New Paper on Developmental Braitenberg Vehicles now live!

 

The special issue of Artificial Life on Embodied Intelligence is now live! Inside you will find our paper "Braitenberg Vehicles as Developmental Neurosimulation", which has lived on the arXiv since 2020. This paper lays out an approach to Developmental Neurosimulation, involving three adversarial approaches to the agent-based development of embodied brains and embodied cognition. Here is the abstract:

Connecting brain and behavior is a longstanding issue in the areas of behavioral science, artificial intelligence, and neurobiology. As is standard among models of artificial and biological neural networks, an analogue of the fully mature brain is presented as a blank slate. However, this does not consider the realities of biological development and developmental learning. Our purpose is to model the development of an artificial organism that exhibits complex behaviors. We introduce three alternate approaches to demonstrate how developmental embodied agents can be implemented. The resulting developmental Braitenberg vehicles (dBVs) will generate behaviors ranging from stimulus responses to group behavior that resembles collective motion. We will situate this work in the domain of artificial brain networks along with broader themes such as embodied cognition, feedback, and emergence. Our perspective is exemplified by three software instantiations that demonstrate how a BV-genetic algorithm hybrid model, a multisensory Hebbian learning model, and multi-agent approaches can be used to approach BV development. We introduce use cases such as optimized spatial cognition (vehicle-genetic algorithm hybrid model), hinges connecting behavioral and neural models (multisensory Hebbian learning model), and cumulative classification (multi-agent approaches). In conclusion, we consider future applications of the developmental neurosimulation approach.

There are many themes to follow up on in this paper. Just of few examples include:

* brain/body scaling in an embodied agent.

* the role of multisensory integration in the development of cognition.

* ways to classify shapes and motifs in the emergence of multi-agent collectives. 

* spatial cognition and transfer learning in developmental embodied systems.

Congratulations to Stefan Dvoretskii, Ziyi Gong, Ankit Gupta, Jesse Parent, and Bradly Alicea for their hard work.

MAIN and Neuromatch Conference Presentations

The Orthogonal Research and Education Lab is on the virtual move! We have been featured at two conferences this week. The first conference is MAIN (Montreal Artificial Intelligence-Neuroscience) conference, a hybrid conference that focused on cutting-edge research in Neuro-AI. Our submission (Developmental Embodied NeuroSimulation) is a group effort and summarizes our work in this area over the past few years. The graphical abstract can be found below.

Graphical Abstract for Developmental Embodied NeuroSimulation.

We also had a presence at Neuromatch 4, with four flash talk presentations on four different topics. Neuromatch 4 was a great time, with two days of keynote talks, short talks, flash talks, and debate panels. 

Each flash talk was 7.5 minutes long, which requires an efficiency of words and ideas not typical of a longer format. The first talk is "The Universal Theory of Switching", which focuses on transitory "switching" phenomena. Switching behavior is ubiquitous across biological, physical, and algorithmic systems, and is controlled by sudden, first-order phase transition-like behavior we characterize as zeroth-order cybernetic regulation. 

Another talk is on "Allostatic Kinds". Allostatic Kinds are a way to regulate the boundaries of meaning and regulation of internal emotional and conscious states. This talk is presented by Jesse Parent, and features a mix of complex systems regulation, philosophy of mind, and consciousness studies. This talk was in conjunction with CEEALAR (Center for Enabling EA Learning and Research), an academic hostel located in Blackpool, UK.

Daniela Cialfi has built upon the lab's work on Meta-brain Models to develop "Economic Meta-brains", which are bio-economic agents that behave according to the free energy principle. Meta-brains are layered computational models that enable different levels of representation in the same agent. These model layers can be configured in geometrically specific ways, which in turn affects their function. The free energy principle enriches the meta-brains approach by adding a mathematically rigorous energetic component to a meta-brain agent. 

Finally, our presentation on "Gibsonian Information" comes with a preprint. Gibsonian Information is the information content of direct perceptual processing (sensu J.J. Gibson). We draw parallels between Shannon and Gibsonian Information, in addition to the role of such information in the dynamic interactions between agents and their environments. See our graphical abstract below, which simplifies the mathematics in the preprint. The talk also features a number of naturalistic settings in which Gibsonian Information can be demonstrated.

Graphical abstract for the Gibsonian Information paper/presentation (direct perception as information content).