What’s new in Simbrain 4

Simbrain 4.0 involved a complete rewrite of Simbrain that occurred between 2015 and 2025. During this time, most of the code was migrated to Kotlin and refactored, unit-tests were added, and hundreds of new features were introduced.

Highlights

• Integration of array-based computations throughout the network system, for example via neuron arrays and weight matrices.
• A flexible supervised learning framework with multiple optimizers allows training networks on the fly through complex architectures.
• Transformer blocks and modern ML components (including GELU activation).
• Image World was rewritten with an image processing pipeline system, image albums, and a live editor.
• Text World merged separate components into a unified NLP environment with token embeddings, multiple tokenizers including byte-pair encoding, and language model integration.
• Odor World now functions like a 2D game engine with tilemaps, animated sprites, and new object sensors.
• A new pixel plot was introduced for visualizing activation patterns as pixel grids.
• The projection plot was rewritten with dimensionality reduction techniques (PCA, t-SNE, Sammon mapping) and improved coloring managers.
• A neuro-evolutionary framework was added with evolution simulations for evolving neural networks, virtual robots, and multi-agent systems.
• A full simulation environment was introduced with over 70 simulations, replacing the earlier scripting environment.
• A classifier object was added, wrapping the Smile machine learning framework.

Network

• Undo/redo (Finally!)
• The major change is array-based neural networks, including a carefully developed GUI (see neuron arrays and matrices) for visualizing array and matrix operations in Simbrain’s signature intuitive style. Array operations enable matrix-based computations for neuron arrays and weight matrices, with support for batch processing and vectorized operations.
• Introduction of transformer blocks and support for language models.
• Supervised models that allow for using backprop on the fly on a source and target set of neurons or neuron arrays, including through paths of diverging and reconverging weights. The framework includes multiple optimizers (Adam, SGD) and flexible weight initialization strategies.
• Neuron collections were introduced as a new way of organizing neurons. These are lightweight wrappers on neurons that can overlap.
• Neuron groups were reorganized and expanded. Competitive, SOM, and winner-take-all became specialized neuron group types rather than subnetworks, making it easier to create these architectures. New normalization and softmax group types were also added.
• Over 20 neuron update rules are now available, including biologically detailed models (Morris-Lecar, Izhikevich, FitzHugh-Nagumo) and modern ML activations like GELU (for transformers), Kuramoto, Allostatic, and Softmax.
• New connection strategies including distance-based and fixed-degree connections.
• New subnetwork types including restricted Boltzmann machines
• A classifier object was added, which wraps the Smile machine learning framework.
• Spike responders were reorganized into a dedicated framework with improved support for modeling synaptic dynamics.

Plots

• A new pixel plot was introduced for visualizing neuron arrays and matrices as pixel grids, useful for displaying activation patterns and image data.
• The projection plot was rewritten with support for dimensionality reduction techniques (PCA, t-SNE, Sammon mapping) to visualize high-dimensional network dynamics. New coloring managers enable better visualization of temporal patterns and data relationships.
• Time series plots were updated with improved auto-range capacities for better tracking of dynamic data ranges.
• Raster plots were updated with improved spike visualization and performance.
• Other plot types (bar charts, histograms, pie charts) remain available for data analysis.

Odor World

Odor World has been significantly rewritten to function as a standard 2D game engine for embodied agent simulations.

• Zooming and panning support for navigating larger environments
• Tilemap system with Tiled editor integration for creating complex worlds
• Animated sprites for dynamic entity visualization
• New object sensor type provides a simplified alternative to smell sensors for detecting specific entity types by name (e.g., “Swiss cheese”) without requiring full vector-based smell source configuration
• Sensors and effectors for movement (straight movement, turning), speech, hearing, bump detection, and tile sensing enable sophisticated embodied agent behaviors

Image World

Image World was completely rewritten (formerly Vision World) with an image processing pipeline architecture.

• Image album system allows loading and cycling through multiple images for training and testing
• Live image editor for on-the-fly modification and experimentation
• Image processing operations including edge detection, Gabor filters, grayscale conversion, thresholding, and resizing for preprocessing visual inputs
• Pipeline system enables chaining multiple operations for complex image transformations

Text World

Text World was completely rewritten with improved NLP capabilities (merged from separate Display and Reader components).

• Token embeddings management with support for custom embeddings and pre-trained models
• Multiple tokenization methods (simple word-based and byte-pair encoding) and embedding types (one-hot, co-occurrence with sliding window, custom pre-trained)
• Bytepair encoding visualization for understanding subword tokenization using the “show token boundaries” feature
• Integration with language model workflows through token embeddings

Other

• The preference framework was rewritten
• The tables and data world have been completely rewritten on top of the Smile framework.