Dr Alexander D. Shaw

Neurocomputational Consulting

Dr Alexander Shaw – Neurocomputational Consulting

Inference, Decision Systems & Mechanistic AI for Complex Problems

I help organisations build models that explain, predict, and act under uncertainty: from neural data to autonomous systems and strategic decision-making.

View Services Book a Consultation
Dr Alexander D. Shaw

About

Alex is a neuroscientist and AI researcher at the University of Exeter, where he leads the Computational Psychiatry & Neuropharmacological Systems (CPNS) Lab and serves as Director of Business Engagement & Innovation for Psychology.

His work focuses on building models that explain complex data, recover hidden structure, and enable adaptive decision-making. This combines dynamical systems modelling with Variational Bayes/Laplace inference, and extends across neuroscience, AI systems, and real-world optimisation problems.

He develops inference methods and generative modelling frameworks used internationally, including Dynamic Causal Modelling, thermodynamic variational inference, and active inference. These approaches are designed to handle nonlinear, uncertain, and high-dimensional systems where standard methods often break down.

Alex works with neurotechnology companies, AI research teams, and industry partners to solve hard modelling and decision problems — from extracting latent structure in complex data to designing adaptive, uncertainty-aware systems.

Much of this work is demonstrated through interactive simulations and agents, spanning control systems, planning under uncertainty, and embodied AI in physical environments.

Explore live demos →

Selected Demonstrations

Working examples of inference, decision-making, and adaptive systems applied to control, planning, and real-world environments.

These are not toy models; they illustrate how principled generative modelling and probabilistic inference can be deployed in real-world settings.

I’ve been applying similar approaches in industry contexts - happy to discuss if relevant.

Real-time control

Thermodynamic Active Inference Pong Agent

Real-time inference and action selection using variational free energy minimisation.

View demo
Decision systems

Hierarchical Active Inference Gridworld

Planning and navigation under uncertainty using belief-driven policies.

View demo
Multi-objective AI

Polyphonic Active Inference

Negotiation between competing objectives (safety, goals, uncertainty, energy).

View demo
Embodied AI

Active Inference Robot

Embodied agent operating in a 3D physics environment with perception-action loops.

View demo

Core Services

Four service lines focused on solving complex modelling, inference, and decision problems: from scientific data to AI systems and real-world optimisation.

Enquire

Strategic & Technical Advisory

Rigorous modelling and scientific insight applied to product, research, and investment decisions.

  • Technical due diligence (AI, neuro, complex systems)
  • R&D strategy & model-driven product design
  • Mechanistic insight for high-uncertainty domains
  • Experimental design & evaluation frameworks

Decision Systems & AI

Designing systems that act under uncertainty using principled inference and planning frameworks.

  • Planning & control beyond reinforcement learning
  • Generative model architectures for agents
  • Simulation-based optimisation & policy design
  • Adaptive systems with uncertainty-aware behaviour

Inference & Modelling Systems

Fitting complex models to data to recover hidden structure, dynamics, and mechanisms.

  • Nonlinear dynamical systems modelling
  • Variational inference (Laplace, thermodynamic VI)
  • Multimodal & non-Gaussian posterior estimation
  • Model comparison, optimisation & validation

Applied Analytics & Measurement

Turning complex data into actionable signals for research, products, and decision-making.

  • Time-series modelling & feature extraction
  • Behavioural and physiological data analysis
  • Biomarker discovery & latent variable modelling
  • Real-world sensing & longitudinal data pipelines

Example Projects

Representative projects illustrating how principled modelling and inference unlock value across AI systems, data analysis, and real-world decision-making.

AI Systems

Adaptive decision systems under uncertainty

Designing agents that plan and act using probabilistic inference rather than fixed reward functions.

Inference

Fitting nonlinear dynamical systems

Using advanced variational inference (e.g. thermodynamic VI) to estimate parameters in complex, unstable models.

Mechanistic Modelling

Interpretable models vs black-box ML

Recovering underlying mechanisms and latent structure from data, rather than relying on opaque predictive models.

Real-world Data

Biomarkers & latent structure in complex signals

Extracting meaningful structure from noisy time-series data (e.g. EEG, behavioural, or sensor data).

Engagement & Pricing

Clear tiers for rapid advisory, scoped projects, or ongoing partnership.

Advisory

High-leverage strategy or technical consults.

£500–£1200 per session

  • 1–2 hr video call
  • Written summary report
  • Reading list / next steps
Book a call

Project-Based

Bespoke analysis or modelling with deliverables.

£2,000–£10,000 per project

  • Defined scope, milestones & success criteria
  • Data audit & preprocessing plan
  • Reproducible analyses, code & figures (version-controlled)
  • VB/Laplace model fitting & comparison
  • Mechanistic interpretation & briefing
  • Technical report & documentation
Request a proposal

Retainer

Ongoing collaboration and technical advisory.

£1,500–£3,000 per month

  • Guaranteed availability
  • Pipeline design & oversight
  • Monthly reporting
Start a partnership

Day-rate option: £800–£1,200/day depending on scope and IP terms.

Contact

If you'd like to discuss collaboration, consulting, speaking, research partnerships, or applied AI / modelling work, send a message below.