑assessment, a process research shows is skewed: women tend to understate and men often overstate their abilities. nexus replaces verbal answers with a biometric read‑out, asking whether unbiased neural data can surface genuine potential more equitably - and, conversely, make exaggeration visible. At the same time it confronts visitors with the unsettling possibility that employers might predict personality traits before any conscious self‑disclosure. The work therefore opens a debate on brain‑data ownership, security architectures, and the regulatory guard‑rails society will need as BCIs move from labs into workplaces.

PROJECT DESCRIPTION

Technical Architecture

Headset → Server: Emotiv EPOC+ streams 14‑channel EEG via WebSocket. Real‑time clean‑up: 1‑s sliding windows, artefact removal, normalisation. Inference: EEGNet‑v2 served in Docker / TensorFlow; ≈ 40 ms per window. Decision: Binary OCEAN vector matched to archetypes → “Fit / No Fit”. Interface: Unity WebGL dashboard renders spectrum & OCEAN meter. Latency: End‑to‑end ≈ 650 ms. Training: 1.3 h in‑house EEG (40 participants). Our tuned EEGNet reaches macro‑F1 0.81 — +7 pp versus the original EEGNet benchmark (macro‑F1 0.74; Lawhern et al., 2018).

Headset → Server: Emotiv EPOC+ streams 14‑channel EEG via WebSocket. Real‑time clean‑up: 1‑s sliding windows, artefact removal, normalisation. Inference: EEGNet‑v2 served in Docker / TensorFlow; ≈ 40 ms per window. Decision: Binary OCEAN vector matched to archetypes → “Fit / No Fit”. Interface: Unity WebGL dashboard renders spectrum & OCEAN meter. Latency: End‑to‑end ≈ 650 ms. Training: 1.3 h in‑house EEG (40 participants). Our tuned EEGNet reaches macro‑F1 0.81 — +7 pp versus the original EEGNet benchmark (macro‑F1 0.74; Lawhern et al., 2018).

nexus

Achievements

Awards

Exhibitions

Ars Electronica, Linz
MANIFEST:IO, Berlin

nexus

nexus

nexus

Credits

Prof. Andreas Ingerl

Prof. Andreas Ingerl

Felix Henke

Felix Henke

Emilia Gentis

Emilia Gentis

Chantal Pisarzowski

Chantal Pisarzowski

Academic Supervision
Academic Supervision
Academic Supervision
Cinematography & Post-Production
Cinematography & Post-Production
Cinematography & Post-Production
Art Direction & Exhibition Design
Art Direction & Exhibition Design
Art Direction & Exhibition Design
Technical Direction & Interface
Technical Direction & Interface
Technical Direction & Interface