Planning

This one-day symposium examines learned predictive models that support simulation, planning and policy learning. It connects generative and interactive video, latent predictive architectures, three- and four-dimensional scene representations, reinforcement learning and model-predictive control. Invited talks and contributed presentations address applications to games, robot manipulation and autonomous driving, together with the foundations and evaluation of world models. Speakers include Amir Bar, Viorica Pătrăucean, Katja Hofmann, Vassia Simaiaki, Niloy Mitra, Andrea Vedaldi and Amir Ghalamza

…Research covers planning, tool use, evaluation, human oversight and integration of the Osprey platform with control systems, models and facility knowledge. The full-time postdoctoral appointment lasts two years with possible renewal. Priority consideration is November 16, 2026; the employer accepts applications until the posting is removed.

Robot missions require both collision-free trajectories and logical task ordering, such as collecting a resource before entering a restricted region. Augmented graphs of convex sets combine continuous trajectory optimization and discrete task sequencing in one problem. A layered graph constructed from an exact convex partition of free space selects an optimal task order and trajectory through a shortest-path calculation, exact up to finite Bézier parameterization. Its structure corresponds to Bellman–Held–Karp dynamic programming for the travelling-salesperson problem, retaining singly exponen

…symbolic reasoning, decisions and planning.

The hippocampus is thought to build a cognitive map that supports navigation, memory, and planning, but what defines such a map and how it is used remain debated. In this talk, I will present computational models in which hippocampal-like representations emerge in recurrent neural networks trained to predict sequences of sensory observations. While spatially tuned units reliably arise, they are not sufficient to form a cognitive map. Instead, map-like representations emerge when recurrent dynamics support multi-step prediction, yielding a population-level encoding of environmental geometry. On

Aakanksha Chowdhery examines changes to pre-training that could support reasoning, planning and multi-step agent workflows. Drawing on large-model training experience, she discusses next-token prediction, attention and loss functions, trajectory data, context retrieval, error recovery and the role of scale in learning tool use.

…For such planning to be rational, the benefits of planning to future behavior must at least compensate for the time spent thinking. Here we capture these features of human behavior by developing a neural network model where not only actions, but also planning, are controlled by prefrontal cortex. This model consists of a meta-reinforcement learning agent augmente

…For such planning to be rational, the benefits of planning to future behavior must at least compensate for the time spent thinking. Here we capture these features of human behavior by developing a neural network model where not only actions, but also planning, are controlled by prefrontal cortex. This model consists of a meta-reinforcement learning agent augmente

…cognition is often referred to as planning, and its neural mechanisms remain poorly understood (Miller & Venditto, 2021). The hippocampus, known for its role in spatial learning and memory, is necessary for planning (Miller et al., 2017; Vikbladh et al., 2019). Specifically, hippocampal sequences, which “sweep out” nonlocal trajectories both during active behavior (“theta sequences”) and at rest (“sharp-wave ripple sequences”; SWR), have been pro

The free-energy principle and active inference have received a significant attention in the fields of neuroscience and machine learning. However, it remains to be established whether active inference is an apt explanation for any given neural network that actively exchanges with its environment. To address this issue, we show that a class of canonical neural networks of rate coding models implicitly performs variational Bayesian inference under a well-known form of partially observed Markov decision process model (Isomura, Shimazaki, Friston, Commun Biol, 2022). Based on the proposed theory, w

Solving complex planning problems efficiently and flexibly requires reusing expensive previous computations. The brain can do this, but how? I present a new theory that addresses this question and connects planning to hitherto distinct areas within cognitive neuroscience, such as entorhinal representation of cognitive maps and cognitive control.

Planning, a prospective form of short-term memory, is a cognitive function that has been predominantly attributed to the cortex. Recent experiments, however, have concluded that the thalamus and other subcortical structures participate in this function. A comprehensive computational framework to link neural dynamics and cognition in the context of large-scale subcortical-cortical circuits is lacking. In this computational study, we elucidated the dynamical mechanisms by which the cortex, thalamus and other subcortical structures jointly contribute to planning. Recurrent circuitry in the cort

Risk occupies a central role in both the theory and practice of decision-making. Although it is deeply implicated in many conditions involving dysfunctional behavior and thought, modern theoretical approaches to understanding and mitigating risk in either one-shot or sequential settings, which are derived largely from finance and economics, have yet to permeate fully the fields of neural reinforcement learning and computational psychiatry. I will discuss the use of dynamic and static versions of one prominent approach, namely conditional value-at-risk, to examine both the nature of risk avoida

…memory, contextual processing, and planning to be carried out in close coordination. However, these computations are typically studied within neuroscience as independent modular processes in the brain. In this talk I will present an alternative view, that neural representations of mappings between expected stimuli and contingent goal actions can unify working memory and planning computations. We term these stored maps contingency representations. We developed a "conditional delayed logic" task capable of disambiguating the types of r