Our goal is to build a Computational Theory of Mind, based on a new approach to Hyperdimensional Computing. It proposes an alternative to classical computationalist models and to neural-network-based connectionist models of the mind.
In this theory, the mind performs computations on sparse binary data. The fundamental data structures are Sparse Distributed Representations (SDRs) for perceptual information and Sparse Holographic Representations (SHRs) for symbols.
The centerpiece of this project is the Topological Associative Memory — a newly discovered associative memory for sparse data, serving as the computational kernel that models the brain's core algorithm. A software frontend for modeling and simulating neural circuits and pathways complements the memory backend.
The methods introduced here lend themselves to binary in-memory computation, opening a new route toward synthetic intelligence that does not rely on GPU acceleration.
- Website: https://creatingintelligence.org
- Download PDF: https://creatingintelligence.org/CreatingIntelligence.pdf
- Preprint: https://arxiv.org/abs/2606.31819
- Discussions: https://github.com/PeterOvermann/CreatingIntelligence/discussions
- Issue Tracker: https://github.com/PeterOvermann/CreatingIntelligence/issues
On Unix, Linux, BSD, and macOS, run
src/python/setup.sh
This will create a Python virtual environment ~/.ci in your home directory and build the creating_intelligence package.
It includes a high-performance memory backend written in Standard C, and an alternative native Python version.
To build stand-alone zero-dependency versions of the Standard C memory kernel (a command line interface and test program), run
make -C src/c/
Learn how to set up neural architectures via the JSON dataflow description language: GettingStarted.pdf
Run a first example:
python3 experiments/hello-world/hello-world.py
To test the circuit package (and indirectly the memory backend), run
python3 experiments/circuit-tests/run.py
Test the stand-alone memory kernel (storage capacity, retrieval accuracy, and I/O performance) with
python3 experiments/memory-tests/run.py
or as a stand-alone C binary with
bin/memorytest
Copyright © 2026 Peter Overmann. All rights reserved.
This repository contains two distinct categories of content, which are licensed separately:
-
The Software: Licensed under the MIT License.
-
The Manuscript & Assets: All Rights Reserved. These materials may not be reproduced, distributed, modified, translated, or commercialized without explicit, prior written permission from the author.
See the full LICENSE file for details.