NOTE The precise details and abstracts for each of the sessions below may change. We are actively working on the conference schedule and will send an update to participants when we publish the final program.

Present and Future of Bonsai

Moderators: Gonçalo Lopes, David Maas, João Frazão

Abstract: Bonsai has become a powerful platform for real-time data acquisition and closed-loop experimentation, widely used in neuroscience and other scientific domains. We will start the Bonsai Developer Conference by presenting the most recent developments in the language, visual editor, and package ecosystem, and share details on the future development roadmap. We will emphasize how our growing community is playing an increasing role in the development of Bonsai and how we hope to facilitate contributions and collaborations going forward.

Package Development

Moderators: Shawn Tan, Hugo Marques, Brandon Parks

Abstract: This session will explore the process of developing new operators for Bonsai, highlighting strategies for designing modular and reusable code. In addition, we will cover recommendations for organizing and publishing community packages—such as choosing intuitive naming conventions, creating clear and meaningful operator icons, and ensuring efficient long-term maintenance. Finally, we will discuss the importance of comprehensive documentation, including how-to guides, technical references, and examples, to ensure that these tools can be effectively utilized and adopted by the broader scientific community.

Machine Learning

Moderators: Nicholas Guilbeault, Joaquín Rapela

Abstract: Typically, neuroscientists use Bonsai for acquisition but rely on other programming languages, such as Python and MATLAB, for offline processing with machine learning (ML) algorithms. Recently, we started adding ML functionality to Bonsai by developing packages that interface with powerful ML tools for online data analysis. In this session we will discuss the basics of online probabilistic machine learning, and introduce a new package Bonsai.ML which aims at extending Bonsai with online machine learning techniques, such as linear dynamical systems, hidden Markov models, online Bayesian linear regression and neural decoding models. We will provide examples and tutorials throughout the session and end with a discussion on the future of machine learning in Bonsai.

Projects Showcase

Moderators: Adam Kampff, Hugo Marques

Abstract: One of the main goals of the Bonsai Developer Conference is to promote collaborations and sharing across the Bonsai community. Projects Showcase will be a unique opportunity to share and discuss your workflows and projects with other power users in the community. We look forward to showcase a broad collection of projects and applications in neuroscience and beyond. We strongly encourage anyone participating in the conference to share projects they would like to present in this session in the registration form.

Immersive Environments

Moderators: Andrew Erskine, David Maas

Abstract: Real-time closed-loop immersive environments are fundamental for next-generation understanding of brain function and behaviour, and also one of the most fun ways to learn reactive programming. With the introduction of BonVision and other high-level rendering packages in Bonsai, we are making it easier to do rapid prototyping of interactive virtual and augmented reality displays, as well as standard visual stimuli. In this session we will discuss how to create different kinds of immersive environments in Bonsai covering VR displays; vestibular, auditory and other sensory perturbations; and coupling of render engines to real-time input controllers and experimental hardware.

Hardware

Moderators: Josh Siegle

Abstract: The use of Bonsai as a hardware integration platform is well known, but the full breadth of support for external hardware is often underappreciated. In this session we will discuss how hardware packages can leverage data standards to ensure interoperability with processing pipelines; present new and forthcoming device integrations including high-speed and high-resolution cameras, the Harp ecosystem, and state-of-the-art electrophysiology recording systems like Open Ephys and Neuropixels; and highlight different approaches being used to accelerate integration of new devices, such as code generation and modular configuration. Finally, we will explore emerging trends in hardware package distribution, including automatic firmware upgrades.

Reproducible Research Practices for Data Analysis

Moderators: Bruno Cruz

Abstract: One of the most exciting opportunities introduced by the widespread adoption of Bonsai is the possibility of standardizing data formats and making reproducible data acquisition and analysis pipelines, allowing researchers to access and explore data as soon as it is collected. In this session we will discuss best practices for reproducible research using Bonsai, and showcase a set of flexible code generation tools and data formats developed to quickly integrate rich experimental metadata into data acquisition workflows. These tools and data formats provide a seamless bridge for analyzing both data and metadata using Python without sacrificing the flexibility and rapid prototyping needs of modern experimental design in neuroscience. We will finish with a discussion on current limitations and future research directions.

Task Control

Moderators: João Frazão

Abstract: Bonsai is still mostly used to drive data acquisition pipelines, but its potential to flexibly drive the logic of complex experiments is becoming more appreciated. In this session we will survey new and existing workflow patterns for complex and flexible control of tasks using Bonsai. These patterns allow the design of dynamic and parameterizable task structures that can be specified from external configuration files, external APIs, or graphical user interfaces. We will also present emerging approaches to create workflows for reconfigurable hardware, and new packages for designing graphical user interfaces and visualizations for online monitoring.

Quantifying Animal Behavior

Moderators: Jakob Voigts

Abstract: The potential for closed-loop experiments and manipulations in neuroscience is most limited by our ability to measure animal behavior of interest in real-time. In this session we want to bring together and showcase the many growing options for using Bonsai and state-of-the-art machine learning and computer vision techniques to measure the behavior of animals in real-time, including 2D and 3D pose estimation, statistical techniques for inferring kinematics and behavior syllables, as well as hybrid approaches combining hardware and signal processing techniques to follow the behavior of animals over long periods of time in naturalistic and freely-moving settings.

Roundtable

Moderators: Gonçalo Lopes

Abstract: We will invite both participants and moderators to take part in a discussion on the future of Bonsai, its developer community, and the emerging role for the Bonsai Foundation in supporting this community-driven project. We will collect questions and topics for discussion for this roundtable from each of the conference sessions, to help us discover together what is unique about our developer community, its strengths and weaknesses, and what may be holding us back from tapping the potential of Bonsai for enabling broader neuroscientific enquiry.