Synapse
⌘+K
Synapse
PulseExploreJournal ClubResearchersJournals
Instagram
HomeJournal ClubExplore
July 26, 2026Topics in Cognitive ScienceOpen Access

Infinite Ends From Finite Samples: Open‐Ended Goal Inference as Top‐Down Bayesian Filtering of Bottom‐Up Proposals

View Full Paper
Ask AI
Bookmark
Share

Authors

TZTan Zhi-XuanGKGloria KangVMVikash Mansinghka

Discussion

Loading...

Member takes

Overview

Randomized trial demonstrates goal inference enhancements via an open-ended model, suggesting improved cognitive processing efficiency.

Key Points

  • The aim is to explore how humans infer extensive goals from limited observations using a new model.
  • Introduced open-ended sequential inverse plan search (SIPS) model for goal inference.
  • Validated the model using a Block Words task where participants guess word formations.
  • Compared predictions of the SIPS model with heuristic guessing and exact Bayesian inference.
  • SIPS model predicts mean and variance of human goal inferences significantly better than heuristic methods.
  • Achieved similar accuracy to exact Bayesian models with reduced cognitive cost.
  • Demonstrates rational pruning of unrealistic goals, enhancing model's efficacy in goal inference.

Cite This Study

Zhi-Xuan et al. (2026) studied this question.

synapsesocial.com/papers/6a65a2e2d3aea3239cd763a5https://doi.org/10.1111/tops.70075
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Symbolic Knowledge Distillation: from General Language Models to Commonsense Models2022 · 158 citations
  2. 2Gen: a general-purpose probabilistic programming system with programmable inference2019 · 130 citations
  3. 3How We Know What Not To Think2019 · 120 citations
  4. 4Adapting the Number of Particles in Sequential Monte Carlo Methods Through an Online Scheme for Convergence Assessment2016 · 69 citations