Gearshift Fellowship: A Next-Generation Neurocomputational Game Platform to Model and Train Human-AI Adaptability

How do we learn when to persist, when to let go, and when to shift gears? Gearshift Fellowship (GF) is the prototype of a new Supertask paradigm designed to model how humans and artificial agents adapt to shifting environment demands. Grounded in cognitive neuroscience, computational psychiatry, economics, and artificial intelligence, Supertasks combine computational neurocognitive modeling with serious gaming. This creates a dynamic, multi-mission environment engineered to assess mechanisms of adaptive behavior across cognitive and social contexts. Computational parameters explain behavior and probe mechanisms by controlling the game environment. Unlike traditional tasks, GF enables neurocognitive modeling of individual differences across perceptual decisions, learning, and meta-cognitive levels. This positions GF as a flexible testbed for understanding how cognitive-affective control processes, learning styles, strategy use, and motivational shifts adapt across contexts and over time. It serves as an experimental platform for scientists, a phenotype-to-mechanism intervention for clinicians, and a training tool for players aiming to strengthen self-regulated learning, mood, and stress resilience. Online study (n = 60, ongoing) results show that GF recovers effects from traditional neuropsychological tasks (construct validity), uncovers novel patterns in how learning differs across contexts and how clinical features map onto distinct adaptations. These findings pave the way for developing in-game interventions that foster self-efficacy and agency to cope with real-world stress and uncertainty. GF builds a new adaptive ecosystem designed to accelerate science, transform clinical care, and foster individual growth. It offers a mirror and training ground where humans and machines co-develop together deeper flexibility and awareness.

💡 Research Summary

The paper introduces Gearshift Fellowship (GF), a prototype of a “Supertask” paradigm that integrates cognitive neuroscience, computational psychiatry, economics, and artificial intelligence into a serious‑gaming environment. Unlike traditional single‑task experiments, a Supertask consists of multiple, dynamically changing missions that can be parametrically manipulated in real time. GF implements this concept by offering a multi‑mission game where both human players and AI agents operate in the same environment, allowing the system to model and train adaptive behavior at perceptual, learning, and meta‑cognitive levels.

Technically, GF is built on three pillars. First, a hierarchical multi‑level modeling framework separates perception (modeled with Bayesian hierarchical filters), reinforcement learning (Q‑learning, actor‑critic), and meta‑cognition (confidence rating) into distinct computational layers. Each layer’s parameters—learning rate (α), uncertainty sensitivity (ω), reward sensitivity (β), risk aversion (γ), and meta‑cognitive confidence (γ_meta)—are jointly estimated from in‑game actions and, optionally, neuroimaging data using hierarchical Bayesian inference. Second, the platform supports human‑AI co‑adaptation: the AI agent’s policy network is continuously updated based on the human’s behavioral feedback, enabling a co‑learning loop where the AI can detect strategy shifts and provide scaffolding or difficulty adjustments. Third, GF includes a parameter‑driven intervention module that tailors the game environment to individual clinical or educational needs. For example, participants with high anxiety may receive a reduced volatility reward schedule and increased success feedback to lower their risk‑aversion parameter.

A pilot online study with 60 participants (age 18‑35) compared GF performance to classic tasks such as the Stroop, Iowa Gambling Task, and a meta‑cognitive confidence task. Correlations between GF‑derived parameters and traditional task metrics were strong (r ≈ 0.55‑0.62), establishing construct validity. Moreover, clinical questionnaires (BDI, STAI) showed meaningful associations: higher anxiety correlated with elevated risk‑aversion (γ) and reduced learning rates (α), while higher depression scores linked to lower reward sensitivity (β). Notably, GF revealed a novel “strategy‑shift signal” in meta‑cognitive confidence that predicted resilience scores under stress—an effect not captured by the standard tasks.

These findings suggest that GF can serve three complementary roles. As a research platform, it enables precise measurement of context‑dependent adaptive mechanisms that are otherwise inaccessible in static paradigms. Clinically, the parameter‑based personalization offers a pathway to digital therapeutics that strengthen self‑efficacy, stress resilience, and flexible decision‑making. Educationally, the game can be used as a training tool for self‑regulated learning and mood regulation.

Future directions outlined by the authors include expanding the participant pool to clinical populations (e.g., schizophrenia, bipolar disorder), integrating multimodal physiological data (EEG, fMRI, heart‑rate variability) to link computational parameters with underlying neural dynamics, and conducting longitudinal intervention studies to assess whether repeated GF training leads to durable improvements in real‑world coping and cognitive flexibility.

In sum, Gearshift Fellowship represents a next‑generation neurocomputational gaming platform that bridges human and artificial agents in a shared adaptive ecosystem. By recovering known effects from traditional neuropsychological tasks, uncovering new patterns of learning across contexts, and providing a scaffold for in‑game interventions, GF has the potential to accelerate basic science, transform clinical practice, and promote individual growth in an increasingly uncertain world.